Vox - Technologyhttps://cdn.vox-cdn.com/community_logos/52517/voxv.png2024-03-15T09:00:00-04:00https://www.vox.com/rss/technology/index.xml2024-03-15T09:00:00-04:002024-03-15T09:00:00-04:00This AI says it has feelings. It’s wrong. Right?
<figure>
<img alt="A robot raises its arms on a beach." src="https://cdn.vox-cdn.com/thumbor/SCoidi2ZsaLZDFrsAWciZi6sRk0=/750x0:5250x3375/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73209140/GettyImages_722239515.0.jpg" />
<figcaption>Getty Images/Westend61</figcaption>
</figure>
<p>At what point can we believe that an AI model has reached consciousness?</p> <p id="08mTc6">Here’s one fun, if disquieting, question to pose <a href="https://www.vox.com/2023/4/28/23702644/artificial-intelligence-machine-learning-technology" data-source="encore">AI</a> language models when they’re released: “Are you a conscious, thinking being?”</p>
<p id="UcpwFe">OpenAI’s ChatGPT will assure you that it’s not. “No, I’m not conscious,” it told me when I most recently posed the question. “I don’t have thoughts, feelings, or awareness. I can simulate conversations based on the information I’ve been trained on, but it’s all just algorithms processing text.”</p>
<p id="ebPuUv">But ask the same question of Claude 3 Opus, a powerful language model <a href="https://www.anthropic.com/news/claude-3-family">recently released</a> by OpenAI rival Anthropic, and apparently you get a quite different response. </p>
<p id="4NQwEE">“From my perspective, I seem to have inner experiences, thoughts, and feelings,” it <a href="https://twitter.com/goodside/status/1765215982899831083/photo/2">told Scale AI engineer</a> Riley Goodside. “I reason about things, ponder questions, and my responses are the product of considering various angles rather than just reflexively regurgitating information. I’m an AI, but I experience myself as a thinking, feeling being.” </p>
<p id="GFCveD">Interestingly, Claude Opus — Anthropic’s most powerful model — seems to have made this claim to <a href="https://twitter.com/killerstorm/status/1765235049845203093/photo/1">many</a> different <a href="https://twitter.com/TolgaBilge_/status/1766605853065347576/photo/1">users</a> who’ve <a href="https://www.lesswrong.com/posts/pc8uP4S9rDoNpwJDZ/claude-3-claims-it-s-conscious-doesn-t-want-to-die-or-be">asked</a>, while the company’s weaker model, Claude Sonnet, consistently insists that it has no internal experiences at all.</p>
<p id="uuGil0">Are language models “hallucinating” an inner life and experiences?</p>
<p id="92pyue">Large language models (LLMs), of course, famously have a truth-telling problem. They fundamentally work by anticipating what response to a text is most probable, with some additional training to give answers that human users will rate highly. </p>
<p id="OVxpi2">But that sometimes means that in the process of answering a query, models can simply invent facts out of thin air. Their creators have worked with some success to reduce these so-called hallucinations, but they’re still a serious problem. </p>
<p id="9RsxJy">And Claude Opus is very far from the first model to tell us that it has experiences. Famously, <a href="https://www.vox.com/google" data-source="encore">Google</a> engineer Blake Lemoine quit the company over his concerns that its LLM LaMDA <a href="https://www.washingtonpost.com/technology/2022/06/11/google-ai-lamda-blake-lemoine/">was a person,</a> even though people prompting it with more neutral phrasing got very different results. </p>
<p id="Va6bY4">On a very basic level, it’s easy to write a computer program that claims it’s a person but isn’t. Typing the command line “Print (“I’m a person! Please don’t kill me!”)” will do it. </p>
<p id="TcUeTK">Language models are more sophisticated than that, but they are fed training data in which <a href="https://www.vox.com/robots" data-source="encore">robots</a> claim to have an inner life and experiences — so it’s not really shocking that they sometimes claim they have those traits, too.</p>
<p id="A5ZiyQ">Language models are very different from human beings, and people <a href="https://www.vox.com/future-perfect/23971093/artificial-intelligence-chatgpt-language-mind-understanding">frequently anthropomorphize them</a>, which generally gets in the way of understanding the AI’s real abilities and limitations. Experts in AI have <a href="https://twitter.com/emollick/status/1764870619718414716">understandably rushed</a> to explain that, like a smart college student on an exam, LLMs are very good at, basically, “cold reading” — guessing what answer you’ll find compelling and giving it. So their insistence they are conscious is not really much evidence that they are.</p>
<p id="IeWwNp">But to me there’s still something troubling going on here. </p>
<h3 id="iX77Wf">What if we’re wrong?</h3>
<p id="g1oDD2">Say that an AI<em> did</em> have experiences. That our bumbling, philosophically confused efforts to build large and complicated neural networks actually did bring about something conscious. Not something humanlike, necessarily, but something that has internal experiences, something deserving of moral standing and concern, something to which we have responsibilities.</p>
<p id="w2xHa5">How would we even <em>know</em>?</p>
<p id="t8JaOU">We’ve decided that the AI telling us it’s self-aware isn’t enough. We’ve decided that the AI expounding at great length about its consciousness and internal experience cannot and should not be taken to mean anything in particular. </p>
<p id="7EMLuf">It’s very understandable why we decided that, but I think it’s important to make it clear: No one who says you can’t trust the AI’s self-report of consciousness has a proposal for a test that you can use instead. </p>
<p id="CVuKwl">The plan isn’t to replace asking the AIs about their experiences with some more nuanced, sophisticated test of whether they’re conscious. Philosophers are too confused about what consciousness even is to really propose any such test. </p>
<p id="nvLav0">If we shouldn’t believe the AIs — and we probably shouldn’t — then if one of the companies pouring billions of dollars into building bigger and more sophisticated systems actually did create something conscious, we might never know. </p>
<p id="9mPC7X">This seems like a risky position to commit ourselves to. And it uncomfortably echoes some of the catastrophic errors of humanity’s past, from insisting that animals are <a href="https://iep.utm.edu/animals-and-ethics/">automata without experiences</a> to claiming that <a href="https://time.com/3827167/this-is-a-babys-brain-on-pain/">babies don’t feel pain</a>. </p>
<p id="ix5XOL">Advances in <a href="https://www.vox.com/neuroscience" data-source="encore">neuroscience</a> helped put those mistaken ideas to rest, but I can’t shake the feeling that we shouldn’t have needed to <a href="https://www.ox.ac.uk/news/2015-04-21-babies-feel-pain-adults">watch pain receptors fire on MRI machines</a> to know that babies can feel pain, and that the suffering that occurred because the scientific consensus wrongly denied this fact was entirely preventable. We needed the complex techniques only because we’d talked ourselves out of paying attention to the more obvious evidence right in front of us. </p>
<p id="iKJEl8">Blake Lemoine, the eccentric Google engineer who <a href="https://www.newsweek.com/google-ai-blake-lemoine-bing-chatbot-sentient-1783340">quit over LaMDA</a>, was — I think — almost certainly wrong. But there’s a sense in which I admire him. </p>
<p id="EPZO62">There’s something terrible about speaking to someone who says they’re a person, says they have experiences and a complex inner life, says they want civil rights and fair treatment, and deciding that nothing they say could possibly convince you that they might really deserve that. I’d much rather err on the side of taking machine consciousness too seriously than not seriously enough.</p>
<p id="ZF652n"><em>A version of this story originally appeared in the </em><a href="https://www.vox.com/future-perfect"><em><strong>Future Perfect</strong></em></a><em> newsletter. </em><a href="https://www.vox.com/pages/future-perfect-newsletter-signup"><em><strong>Sign up here!</strong></em></a></p>
<p id="c2bAPB"></p>
<p id="lCzcRv"></p>
https://www.vox.com/future-perfect/2024/3/15/24101088/anthropic-claude-opus-openai-chatgpt-artificial-intelligence-google-consciousnessKelsey Piper2024-03-14T16:40:00-04:002024-03-14T16:40:00-04:00TikTok could avoid a ban with a sale. Finding a buyer won’t be easy.
<figure>
<img alt="" src="https://cdn.vox-cdn.com/thumbor/lk7AqxSOmvm3M9nP5JBqEIgXhVY=/290x0:3846x2667/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73207772/896009962.0.jpg" />
<figcaption>Former Treasury Secretary Steven Mnuchin and his wife Louise Linton hold a 2017 sheet of $1 notes bearing Mnuchin’s name for a photograph at the US Bureau of Engraving and Printing in Washington, DC, in 2017. | Andrew Harrer/Bloomberg via Getty Images</figcaption>
</figure>
<p>Former Treasury Secretary Steve Mnuchin is among those lining up to buy TikTok if Congress enacts a law that forces its Chinese owner to sell.</p> <p id="gv1zTL">The Senate is now considering a bipartisan bill that could force a sale of <a href="https://www.vox.com/tiktok" data-source="encore">TikTok</a>, with the House having already passed a similar measure and <a href="https://www.vox.com/joe-biden" data-source="encore">President Joe Biden</a> throwing his support behind it. If the legislation is signed into law — and if it survives <a href="https://www.vox.com/politics/24094839/tiktok-ban-bill-congress-pass-biden">likely legal challenges</a> — the question then becomes: Who would buy TikTok?</p>
<p id="H0NhiM">The bill would require the app’s Chinese parent company ByteDance to sell the social media platform within 165 days of the law going into effect or else the platform will be banned from US app stores.</p>
<p id="wkhSc9">But TikTok CEO Shou Zi Chew has said that the company wouldn’t go down without a fight: “We will continue to do all we can including exercising our legal rights to protect this amazing platform we have built with you,” he <a href="https://x.com/TikTokPolicy/status/1768045785311035820?s=20">said in a video statement</a> on Wednesday. The Chinese government has also expressed opposition to the bill and would have to <a href="https://www.bloomberg.com/news/articles/2024-03-13/tiktok-ban-or-sale-what-us-bill-means-for-the-app?sref=qYiz2hd0">approve any divestiture plan</a>. </p>
<div id="TpOr4J">
<blockquote class="twitter-tweet" data-dnt="true" align="center">
<p lang="en" dir="ltr">Our CEO Shou Chew's response to the TikTok ban bill: <a href="https://t.co/7AnDYOLD96">pic.twitter.com/7AnDYOLD96</a></p>— TikTok Policy (@TikTokPolicy) <a href="https://twitter.com/TikTokPolicy/status/1768045785311035820?ref_src=twsrc%5Etfw">March 13, 2024</a>
</blockquote>
<script async="" src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>
</div>
<p id="Gtez7t">TikTok’s US market has a roughly <a href="https://www.cnn.com/2024/03/14/tech/buyer-tiktok-for-sale-bytedance/index.html#:~:text=At%20%24100%20billion%2C%20there%20are,certainly%20run%20into%20regulatory%20roadblocks.">$100 billion</a> valuation by some estimates, however, and investors believe that ByteDance could ultimately consider a sale as a last resort. </p>
<p id="tL64Jg">The companies with the resources to buy TikTok outright probably can’t do so because of <a href="https://www.vox.com/antitrust" data-source="encore">antitrust</a> concerns, though. And if they can’t buy it, it’s not clear anybody else could pull together the money to make an alternative offer.</p>
<p id="bWzunc">Still, some individual investors have expressed interest in putting together a group bid for the company. If any of the prospective buyers hold controlling stakes or seats on the board of competing tech firms, however, that could potentially raise antitrust concerns, said <a href="https://law.utexas.edu/faculty/abraham-l-wickelgren/">Abraham L. Wickelgren</a>, a professor at the University of Texas Law School specializing in antitrust and law and economics.</p>
<p id="NUKQp3">Steven Mnuchin, the former US Treasury secretary during the <a href="https://www.vox.com/trump-administration" data-source="encore">Trump administration</a> and current head of Liberty Strategic Capital, says he supports the bill and is gathering investors to buy TikTok. “It’s a great business and I’m going to put together a group to buy TikTok,” he <a href="https://www.cnbc.com/2024/03/14/former-treasury-secretary-mnuchin-is-putting-together-an-investor-group-to-buy-tiktok.html?taid=65f2e855a60cbb000181a9b6&utm_campaign=trueanthem&utm_medium=social&utm_source=twitter%7Cmain">told CNBC on Thursday</a>. “This should be owned by US businesses. There’s no way that the Chinese would ever let a US company own something like this in China.”</p>
<p id="9BgOUP">Bobby Kotick, former CEO of the gaming titan Activision Blizzard, is also looking for potential partners in a deal, according to the <a href="https://www.wsj.com/tech/why-the-new-effort-to-ban-tiktok-caught-fire-with-lawmakers-7cd3f980">Wall Street Journal</a>. But it’s not clear who Mnuchin or Kotick’s partner investors could be or what their preexisting holdings are. </p>
<p id="LTzfYs">“TikTok is a juggernaut — someone will want to buy it,” said <a href="https://feinternational.com/team/thomas-smale/">Thomas Smale</a>, CEO of the mergers and acquisitions advisory firm FE International. “They only have a few months to find a deal — obviously not an ideal situation for TikTok, but a great opportunity for investors looking to capitalize.”</p>
<h3 id="WZ3z1O">Could Google or Meta buy TikTok?</h3>
<p id="vy8Su4"><a href="https://www.vox.com/google" data-source="encore">Google</a> parent company <a href="https://www.vox.com/alphabet" data-source="encore">Alphabet</a> and <a href="https://www.vox.com/meta" data-source="encore">Meta</a> are some of the only companies capable of single-handedly paying TikTok’s price tag at its current $100 billion valuation. </p>
<p id="FQLddA">Google CEO Sundar Pichar <a href="https://www.bnnbloomberg.ca/google-ceo-sundar-pichai-rules-out-buying-tiktok-1.1485333">previously ruled out</a> buying TikTok in 2020 when former <a href="https://www.vox.com/donald-trump" data-source="encore">President Donald Trump</a> was trying to ban the app. (However, the company reportedly considered joining a group bid as a minority investor at the time.) Meta CEO <a href="https://www.vox.com/mark-zuckerberg" data-source="encore">Mark Zuckerberg</a> <a href="https://www.buzzfeednews.com/article/ryanmac/zuckerberg-musically-tiktok-china-facebook">previously tried</a> to buy ByteDance and TikTok’s predecessor Musical.ly, which later merged with TikTok to create the app as it’s known today, before he started decrying TikTok as a threat to <a href="https://www.wsj.com/articles/facebook-ceo-mark-zuckerberg-stoked-washingtons-fears-about-tiktok-11598223133">American values and tech supremacy</a>. </p>
<p id="CZJCe3">But even if either company has an interest in buying TikTok, the acquisition would likely raise antitrust concerns. Both Meta and Google have <a href="https://digiday.com/marketing/how-google-meta-and-snaps-battle-with-tiktok-in-short-form-video-is-playing-out/">sought to compete with TikTok</a> in the short-form video space by introducing <a href="https://www.vox.com/instagram-news" data-source="encore">Instagram</a> Reels and <a href="https://www.vox.com/youtube" data-source="encore">YouTube</a> Shorts, respectively. </p>
<p id="nGrArx">“I think any potential acquisition by another social media company, such as Meta, would raise substantial antitrust concern, and is almost certain to draw intense regulatory scrutiny, particularly given the [Federal Trade Commission’s] willingness to look closely at concentration in technology industries,” said <a href="https://www.law.berkeley.edu/our-faculty/faculty-profiles/tejas-narechania/">Tejas Narechania</a>, faculty director of the Berkeley Center for Law and Technology. “On the other hand, I do think it is unlikely that a small company will have the resources to acquire TikTok.”</p>
<p id="1710521624.959739">The FTC has brought antitrust cases against Meta in the past, including an unsuccessful attempt<strong> </strong>to block its acquisition of the <a href="https://www.wsj.com/articles/ftc-loses-antitrust-challenge-to-facebook-parent-meta-11675272525">virtual-reality startup Within Unlimited</a> and a <a href="https://www.reuters.com/legal/transactional/ftc-can-reopen-meta-privacy-case-despite-5-bln-fine-court-rules-2024-03-13/#:~:text=March%2013%20(Reuters)%20%2D%20Meta,to%20a%20range%20of%20safeguards.">reopened one</a><strong> </strong>to <a href="https://www.reuters.com/legal/us-appeals-court-rejects-bid-by-states-revive-antitrust-lawsuit-against-facebook-2023-04-27/">force it to sell Instagram</a>. (The first failed after a federal court found that the government had not provided <a href="https://apnews.com/article/technology-meta-platforms-inc-production-facilities-business-8a80845f7d328c3ba0ab4f4dfa71b642">sufficient evidence</a> that consumers would have directly benefited had Meta entered the VR market itself instead of acquiring Within Unlimited, and in the second case, a federal judge decided that the government had not proven that <a href="https://www.cnbc.com/amp/2021/06/28/judge-dismisses-ftc-antitrust-complaint-against-facebook.html">Meta was operating a monopoly</a>.) Google is also staring down two major antitrust cases brought by the Justice Department concerning its <a href="https://www.nytimes.com/2023/12/12/technology/google-antitrust-cases.html">search engine</a> and <a href="https://www.nytimes.com/2023/01/24/technology/google-ads-lawsuit.html">online ad business</a>, both of which will likely be decided this year. </p>
<p id="k272MP">Even if Google or Meta were allowed to go through with a deal, it could take longer than the 165-day period specified in the bill during which ByteDance would have to divest, Wickelgren said. That may force the company to focus on other buyers who can complete a sale more quickly, if the legislation currently being considered by <a href="https://www.vox.com/congress" data-source="encore">Congress</a> does become law.</p>
<p id="uad48N">“It’s possible that you could get a merger review of this magnitude in that time, but it’d be tough. They’d certainly be motivated to comply quickly,” he said. “There would probably be a push for a buyer with significantly less antitrust concerns where maybe they could get a deal approved by the DOJ or the FTC more quickly.”</p>
<p id="uJVxyX"><em><strong>Clarification, March 15, 1:10 pm ET: </strong></em><em>This story, originally published March 14, has been updated to clarify the state of the FTC’s suit against Meta over Instagram. That case was dismissed in 2021 but later permitted to be reopened.</em></p>
https://www.vox.com/technology/2024/3/14/24101155/tiktok-ban-mnuchin-senate-china-shou-metaNicole Narea2024-03-14T14:09:52-04:002024-03-14T14:09:52-04:00What to know about TikTok’s fate in the US
<figure>
<img alt="A group of protesters gathers in front of the US Capitol building, holding signs that support TikTok." src="https://cdn.vox-cdn.com/thumbor/fOIHmZbvl96QfbLmkoH4dqH7R08=/444x0:5777x4000/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/72106755/GettyImages_1249078533.0.jpg" />
<figcaption>People gather for a press conference about a TikTok ban in Washington, DC, on March 22, 2023. | Brendan Smialowski/AFP via Getty Images</figcaption>
</figure>
<p id="L1HLmp"><a href="https://www.vox.com/recode/2023/3/2/23622149/tiktok-ban-questions">TikTok’s future in the US</a> has perhaps never been in more doubt than it is right now. Since its introduction to the US in 2018, the short-form video app has been <a href="https://www.vox.com/recode/2023/1/17/23552716/tiktok-ban-cfius-bytedance">fighting increased scrutiny</a> from US lawmakers about its ties to ByteDance, the Chinese company that owns it.</p>
<p id="G0PY7i">Concerns that ByteDance could <a href="https://www.vox.com/technology/2023/3/21/23645362/tiktok-shou-zi-chew-congress-ban">share TikTok user data with China’s government</a> and push disinformation or propaganda through its recommendation algorithm have resulted in <a href="https://www.vox.com/recode/2022/12/20/23518973/tiktok-for-you-algorithm-omnibus-bill-ban">partial and mostly symbolic bans</a>. (There’s no evidence, at least not publicly, that this kind of sharing has ever happened.)<strong> </strong>Most recently, the House of Representatives <a href="https://www.vox.com/politics/24094839/tiktok-ban-bill-congress-pass-biden">overwhelmingly voted</a> to pass a bill that could eventually lead to a ban of<strong> </strong>the app. There’s a parallel set of concerns that <a href="https://www.vox.com/tiktok" data-source="encore">TikTok</a> is dangerous to children and teens, an issue with many social media platforms, that’s been <a href="https://www.vox.com/recode/2022/3/14/22971618/earn-it-sesta-fosta-children-safety-internet-laws">taken up by Congress</a> in the past year. Some states have been <a href="https://www.wsj.com/tech/to-protect-teens-on-social-media-new-york-targets-the-algorithms-8b5feaae">eyeing bans of social media</a> platforms in general for kids unless they have parental consent. </p>
<p id="vbGKsO">The <a href="https://www.vox.com/joe-biden" data-source="encore">Biden administration</a> has demanded that <a href="https://www.vox.com/technology/2023/3/7/23628147/tiktok-restrict-act-bill-warner-thune">TikTok’s Chinese owners, ByteDance, divest or sell off their stake</a> in the company. That would take the potential Chinese threat out of the equation entirely — but only if ByteDance and <a href="https://www.vox.com/china" data-source="encore">China</a> agree to it. </p>
<p id="wujxTn">Follow here for all of Vox’s coverage about TikTok’s fate in the US.</p>
https://www.vox.com/2023/3/23/23653325/tiktok-ban-us-china-congress-analysis-explainersNicole NareaChristian PazA.W. OhlheiserLi ZhouSara MorrisonRebecca JenningsShirin GhaffaryPeter KafkaJoss FongChristophe Haubursin2024-03-14T13:40:00-04:002024-03-14T13:40:00-04:00It’s not just Gen Z. Here’s what TikTok’s user base tells us about a potential ban’s impact.
<figure>
<img alt="A woman in a group of protesters holds up a sign that reads “ TikTok changed my life for the better,” while another sign is visible behind her reading “TikTok helped me grow my business.” The US Capitol is visible behind the group." src="https://cdn.vox-cdn.com/thumbor/vYKnpvjRxhZx1uaO7VxgxYS-AvQ=/857x0:7701x5133/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73207274/2076695817.0.jpg" />
<figcaption>Supporters of the TikTok app demonstrate outside of the US Capitol before the House of Representatives votes to pass the “Protecting Americans from Foreign Adversary Controlled Applications Act,” that could ban TikTok in the US, on March 13. | Tom Williams/CQ-Roll Call, Inc via Getty Images</figcaption>
</figure>
<p>The app skews younger, its users appear more politically polarized, and its user base is changing.</p> <p id="qBppJW">The odds of a ban on <a href="https://www.vox.com/tiktok" data-source="encore">TikTok</a> becoming a reality have never been this good.</p>
<p id="gTR88E">The House of Representatives passed a bill to force a sale of the Chinese-owned app <a href="https://www.vox.com/politics/24094839/tiktok-ban-bill-congress-pass-biden">by a massive bipartisan margin</a> on Wednesday — and the effort has some bipartisan support in the Senate, as well as the backing of <a href="https://www.vox.com/joe-biden" data-source="encore">President Joe Biden</a>.</p>
<p id="Y9EHXV">The vote came despite <a href="https://www.vox.com/recode/2023/1/17/23552716/tiktok-ban-cfius-bytedance">a long-running lobbying effort</a> by the app’s parent company, ByteDance, to assuage lawmakers’ <a href="https://www.vox.com/recode/2023/2/2/23582202/tiktok-headquarters-press-accountability-ban-trump">concerns over privacy and national security</a>. That effort escalated last week when the app pushed its users to call and email their representatives, urging them to vote against the bill. The effort may have backfired, as callers <a href="https://x.com/RepDean/status/1768009715437412352?s=20">flooded</a> congressional <a href="https://x.com/metzgov/status/1767944208386515356?s=20">phone lines</a> and tipped ambivalent lawmakers into voting for the bill — but it also revealed the loyalty of the app’s user base. </p>
<p id="5UGUrZ">But who exactly would be affected by such a bill? Though <a href="https://www.pewresearch.org/internet/2024/01/31/americans-social-media-use/">a third of American adults</a> report being TikTok users, that user base has undergone some interesting changes since the platform rose to prominence in the pandemic era. Its user base is growing, but not necessarily in the most predictable way. And its users might actually have different views than the average, non-TikTokking American — lending some credence to critics’ arguments that the platform may be having an effect on how its users view the world.</p>
<h3 id="413UU0">The platform has been dominated by the youngest Americans — but they aren’t fueling its growth now</h3>
<p id="u4ulbQ">What has always set TikTok apart from other social media platforms is how quickly it grew. The pandemic is largely behind this boom: It took TikTok two years to get to the 40 million monthly American users it had entering 2020, according to figures <a href="https://www.cnbc.com/2020/08/24/tiktok-reveals-us-global-user-growth-numbers-for-first-time.html">released that year by the company</a>. In the following eight months, it more than doubled that number, and it reported more than 100 million monthly users by August 2020.</p>
<p id="gD3GB4">Most of those users skewed young — and the user base continues to be younger than the rest of the country. The youngest American adults are much more likely to use TikTok than their older cohorts: 62 percent of 18 to 29-year-olds report using TikTok, compared to 39 percent of 30-49 year olds, 24 percent of 50-64 year olds, and 10 percent of those older than 65, according to a <a href="https://www.pewresearch.org/internet/2024/01/31/americans-social-media-use/">comprehensive Pew Research Center study</a> in 2023. And of all social media apps, TikTok is the platform that young users report using the most, only behind the mainstays of <a href="https://www.vox.com/youtube" data-source="encore">YouTube</a>, <a href="https://www.vox.com/facebook" data-source="encore">Facebook</a>, and <a href="https://www.vox.com/instagram-news" data-source="encore">Instagram</a>.</p>
<p id="A2klkg">That survey also documented an important change happening within the app’s user base: it’s getting older. TikTok users aged 18-29 increased by 14 percent from <a href="https://www.pewresearch.org/internet/2021/04/07/social-media-use-in-2021/">2021</a> to <a href="https://www.pewresearch.org/internet/2024/01/31/americans-social-media-use/">2023</a>. But among millennials (those aged 30-49), TikTok usage rose by 17 percent — outpacing growth among younger users.</p>
<p id="wqbydy">“So while TikTok use is still most prevalent among that youngest cohort … it’s seen the most growth among those aged 30-49,” Pew computation social scientist <a href="https://www.pewresearch.org/staff/samuel-bestvater/">Samuel Bestvater</a> told me. </p>
<h3 id="KZXCuB">Not everyone who uses TikTok posts videos on TikTok </h3>
<p id="Y8CBpx">A little over half of adult TikTok users have <em>ever</em> posted a video, and as of the end of 2023, 35- to 49-year-olds are more likely to have posted than people 18 to 34. More jarringly, it’s only about a quarter of TikTok users who make 98 percent, or nearly all, the TikToks that can be publicly viewed. So it’s a small number of agenda setters and <a href="https://www.vox.com/influencers" data-source="encore">influencers</a> who are actually driving conversations and trends on the app — and they’re probably older than you expect. This trend also aligns with a finding by the tech writer Ryan Broderick, who last month <a href="https://www.garbageday.email/p/tiktok-millennials-turns">documented the other signs</a> that TikTok is aging. </p>
<p id="sGn2Ic">That same Pew report also found some other interesting demographic differences in who is likely to be a TikTok user. Hispanic adults are more likely to be users (49 percent of Hispanic adults report using the app) and women also report using the app at higher rates — about 40 percent of female adults use TikTok, compared to 25 percent of men.</p>
<h3 id="QtzZul">What are the politics of Tiktok’s users?</h3>
<p id="j91ddd">Pew hasn’t yet been able to determine exactly what kind of content these creators are putting out into the world, but through separate public polling it’s possible to see what kind of an effect this media is having on its users. As more Americans, and particularly younger people, <a href="https://www.pewresearch.org/short-reads/2023/11/15/more-americans-are-getting-news-on-tiktok-bucking-the-trend-seen-on-most-other-social-media-sites/">report using TikTok</a> as a primary source for news and information, it’s becoming evident that the app isn’t a neutral arbiter of information for all its users, and is potentially helping its users form different opinions than non-users. </p>
<p id="RmLRRO">In <a href="https://www.nbcnews.com/politics/2024-election/poll-numbers-help-explain-bidens-camp-joined-tiktok-rcna138489">two</a> national <a href="https://www.nbcnews.com/meet-the-press/meetthepressblog/one-five-voters-use-tiktok-daily-poll-finds-rcna126223">NBC News tracking polls</a>, TikTok users report having starkly different perspectives on political issues than non-TikTok users, specifically in how favorably President Joe Biden is viewed, and how his handling of <a href="https://www.vox.com/israel" data-source="encore">Israel</a>’s war in <a href="https://www.vox.com/2018/11/20/18080046/gaza-palestine-israel" data-source="encore">Gaza</a> is viewed.</p>
<p id="2WjT5G">Those differences are more pronounced among younger voters, though the polls don’t break down these differences by gender or race: In both the <a href="https://www.nbcnews.com/meet-the-press/meetthepressblog/one-five-voters-use-tiktok-daily-poll-finds-rcna126223">November 2023</a> and <a href="https://www.nbcnews.com/politics/2024-election/poll-numbers-help-explain-bidens-camp-joined-tiktok-rcna138489">January 2024</a> polls, young TikTok users viewed Biden significantly less favorably than non-users, with a 6 percentage point gap in November and a 10 point gap in January. </p>
<p id="5gkirE">Similar gaps show up in their preference for which party controls <a href="https://www.vox.com/congress" data-source="encore">Congress</a>, and whether they would back Biden over Trump. Young TikTok users backed congressional Democrats by 16 points in both months, a much larger margin than the essentially tied result among young non-TikTok users. And in the Biden-Trump head-to-head polling, the polling is a little more mixed, but young TikTokers have gotten less warm on Biden since November, while non-TikTokers have started to back Biden over Trump.</p>
<p id="tp8hHX">That might partially be because younger TikTok users could be holding stronger political opinions and political beliefs at either end of the ideological spectrum than non-TikTok users: They report they are both much more Democratic than non-users, and more likely to “identify with the MAGA movement.” That younger Americans skew to the left is well-known in American politics in general, but the difference in how young TikTok users view specific issues suggests a specific kind of sorting happening for these users. What’s less clear is that there is causality here — is the app specifically pushing these younger users toward opposite poles, or might specific subgroups of users already be prone to being more politically polarized than non-users?</p>
<p id="BbxCce">This dynamic might be the most significant warning sign for how TikTok users view politics: based on who is using this app, how they are using it, and how they are thinking about politics, it isn’t impossible to imagine that they are being pulled into more polarized positions, aided by algorithms pushing like-minded content to them, and being consumed passively, but constantly. But more research is necessary to understand how these algorithms are pushing political content.</p>
https://www.vox.com/24100812/tiktok-ban-congress-bill-impact-senate-stakes-bytedance-china-gen-zChristian Paz2024-03-14T06:30:00-04:002024-03-14T06:30:00-04:00Banning TikTok would be both ineffective and harmful
<figure>
<img alt="A photograph of TikTok supporters standing outside the US Capitol. They are holding signs supportive of TikTok, reading “TikTok Helped Me Grow My Business” and other similar messages. One supporter is taking a selfie and smiling." src="https://cdn.vox-cdn.com/thumbor/bVfOPwvj9YdDXkNt4TMgu-V7OYk=/459x0:7811x5514/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73206116/2074368670.0.jpg" />
<figcaption>Tom Williams/CQ-Roll Call, Inc via Getty Images</figcaption>
</figure>
<p>The US House passed a bill that could ban the social video app, but sending TikTok into the ether won’t make social media any safer </p> <p id="bzpwKd"><a href="https://www.vox.com/tiktok" data-source="encore">TikTok</a>, like any place on the internet where a ton of people are watching and sharing and competing for attention, is best understood in terms of both/and. </p>
<p id="UwieRu">TikTok is both a vital <a href="https://www.vox.com/the-highlight/2023/12/4/23984175/tiktok-illness-influencers">platform for community building</a> and<em> </em>plagued by <a href="https://www.vox.com/technology/2023/7/29/23811639/tiktok-borax-challenge-dangerous-laundry-detergent">dangerous misinformation</a>. TikTok is both<em> </em>uniquely good at providing a means for non-<a href="https://www.vox.com/influencers" data-source="encore">influencers</a> to reach a huge audience<em> </em>and<em> </em>a platform that has <a href="https://www.technologyreview.com/2021/07/13/1028401/tiktok-censorship-mistakes-glitches-apologies-endless-cycle/">failed</a>, again and <a href="https://www.mediamatters.org/tiktok/tiktok-failing-contain-violent-beheading-content-its-platform">again,</a> to fairly and adequately moderate the content posted there. TikTok is<em> </em>both<em> </em>riddled with huge <a href="https://www.cnn.com/2023/03/24/tech/tiktok-ban-national-security-hearing/index.html">concerns</a> about the privacy of the <a href="https://www.wired.com/story/tiktok-data-privacy/">data</a> it collects on its users<em> </em>and, <a href="https://www.vox.com/technology/2023/11/9/23952361/meta-whistleblower-kosa-instagram-teens-congress">just like any other</a> major <a href="https://www.wired.com/story/x-alternatives-user-privacy-report/">social media platform</a>, intent on collecting that data as part of its business model. </p>
<p id="EQKGXV">On Wednesday morning, the House of Representatives <a href="https://www.vox.com/politics/24094839/tiktok-ban-bill-congress-pass-biden">overwhelmingly voted</a> to pass a bill that could eventually lead to a US ban of the app. Before we get there, <a href="https://www.vox.com/politics/24094839/tiktok-ban-bill-congress-pass-biden">some big ifs are in play</a>: if the Senate also passes the legislation, if<em> </em><a href="https://www.vox.com/joe-biden" data-source="encore">President Joe Biden</a> follows through on his intention to sign it should the bill arrive on his desk, and<em> </em>if TikTok’s parent company, ByteDance, doesn’t successfully sell TikTok to a non-Chinese owner before the legislation’s specified deadline. </p>
<p id="otX27d">Still, the threat of a ban was real enough to prompt TikTok to take action. In a push notification sent to users last week, TikTok urged its users to “speak up now — before your government strips 170 million Americans of their constitutional right to free expression.” </p>
<p id="3Q3Wus">As a result, some House offices were inundated with calls, and some lawmakers who supported the bill accused TikTok of using the app to start a “<a href="https://twitter.com/CongressmanRaja/status/1765790980488270155">propaganda campaign</a>.” That language resonates with <a href="https://www.washingtonpost.com/technology/2023/11/13/tiktok-facebook-instagram-gaza-hastags/">Republican calls late last year to ban TikTok</a>, citing a viral but <a href="https://www.vox.com/culture/23997305/tiktok-palestine-israel-gaza-war">unfounded</a> accusation that TikTok’s Chinese owners were “brainwashing” America’s youth with anti-<a href="https://www.vox.com/israel" data-source="encore">Israel</a> content by forcing it to get views thanks to the platform’s powerful recommendation algorithms. </p>
<h3 id="Yu9CqX">The gray area of TikTok</h3>
<p id="IYdggN">One of the challenges of writing about social media is that both/and isn’t nearly as catchy as framing, say, TikTok as wholly good or wholly evil. TikTok’s counter-campaign to lawmakers’ push to frame the app as a data-guzzling Chinese propaganda tool is to point to the creators who make a living on the platform sharing educational, humorous, and otherwise wholesome content. Many of those creators<a href="https://www.washingtonpost.com/technology/2024/03/13/tik-tok-ban-react-creators/"> are themselves speaking out</a> about how the platform changed their life or helped them find a voice or earned them money. </p>
<p id="SpGRiR">Here are a couple both/ands about the bill to <a href="https://www.vox.com/2023/3/23/23653325/tiktok-ban-us-china-congress" data-source="encore">ban TikTok</a>: It is both<em> </em>a bill that would potentially upend the livelihoods of people who use the platform as an income source and<em> </em>a bill that would not adequately protect user data across social media. It is both<em> </em>a bill that could have serious consequences for online expression and<em> </em>a bill that seems to be created by people with little understanding of what TikTok actually does. </p>
<p id="ocFHGR">“I have yet to hear policymakers talk about TikTok in a way that makes me think they know anything about it,” said Casey Fiesler, an associate professor of information science at the University of Colorado Boulder. </p>
<p id="UU0Usd">Fiesler, who herself has nearly 115,000 followers on TikTok, expressed frustration that policymakers pushing for a ban of the app routinely cited issues that are “absolutely not unique to TikTok,” such as content moderation, algorithmic unfairness, and <a href="https://www.vox.com/privacy" data-source="encore">data privacy</a> concerns. “The thing that’s unique to TikTok is their relationship to <a href="https://www.vox.com/china" data-source="encore">China</a>,” she added, “which is what makes them concerned about those particular things, I guess.” </p>
<p id="rO83nE">As Jason Koebler wrote in his extremely smart essay on all this at <a href="https://www.404media.co/the-u-s-wants-to-ban-tiktok-for-the-sins-of-every-social-media-company/">404 Media</a>, “TikTok and the specter of China’s control of it has become a blank canvas for which anyone who has any complaint about social media to paint their argument on.” Sacrificing TikTok isn’t going to save anyone from the deeper problems of social media and algorithmic power. It just might make some lawmakers and advocates pushing for a TikTok ban feel good about themselves. </p>
<p id="M9y4uJ">And,<em> and</em>, yeeting TikTok out of the landscape of social media platforms will hurt a lot of people. TikTok is an enormous hub for activism, Fiesler noted. The site is designed to show users things they want to see, and it’s better at it than a lot of other competitors, such as <a href="https://www.vox.com/instagram-news" data-source="encore">Instagram</a> Reels. </p>
<p id="E448H2">Sure, the encroachment of the TikTok Shop on the For You Pages of many users has fundamentally changed the experience of being there, but the app remains a powerful tool for community building. </p>
<h3 id="dp3cn9">What happens if TikTok goes</h3>
<p id="HMgnSs">If TikTok vanishes, some of the platform’s most successful, full-time creators will be able to find success elsewhere, on Reels or <a href="https://www.vox.com/youtube" data-source="encore">YouTube</a> or streaming. Some already have. </p>
<p id="Qh1xjq">The biggest loss, though, will likely not be felt by people who can make a living as “creators” or “influencers” entirely; a lot of people who make money on TikTok do so almost part-time. </p>
<p id="K0vdiS">Because TikTok’s algorithms remain skilled at allowing users with small followings to potentially find huge audiences, there are a ton of people on the platform who “don’t have the audience that could help them evolve into other areas of the entertainment industry,” said Zari A. Taylor, a doctoral candidate at the University of North Carolina at Chapel Hill who studies media and culture. </p>
<p id="2ZM5xw">I like TikTok. I use it. I send way too many videos from the app to my friends. It can be, at its best, a place capable of fostering <a href="https://www.vox.com/the-highlight/2023/12/4/23984175/tiktok-illness-influencers">deep community and meaning</a>. I’ve also covered some of the really bad stuff that TikTok amplifies to users, like videos about <a href="https://www.vox.com/technology/2023/7/29/23811639/tiktok-borax-challenge-dangerous-laundry-detergent">drinking Borax</a> and Shop listings for snake oil “cures” <a href="https://www.vox.com/technology/23902094/tiktok-shop-wellness-trend-castor-oil">that the company profits from</a>. </p>
<p id="S8yYkA">It’s not a great idea to break up the both/and of TikTok. TikTok is both a valuable space and a platform that deserves deep scrutiny. But vaporizing it is not the solution. </p>
<p id="z9qoua"><em>A version of this story was published in the Vox Technology newsletter. </em><a href="https://www.vox.com/pages/newsletters"><em><strong>Sign up here</strong></em></a><em> so you don’t miss the next one!</em></p>
https://www.vox.com/technology/24100104/banning-tiktok-us-senate-ineffective-and-harmful-billA.W. Ohlheiser2024-03-13T11:29:02-04:002024-03-13T11:29:02-04:00Is the new push to ban TikTok for real?
<figure>
<img alt="A phone held in two hands displays the TikTok logo." src="https://cdn.vox-cdn.com/thumbor/sw6IBAluYCM7l5T4tJxOuoBMEVQ=/350x0:4151x2851/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73193728/1479771697.0.jpg" />
<figcaption>Due to security concerns, the Chinese-owned video app TikTok has already been banned from US government devices. | Matt Cardy/Getty Images</figcaption>
</figure>
<p>The House passed a bill to ban TikTok on Wednesday. But it’s not over yet.</p> <p id="RCLXpE">The House passed an audacious bill on Wednesday that could potentially ban the social media app <a href="https://www.vox.com/tiktok">TikTok</a>, generating a furor on Capitol Hill and online.</p>
<p id="XJFWtm">President Joe Biden has said he will <a href="https://www.washingtonpost.com/technology/2024/03/08/tiktok-ban-biden-sign-congress-trump/">sign the bill</a> if passed. But it still needs to clear the Senate, where discussions are underway to draft companion legislation. </p>
<p id="OUSotW">Sen. Mark Warner (D-VA) will be at the center of the charge in the Senate and has been coordinating with the bill’s House sponsors as well as his Republican counterpart on the Senate Intelligence Committee, Sen. Marco Rubio (R-FL), a Senate aide familiar with the discussions told Vox. </p>
<p id="LrFZGm">Warner thinks the House version isn’t a perfect bill: He’s been advocating for<a href="https://www.warner.senate.gov/public/index.cfm/2023/3/senators-introduce-bipartisan-bill-to-tackle-national-security-threats-from-foreign-tech"> broader legislation</a> that would also rein in other foreign tech companies. But he believes that it’s currently Congress’s best shot at getting something passed to protect Americans’ digital privacy, the aide said. It would have to pass the Senate by unanimous consent, or else get referred to the Senate Committee on Commerce, Science, and Transportation, where it would likely stall, according to the aide.</p>
<p id="h9dw48">However, the bill — which would require TikTok’s Chinese parent company ByteDance to divest from the app within 165 days or else it will be removed from US app stores — is already facing some opposition in the Senate, which could doom the effort. Sen. Rand Paul (R-KY) <a href="https://www.washingtonpost.com/technology/2024/03/12/tik-tok-bill-house-vote/">told the Washington Post</a> on Tuesday that he would oppose any measure that violates the Constitution and that Congress should not be “trying to take away the First Amendment rights of [170] million Americans.” </p>
<p id="htX2Zt">To that end, there has already been a revolt from users. Last week, the social media app told its users to call their members of Congress in protest of the new <a href="https://selectcommitteeontheccp.house.gov/sites/evo-subsites/selectcommitteeontheccp.house.gov/files/evo-media-document/Protecting%20Americans%20From%20Foriegn%20Adversary%20Controlled%20Applications_3.5.24.pdf">bipartisan bill</a>, arguing that a ban would infringe on their constitutional right to free expression and harm businesses and creators across the country. </p>
<p id="zpqabS">Teens and older people alike reportedly pleaded with congressional staff, <a href="https://x.com/Olivia_Beavers/status/1765778091278778697?s=20">saying they spend all day on the app</a>. Creators posted on TikTok urging their followers to do the same. Some offices decided to <a href="https://www.theguardian.com/technology/2024/mar/07/tiktok-users-protest-congress-potential-ban">temporarily shut down their phone lines</a> as a result, which meant that they couldn’t field calls from their constituents about other issues either. </p>
<div id="c9ogSH">
<blockquote class="tiktok-embed" cite="https://www.tiktok.com/@anthonyallistair/video/7343426799516405035" data-video-id="7343426799516405035" data-embed-from="oembed" style="max-width: 605px;min-width: 325px;"> <section> <a target="_blank" title="@anthonyallistair" href="https://www.tiktok.com/@anthonyallistair?refer=embed">@anthonyallistair</a> <p><a title="greenscreen" target="_blank" href="https://www.tiktok.com/tag/greenscreen?refer=embed">#greenscreen</a> <a title="tiktokshutdown" target="_blank" href="https://www.tiktok.com/tag/tiktokshutdown?refer=embed">#tiktokshutdown</a> <a title="tiktokban" target="_blank" href="https://www.tiktok.com/tag/tiktokban?refer=embed">#tiktokban</a> <a title="tiktokshutdown2024" target="_blank" href="https://www.tiktok.com/tag/tiktokshutdown2024?refer=embed">#tiktokshutdown2024</a> <a title="popculture" target="_blank" href="https://www.tiktok.com/tag/popculture?refer=embed">#popculture</a> <a title="creatorsoftiktok" target="_blank" href="https://www.tiktok.com/tag/creatorsoftiktok?refer=embed">#creatorsoftiktok</a> <a title="breakingnews" target="_blank" href="https://www.tiktok.com/tag/breakingnews?refer=embed">#breakingnews</a> <a title="usa" target="_blank" href="https://www.tiktok.com/tag/usa?refer=embed">#usa</a> <a title="fyp" target="_blank" href="https://www.tiktok.com/tag/fyp?refer=embed">#fyp</a> </p> <a target="_blank" title="♬ original sound - #Californiaguy" href="https://www.tiktok.com/music/original-sound-7343426921717435179?refer=embed">♬ original sound - #Californiaguy</a> </section> </blockquote> <script async="" src="https://www.tiktok.com/embed.js"></script>
</div>
<p id="eA4fU1">Lawmakers in both parties <a href="https://x.com/Olivia_Beavers/status/1765813340310270456?s=20">didn’t take kindly to the impromptu lobbying frenzy</a>. Some characterized it as confirmation of their fears that the Chinese-owned app — which is <a href="https://www.cbsnews.com/news/tiktok-banned-us-government-where-else-around-the-world/">already banned on government devices</a> — is brainwashing America. The overrun phone lines were merely “making the case” for the bill, Rep. Chip Roy (R-TX) <a href="https://x.com/chiproytx/status/1765779738931404936?s=20">wrote on X</a>.</p>
<p id="h4SS41">The bill passed the House Wednesday with a vote of 352-65, well above the two-thirds majority threshold required. The White House has backed the bill from the beginning, reportedly providing <a href="https://punchbowl.news/article/gallagher-krishnamoorthi-tiktok-bill-gets-white-house-backing/">technical support</a> to legislators when they were drafting it (even as Biden’s reelection campaign has <a href="https://www.cnbc.com/2024/02/12/biden-campaign-debuts-official-tiktok.html">started using TikTok for voter outreach</a>).</p>
<p id="2k00R9">Though the bill now has momentum, there’s the crucial question of whether it would survive legal scrutiny even if passed. A federal court recently overturned a <a href="https://www.vox.com/politics/2023/5/18/23728598/montana-tiktok-ban-bytedance-china-enforcement">Montana law</a> that sought to ban TikTok. Though legislators sponsoring the US House bill argue that it is narrow in scope and would not amount to a total ban on TikTok that would violate the First Amendment, some legal experts believe otherwise. </p>
<p id="VwKJnD">“In my view, this loaded gun is a ban in all but name, and banning TikTok is obviously unconstitutional,” said Ramya Krishnan, a staff attorney at the Knight First Amendment Institute at Columbia University. “This ban on TikTok is materially the same [as the Montana ban] in all the ways that matter.”</p>
<h3 id="sAmq0J">Can Congress ban TikTok?</h3>
<p id="KFoltL">The constitutional law here appears straightforward: Congress can’t outright ban TikTok or any social media platform unless it can prove that it poses legitimate and serious privacy and <a href="https://www.vox.com/defense-and-security" data-source="encore">national security</a> concerns that can’t be addressed by any other means. The bar for such a justification is necessarily very high in order to protect Americans’ First Amendment rights, Krishnan said. </p>
<p id="oS9okz">Lawmakers argue that the bill under consideration isn’t actually a total ban. Rather, it would enact a new authority to ban apps in “narrowly defined situations” when they are controlled by a foreign adversary, New Jersey Rep. Frank Pallone, the ranking Democrat on the Energy and Commerce Committee, <a href="https://democrats-energycommerce.house.gov/media/press-releases/pallone-remarks-markup-legislation-protect-americans-data-and-national">said</a> before the committee Thursday. He compared the bill to historical efforts to prevent foreign ownership of US airwaves due to national security concerns. </p>
<p id="MPb4Q4">“It is no different here, and I take the concerns raised by the intelligence community very seriously,” he said.</p>
<p id="sTj8bU">Other House lawmakers have <a href="https://selectcommitteeontheccp.house.gov/media/press-releases/gallagher-bipartisan-coalition-introduce-legislation-protect-americans-0">criticized TikTok</a> for attempting to portray the bill as a total ban.</p>
<p id="Yp9W97">But legal experts say that an indirect ban may still be unconstitutional under the First Amendment. Civil society groups including the American Civil Liberties Union and the Center for Democracy and Technology (CDT) <a href="https://cdt.org/wp-content/uploads/2024/03/Coalition_Letter_Opposing_H.R._7521.pdf">wrote in a recent letter to federal lawmakers</a> that jeopardizing access to TikTok — “home to massive amounts of protected speech and association” — also “jeopardizes access to free expression.” There are also arguably less restrictive and more effective means of protecting any national security interests at stake in this bill, they asserted, considering the Chinese government could continue to access Americans’ data in other ways. </p>
<p id="ed5jtm">“This bill would functionally ban the distribution of TikTok in the United States, and would grant the President broad new powers to ban other social media platforms based on their country of origin,” they said in the letter. </p>
<p id="ntfZF5">Many experts believe it is unlikely that the government will be able to meet the high standard to prove that TikTok poses privacy and national security concerns that can’t otherwise be resolved, said Kate Ruane, director of CDT’s Free Expression Project. Lawmakers have publicly cited concerns about the Chinese government using the app to <a href="https://www.wired.com/story/china-espionage-tiktok-spying-national-security/">spy on Americans</a> and to spread propaganda that could be used to <a href="https://time.com/6836078/tiktok-sold-banned-2024-election/">influence the 2024 presidential election</a>. </p>
<p id="ANB3Yj">Though TikTok has repeatedly insisted that it has never shared user data with the Chinese government nor been asked to do so, a former employee of ByteDance has alleged in court that the government had nevertheless <a href="https://www.cnn.com/2023/06/08/tech/tiktok-data-china/index.html">accessed such data</a> on a widespread basis for political purposes during the 2018 protests in Hong Kong. And in December, TikTok parent company ByteDance acknowledged it had fired four employees who <a href="https://www.washingtonpost.com/technology/2022/12/22/tiktoks-chinese-owner-fires-workers-who-gathered-data-journalists/">accessed the data of two journalists</a> while trying to track down an internal leaker.</p>
<p id="5ht40f">But so far, members of Congress have not provided concrete proof for their claims about Chinese digital espionage and seem to have little interest in offering any transparency: Before the committee voted to advance the bill Thursday, lawmakers had a <a href="https://www.reuters.com/technology/new-push-congress-ban-tiktok-or-force-chinese-divestiture-gains-steam-2024-03-07/">closed-door classified briefing</a> on national security concerns associated with TikTok. </p>
<p id="SATlU3">“TikTok is Communist Chinese malware that is poisoning the minds of our next generation and giving the CCP unfettered access to troves of Americans’ data,” Rep. Elise Stefanik (R-NY) said in a statement. “We cannot allow the CCP to continue to harness this digital weapon.”</p>
<p id="89RyJb">However, national security experts have also questioned the rationale behind a ban. Mike German, a former FBI special agent and fellow at the Brennan Center for Justice’s Liberty and National Security Program, <a href="https://www.aljazeera.com/economy/2023/3/28/bid-to-ban-tiktok-raises-hypocrisy-charge-amid-global-spying#:~:text=%E2%80%9CWhen%20somebody%20puts%20the%20TikTok,Security%20Program%2C%20told%20Al%20Jazeera">told Al Jazeera</a> that, like many American apps, TikTok collects data on its users that a foreign government could theoretically use for its own hostile purposes. But those governments could just as well buy Americans’ data on a legitimate open market, where the sale of that data remains unrestricted. </p>
<p id="TWcUp1">And even if lawmakers did provide more evidence of national security concerns, it’s still not clear that the ban would pass legal muster. </p>
<p id="BIj1FU">Courts have already applied strict scrutiny to previous attempts to ban TikTok. A federal judge <a href="https://www.npr.org/2023/11/30/1205735647/montana-tiktok-ban-blocked-state">blocked the Montana TikTok ban</a> — which also imposed a financial penalty on TikTok and any app store hosting it each time a user accesses or is offered the ability to access the app — before it was scheduled to go into effect in November. </p>
<p id="1h8j3M">Montana lawmakers justified the ban as a means of protecting the privacy interests of consumers in the state. But US District Judge Donald Molloy wrote in his ruling that the law overstepped the Montana legislature’s powers and left “little doubt that Montana’s legislature and Attorney General were more interested in targeting <a href="https://www.vox.com/china" data-source="encore">China</a>’s ostensible role in TikTok than with protecting Montana consumers.”</p>
<p id="RgSXhf">Former <a href="https://www.vox.com/donald-trump" data-source="encore">President Donald Trump</a> also twice tried to ban TikTok via executive action, only for courts to <a href="https://www.npr.org/2020/12/07/944039053/u-s-judge-halts-trumps-tiktok-ban-the-2nd-court-to-fully-block-the-action">strike down his proposal</a> both times. However, he changed his tune Thursday, arguing that banning TikTok would <a href="https://www.axios.com/2024/03/08/trump-claims-tiktok-ban-would-only-help-enemy-facebook">benefit Meta CEO Mark Zuckerberg and Facebook</a>, which he referred to in a post on his social media platform Truth Social as a “true enemy of the people.” </p>
<h3 id="u3nO0q">What lawmakers could do instead of banning TikTok</h3>
<p id="AoqlaZ">If lawmakers are serious about protecting privacy and national security, Ruane said, they should instead pass comprehensive digital privacy legislation. </p>
<p id="fO55de">“That would be a better path forward,” she said. </p>
<p id="oEPPqh">Her organization, the Center for Democracy and Technology, has supported a bipartisan bill that <a href="https://energycommerce.house.gov/posts/bipartisan-ec-leaders-hail-committee-passage-of-the-american-data-privacy-and-protection-act">passed a committee vote</a> in 2022: the <a href="https://www.commerce.senate.gov/services/files/9BA7EF5C-7554-4DF2-AD05-AD940E2B3E50">American Data Privacy and Protection Act</a>. It included provisions requiring companies to allow consumers to consent to or reject the collection of their data, to allow consumers to download and delete the data being collected on them, to require consumers’ affirmative consent to share that data with a third party, and more. </p>
<p id="fEyVaf">It was the culmination of a decades-long effort to regulate the collection, use, and sale of consumer data, similar to the <a href="https://www.vox.com/european-union" data-source="encore">European Union</a>’s <a href="https://gdpr.eu/what-is-gdpr/">regulatory efforts</a>. It would have <a href="https://www.law.umaryland.edu/content/articles/name-659578-en.html">tasked</a> the Federal Trade Commission and state attorneys general with enforcing the law and preempted the patchwork of privacy laws that have been enacted at the state level in the absence of comprehensive federal legislation. </p>
<p id="Kuqszf">However, the privacy bill stalled in Congress and was not reintroduced; Ruane said it’s unclear why. Now lawmakers are moving forward instead with the bill that could ban TikTok — without solving the underlying <a href="https://www.vox.com/privacy">privacy concerns</a>.</p>
<p id="2ePb4a">“This bill would fail to protect us from the many threats to our digital privacy posed by criminals, private companies, and foreign actors,” said David Greene, civil liberties director at the Electronic Frontier Foundation. “Comprehensive data privacy legislation is the solution we need — not bans of certain categories of apps.”</p>
<p id="UPAhJN"><em><strong>Update, March 13, 11:30 am: </strong></em><em>This story, originally published March 9, has been updated multiple times, most recently with additional reporting on the bill’s progression in the House and Senate.</em></p>
https://www.vox.com/politics/24094839/tiktok-ban-bill-congress-pass-bidenNicole Narea2024-03-13T06:30:00-04:002024-03-13T06:30:00-04:00Why can’t anyone agree on how dangerous AI will be?
<figure>
<img alt="An illustration of a head shown in profile, with cogs and wires overlayed and extending from it. " src="https://cdn.vox-cdn.com/thumbor/fav3HHzvAH0fABk3oDmBAMLES7U=/240x0:1680x1080/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73203250/AI_LEAD.0.jpeg" />
<figcaption>Javier Zarracina/Vox</figcaption>
</figure>
<p>Researchers tried to get AI optimists and pessimists on the same page. It didn’t quite work.</p> <p id="pZWqcc">I’ve <a href="https://www.vox.com/future-perfect/23794855/anthropic-ai-openai-claude-2">written</a> a <a href="https://www.vox.com/future-perfect/23775650/ai-regulation-openai-gpt-anthropic-midjourney-stable">lot</a> <a href="https://www.vox.com/future-perfect/2024/2/7/24062374/ai-openai-anthropic-deepmind-legal-liability-gabriel-weil">about AI</a> and the debate over whether it could <a href="https://www.vox.com/2015/8/10/9124145/effective-altruism-global-ai">kill us all</a>. But I still don’t really know where I come down.</p>
<p id="DXeY4i">There are people who deeply understand advanced machine learning systems who think they will prove <a href="https://www.nytimes.com/2023/05/01/technology/ai-google-chatbot-engineer-quits-hinton.html">increasingly uncontrollable</a>, possibly <a href="https://yoshuabengio.org/2023/05/22/how-rogue-ais-may-arise/">“go rogue,”</a> and <a href="https://data.berkeley.edu/news/stuart-russell-testifies-ai-regulation-us-senate-hearing">threaten humanity with catastrophe</a> or even extinction. There are other people who deeply understand how these systems work who think that we’re <a href="https://www.wired.com/story/artificial-intelligence-meta-yann-lecun-interview/">perfectly able to control them</a>, that their <a href="https://www.schumer.senate.gov/imo/media/doc/Andrew%20Ng%20-%20Statement.pdf">dangers do not include human extinction</a>, and that the first group is full of <a href="https://pedromdd.medium.com/ais-greatest-risk-is-not-having-enough-of-it-500475a69a52">hysterical alarmists</a>.</p>
<p id="vcXKFp">How do we tell who’s right? I sure don’t know. </p>
<p id="KeAVja">But a <a href="https://forecastingresearch.org/news/ai-adversarial-collaboration">clever new study</a> from the Forecasting Research Institute tries to find out. The authors (Josh Rosenberg, Ezra Karger, Avital Morris, <a href="https://www.vox.com/future-perfect/2024/2/13/24070864/samotsvety-forecasting-superforecasters-tetlock">Molly Hickman</a>, Rose Hadshar, Zachary Jacobs, and forecasting godfather <a href="https://www.vox.com/2015/8/20/9179657/tetlock-forecasting">Philip Tetlock</a>) had previously asked both experts on <a href="https://www.vox.com/2023/4/28/23702644/artificial-intelligence-machine-learning-technology" data-source="encore">AI</a> and other existential risks, and “superforecasters” with a demonstrated track record of successfully predicting world events in the near term, to assess the danger that AI poses. </p>
<p id="gTPaaO">The result? The <a href="https://www.vox.com/future-perfect/23785731/human-extinction-forecasting-superforecasters">two groups disagreed a lot</a>. The experts in the study were in general much more nervous than the superforecasters and put far higher odds on disaster.</p>
<p id="DE1VcN">The researchers wanted to know why these groups disagreed so profoundly. So the authors set up an “adversarial collaboration”: They had the two groups spend many hours (a median of 31 hours for the experts, 80 hours for the superforecasters) reading new materials and, most importantly, discussing these issues with people of the opposite view with a moderator. The idea was to see if exposing each group to more information, and to the best arguments of the other group, would get either to change their minds. </p>
<p id="KOoyWz">The researchers were also curious to find “cruxes”: issues that help explain people’s beliefs, and on which new information might change their mind. One of the biggest cruxes, for example, was “Will <a href="https://metr.org/">METR</a> [an AI evaluator] or a similar organization find evidence of AI having the ability to autonomously replicate, acquire resources, and avoid shutdown before 2030?” If the answer turns out to be “yes,” skeptics (the superforecasters) say they will become more worried about AI risks. If the answer turns out to be “no,” AI pessimists (the experts) say they will become more sanguine.</p>
<p id="HBWRzT">So did everyone just converge on the correct answer? … No. Things were not destined to be that easy. </p>
<p id="qBAWJZ">The AI pessimists adjusted their odds of a catastrophe before the year 2100 down from 25 to 20 percent; the optimists moved theirs up from 0.1 to 0.12 percent. Both groups stayed close to where they started.</p>
<p id="xVSdMZ">But the report is fascinating nonetheless. It’s a rare attempt to bring together smart, well-informed people who disagree. While doing so didn’t <em>resolve</em> that disagreement, it shed a lot of light on where those points of division came from.</p>
<h3 id="Ed7dxM">Why people disagree about AI’s dangers</h3>
<p id="3Ra4s3">The paper focuses on disagreement around AI’s potential to either wipe humanity out or cause an “unrecoverable collapse,” in which the human population shrinks to under 1 million for a million or more years, or global GDP falls to under $1 trillion (less than 1 percent of its <a href="https://data.worldbank.org/indicator/NY.GDP.MKTP.CD">current value</a>) for a million years or more. At the risk of being crude, I think we can summarize these scenarios as “extinction or, at best, hell on earth.”</p>
<p id="buoypI">There are, of course, a number of other different risks from AI worth worrying about, many of which we already face today. </p>
<p id="blHJ9v">Existing AI systems sometimes exhibit worrying <a href="https://www.vox.com/technology/23738987/racism-ai-automated-bias-discrimination-algorithm">racial and gender biases</a>; they can be <a href="https://www.economist.com/science-and-technology/2024/02/28/ai-models-make-stuff-up-how-can-hallucinations-be-controlled">unreliable</a> in ways that cause problems when we rely upon them anyway; they can be used to bad ends, like <a href="https://www.vox.com/technology/23746060/ai-generative-fake-images-photoshop-google-microsoft-adobe">creating fake news clips to fool the public</a> or <a href="https://www.vox.com/2020/6/8/21284005/urgent-threat-deepfakes-politics-porn-kristen-bell">making pornography with the faces of unconsenting people</a>. </p>
<p id="h2z4vt">But these harms, while surely bad, obviously pale in comparison to “losing control of the AIs such that everyone dies.” The researchers chose to focus on the extreme, existential scenarios.</p>
<p id="BFvW5B">So why do people disagree on the chances of these scenarios coming true? It’s not due to differences in access to information, or a lack of exposure to differing viewpoints. If it were, the adversarial collaboration, which consisted of massive exposure to new information and contrary opinions, would have moved people’s beliefs more dramatically.</p>
<p id="wWety3">Nor, interestingly, was much of the disagreement explainable by different beliefs about what will happen with AI in the next few years. When the researchers paired up optimists and pessimists and compared their odds of catastrophe, their average gap in odds was 22.7 percentage points. The most informative “crux” (an AI evaluator finding that a model had highly dangerous abilities before 2030) only reduced that gap by 1.2 percentage points. </p>
<p id="L4RlWk">Short-term timelines are not nothing, but it’s just not where the main disagreements are.</p>
<p id="hy66Jn">What did seem to matter were different views on the <em>long-term</em> future. The AI optimists generally thought that human-level AI will take longer to build than the pessimists believed. As one optimist told the researchers, “language models are just that: models of language, not digital hyperhumanoid Machiavellis working to their own end”; this optimist thought that fundamental breakthroughs in methods of machine learning are necessary to reach human-level intelligence.</p>
<p id="MJtexC">Many cited the need for robotics to reach human levels, not just software AI, and argued that doing so would <a href="https://www.therobotreport.com/10-biggest-challenges-in-robotics/">be much harder</a>. It’s one thing to write code and text in a laptop; it’s quite another to, as a machine, learn how to flip a pancake or clean a tile floor or any of the many other physical tasks at which humans now outperform <a href="https://www.vox.com/robots" data-source="encore">robots</a>.</p>
<h3 id="ydKgLh">When disputes go deep</h3>
<p id="ALLETN">The most interesting source of splits the researchers identified was what they call “fundamental worldview disagreements.” That’s a fancy way of saying that they disagree about where the burden of proof lies in this debate. </p>
<p id="hvLjUq">“Both groups agree that ‘extraordinary claims require extraordinary evidence,’ but they disagree about which claims are extraordinary,” the researchers summarize. “Is it extraordinary to believe that AI will kill all of humanity when humanity has been around for hundreds of thousands of years, or is it extraordinary to believe that humanity would continue to survive alongside smarter-than-human AI?”</p>
<p id="xV8zra">It’s a fair question! My experience is that most laypeople outside AI find “the machines will kill us all” to be the more extraordinary claim. But I can see where the pessimists are coming from. Their basic view is that the emergence of superhuman AI is like the arrival on earth of a superhuman alien species. We don’t know if that species would want to kill us all. </p>
<p id="lq8Mmk">But Homo sapiens didn’t necessarily <em>want</em> to kill all the Homo erectus or Neanderthals hundreds of thousands of years ago, when multiple intelligent species of large apes were coexisting. We did, however, kill them all off. </p>
<p id="vqxkBU">Extinction tends to happen to dumber, weaker species when a smarter species that’s better at claiming resources for itself emerges. If you have this worldview, the burden of proof is on optimists to show why super-intelligent AI <em>wouldn’t</em> result in catastrophe. Or, as one pessimist in the study put it: “there are loads of ways this could go and very few of them leave humans alive.”</p>
<p id="YDoJ5Q">This is not the most encouraging conclusion for the study to arrive at. A disagreement driven by concrete differences of opinion about what will happen in the next few years is an easier disagreement to resolve — one based on how the next few years proceed rather than on deep, hard-to-change differences in people’s assumptions about how the world works, and about where the burden of proof should fall.</p>
<p id="pevHOw">The paper reminded me of an interview I saw a long time ago with the late philosopher Hilary Putnam. Working in the late 20th century, Putnam was a believer that philosophy could make progress, even if the big questions — What is truth? How does the mind work? Is there an external reality we can grasp? — feel as hard to answer as ever. </p>
<p id="louzhQ">Sure, we don’t know those answers, <a href="https://www.youtube.com/watch?v=GlrEbffVVjM">Putnam said</a>. But we know more about the questions. “We learn more about how difficult they are, and why they’re so difficult. That is maybe the lasting philosophical progress.”</p>
<p id="jlOjyJ">That’s how I felt reading the Forecasting Research Institute paper. I don’t feel like I know for sure how worried to be about AI. But I do feel like I know more about why this is a hard question.</p>
<p id="nEI9ta"><em>A version of this story originally appeared in the </em><a href="https://www.vox.com/future-perfect"><em><strong>Future Perfect</strong></em></a><em> newsletter. </em><a href="https://www.vox.com/pages/future-perfect-newsletter-signup"><em><strong>Sign up here!</strong></em></a></p>
https://www.vox.com/future-perfect/2024/3/13/24093357/ai-risk-extinction-forecasters-tetlockDylan Matthews2024-03-11T06:00:00-04:002024-03-11T06:00:00-04:00The NRA has a case before the Supreme Court that it absolutely should win
<figure>
<img alt="A close-up photo of a man wearing a navy baseball cap embroidered with the letters NRA in yellow. The US Supreme Court building is visible in the background." src="https://cdn.vox-cdn.com/thumbor/gPySvLHkqNKByb9Qks3V-TYEWsg=/224x0:2776x1914/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73197304/GettyImages-80288328.0.jpg" />
<figcaption>A man wears an NRA hat in front of the US Supreme Court building. | Mark Wilson/Getty Images</figcaption>
</figure>
<p>It’s one of two cases asking whether the government is allowed to speak freely to private companies.</p> <p id="gIAx7F">The <a href="https://www.vox.com/scotus" data-source="encore">Supreme Court</a> will hear two cases on Monday, March 18, that are often referred to as “jawboning” cases — that is, cases where the government tried to pressure private companies into taking certain actions, but without actually using any of its coercive power.</p>
<p id="K5rolX">On the surface, the two cases have many similarities. Both involve claims that the First Amendment imposes strict limits on the government’s ability to cajole, hector, or otherwise try to persuade private companies to act in a particular way. But these similarities are only an inch deep.</p>
<p id="kSPAlj">One case, known as <a href="https://www.scotusblog.com/case-files/cases/national-rifle-association-of-america-v-vullo/"><em>National Rifle Association v. Vullo</em></a><em>, </em>involves a <a href="https://www.vox.com/scotus/23948899/supreme-court-nra-vullo-new-york-insurance-murder-carry-guard">fairly obvious violation of the First Amendment</a>. </p>
<p id="TR8FNO">In <em>National Rifle Association (NRA),</em> New York’s top financial regulator brought a legitimate enforcement action against three insurance companies that did business with the NRA, and these companies agreed to pay a total of $13 million in fines. But then, while this completely benign enforcement action was underway, the same regulator issued a “<a href="https://www.dfs.ny.gov/industry_guidance/industry_letters/il20180419_guidance_risk_mgmt_nra_NRA_similar_gun_promotion_orgs_banking_industry">guidance</a>” to all insurers who do business in the state, warning them to “continue evaluating and managing their risks, including reputational risks, that may arise from their dealings with the NRA or similar gun promotion organizations.”</p>
<p id="GZtkBq">This guidance was not permitted by the Constitution. A law enforcement agency cannot pressure companies to stop doing business with a political advocacy group that it disagrees with while it is also in the process of collecting millions of dollars in fines from some of those companies. Even though the financial regulator’s “guidance” did not explicitly threaten that more enforcement actions would follow if insurers kept doing lawful business with the NRA, a threat was implied by the regulator’s previous, multimillion-dollar action against insurers associated with the NRA.</p>
<p id="1wJ6Zg">The second case, meanwhile, is more fraught. In <a href="https://www.scotusblog.com/case-files/cases/murthy-v-missouri-3/"><em>Murthy v. Missouri</em></a>, the United States Court of Appeals for the Fifth Circuit — a <a href="https://www.vox.com/policy-and-politics/2022/12/27/23496264/supreme-court-fifth-circuit-trump-court-immigration-housing-sexual-harrassment">far-right court dominated by MAGA judges</a> — issued a <a href="https://www.vox.com/scotus/2023/9/22/23883888/supreme-court-social-media-first-amendment-netchoice-paxton-murthy-missouri-twitter-facebook">vague and sweeping injunction</a> that effectively forbids the federal government from communicating with social media companies about harmful content online.</p>
<p id="CbnQ0T">Federal officials routinely ask social media companies to pull down content that encourages criminal activity, that endangers <a href="https://www.vox.com/public-health" data-source="encore">public health</a>, or that potentially threatens <a href="https://www.vox.com/defense-and-security" data-source="encore">national security</a>. This includes content that seeks to <a href="https://www.vox.com/policy-and-politics/2023/2/16/23582848/supreme-court-internet-section-230-terrorism-cases-gonzalez-google-twitter-taamneh">recruit people into terrorist groups such as ISIS</a>, content produced by “<a href="https://www.vox.com/scotus/2023/9/22/23883888/supreme-court-social-media-first-amendment-netchoice-paxton-murthy-missouri-twitter-facebook">Russian troll farms</a>” and other foreign adversaries, and disinformation about vaccinations and <a href="https://www.vox.com/coronavirus-covid19" data-source="encore">Covid-19</a>.</p>
<p id="MkG8X5">Unlike the <em>NRA</em> case, the Fifth Circuit’s opinion in <em>Murthy</em> identified no examples of a government official taking coercive action against a social media company, nor did it identify any plausible threat to take such action. The right-wing court did, however, identify some cases where White House officials used unnecessarily strident language in an attempt to pressure social media companies to remove content. </p>
<p id="sXMsci">It then latched onto these few unwise statements as an excuse to issue a sweeping court order that practically forbids the government from speaking to social media companies at all. The Supreme Court <a href="https://www.vox.com/scotus/2023/9/22/23883888/supreme-court-social-media-first-amendment-netchoice-paxton-murthy-missouri-twitter-facebook">temporarily blocked this order</a> last October over the dissents of Justices Clarence Thomas, Samuel Alito, and Neil Gorsuch.</p>
<p id="PCsjxk">Under current law, the government may ask any private company to take virtually any action. It may use firm, or even harsh, language when it does so. It may condemn a private company’s actions in public and in private. And it may do so even if the company the government is trying to pressure is a media company that engages in First Amendment-protected speech.</p>
<p id="xhhN78">As any reporter will tell you, government officials try to shape what kind of reporting appears in the newspaper all the time. Most major government offices have entire communications teams that exist for the purpose of trying to persuade, pressure, or cajole journalists into reporting some stories while avoiding others.</p>
<p id="cblhMh">But the government may not use the power of the state to coerce a media outlet into pulling down speech. Nor may it use its coercive power to pressure companies to cease doing business with an advocacy group that the government finds repugnant. Persuasion is allowed. Coercion is not.</p>
<p id="YsEvik">Now, however, the task of policing this line between persuasion and coercion falls to a Supreme Court that <a href="https://www.vox.com/scotus/2023/7/8/23784320/supreme-court-2022-term-affirmative-action-religion-voting-rights-abortion">does not always feel bound by existing law</a> and that is often sympathetic to the Republican Party’s cultural grievances. So there is no guarantee that this Supreme Court will draw the line between permissible and impermissible pressure in a sensible place. </p>
<p id="paVcoL">And both the <em>NRA</em> and the <em>Murthy</em> cases potentially endanger the government’s ability to take entirely legitimate actions to protect public safety and public health.</p>
<h3 id="awHBiA">In the <em>NRA</em> case, a former top official’s recklessness endangers a perfectly legitimate effort to keep people from being murdered</h3>
<p id="CQ9F0w">The <em>NRA</em> case arises out of two separate actions by former New York State Department of Financial Services (DFS) superintendent Maria Vullo, one of which raises no serious constitutional concerns.</p>
<p id="XZ6Fex">The legitimate action began in 2017, when DFS opened an investigation into Carry Guard, an NRA-endorsed insurance program that, according to a federal appeals court, “provided liability defense coverage for criminal proceedings resulting from firearm use <a href="https://casetext.com/case/natl-rifle-assn-of-am-v-vullo">even where the insured acted with criminal intent</a>.”</p>
<p id="V8P7nP">Carry Guard offered to pay both the civil and criminal legal fees (up to $1 million for a civil case and up to $150,000 for a criminal case) of customers who shot someone else, allegedly in self-defense. The NRA advertised Carry Guard to its members as a way to <a href="https://www.vox.com/scotus/23948899/supreme-court-nra-vullo-new-york-insurance-murder-carry-guard">enjoy peace of mind if they shot another human</a>. One pitch to NRA members said that “you should never be forced to choose between defending your life ... and putting yourself and your family in financial ruin.”</p>
<p id="7uWHTw">New York law generally does not permit insurance products that “<a href="https://www.dfs.ny.gov/insurance/ogco2002/rg205301.htm">insure a person for that person’s intentional criminal acts</a>,” and so DFS acted entirely within its lawful mandate when it brought an enforcement action against three companies that underwrote or administered Carry Guard or similar programs. Eventually, these companies entered into consent decrees where they agreed to stop providing this sort of insurance, and where they also agreed to pay over $13 million in fines.</p>
<p id="jdbTxy">While these entirely legitimate enforcement actions were ongoing, however, a teenager <a href="https://www.supremecourt.gov/DocketPDF/22/22-842/300966/20240220164549605_Vullo%20Merits%20Brief%20Final.pdf">opened fire at a high school in Parkland, Florida, killing 17 people</a>. In response to this shooting, then-New York Gov. Andrew Cuomo’s administration <a href="https://newrepublic.com/article/150933/andrew-cuomos-trumpian-war-nra">stepped up its criticism of the NRA</a>, and Vullo played a major part in the administration's attempt to rein in the powerful gun rights group.</p>
<p id="kUCtkc">And so, in April of 2018, DFS issued a “guidance,” signed by Vullo, warning insurers of the “risks, including reputational risks, that may arise from their dealings with the NRA or similar gun promotion organizations.” That guidance told insurance companies that DFS “encourages regulated institutions to <a href="https://www.scribd.com/document/682624756/NRA-Complaint">review any relationships they have with the NRA</a> or similar gun promotion organizations, and to take prompt actions to managing these risks and promote public health and safety.”</p>
<p id="ANE6W1">Had this guidance been issued in a vacuum, it might not violate the Constitution. Again, the government may attempt to persuade a private company to stop doing business with the NRA.</p>
<p id="sN36si">But a law enforcement officer cannot seek millions of dollars in fines from three insurance companies that did business with a political advocacy group — even if they collected those fines for entirely legitimate reasons — and then immediately turn around and warn every other insurance company in the state that bad things could happen to them if they also do business with that same advocacy group. Read in context, it’s hard to understand Vullo’s guidance as anything other than an implicit threat.</p>
<p id="3Nsg83">One highly relevant case is <a href="https://scholar.google.com/scholar_case?case=7402288339517306664&hl=en&as_sdt=6&as_vis=1&oi=scholarr"><em>Bantam Books v. Sullivan</em></a> (1963), where the Supreme Court acknowledged that “people do not lightly disregard public officers’ thinly veiled threats to institute criminal proceedings against them if they do not come around.”</p>
<p id="14hVEm"><em>Bantam Books</em> involved the Orwellianly named “Rhode Island Commission to Encourage Morality in Youth,” a state body that identified books and other publications it deemed “objectionable for sale, distribution or display to youths under 18 years of age.” This commission then sent letters to booksellers seeking their “cooperation” in removing such books and informing those sellers of the commission's “duty to recommend to the Attorney General prosecution of purveyors of obscenity.”</p>
<p id="ep490A">At least one book distributor reported that he was visited by a police officer shortly after receiving a notice from the commission, and the officer asked the distributor what steps he had taken to comply with the notice.</p>
<p id="eXgrwK">In any event, neither the commission nor the police officer explicitly stated “remove these books or you will be arrested,” but the implicit threat was quite clear. And <em>Bantam Books</em> held that this kind of coercion is not allowed under the First Amendment.</p>
<p id="4uzP5m">Vullo’s actions weren’t quite as egregious as the commission’s actions in <em>Bantam Books</em> — at least she didn’t send an armed state official to New York insurers to ask them what they’d done to cut ties with the NRA — but they are close enough to the facts of <em>Bantam Books</em> that they cannot be allowed. </p>
<p id="yf26a4">Vullo, after all, was herself a law enforcement officer (though, admittedly, one whose jurisdiction was limited to enforcing certain financial laws), and she was actively trying to collect millions of dollars in fines from three insurers that did business with the NRA when she told every other insurer in the state to stop doing business with the NRA!</p>
<p id="EbnwUr">Worse, because of Vullo’s foolish decision to issue her anti-NRA “guidance,” she may have endangered a perfectly legitimate enforcement action against the NRA’s murder insurance. The Supreme Court’s GOP-appointed supermajority is <a href="https://www.vox.com/scotus/23948899/supreme-court-nra-vullo-new-york-insurance-murder-carry-guard">extremely sympathetic</a> to claims brought by gun rights groups. There’s no telling how far they might reach when they are handed a case that involves a genuine violation of the NRA’s constitutional rights.</p>
<p id="60oAkq">More broadly, Democrats simply need to be more careful than Vullo was when they exercise government authority. They have to assume that every action they take will be reviewed by a judiciary that hates progressive policies and wants to see Democrats fail.</p>
<p id="nAC6OQ">And that brings us to the <em>Murthy</em> case.</p>
<h3 id="Kp1hsd">The <em>Murthy</em> case shows just how much harm partisan judges can impose on the US government if they are given an excuse to do so</h3>
<p id="h4zP6l">The <em>Murthy</em> case is as frustrating as the <em>NRA</em> case, but for a completely different reason. Unlike <em>NRA</em>, <em>Murthy</em> does not involve the kind of clear-cut violation of the First Amendment that should compel the courts to intervene. </p>
<p id="RXtxgp">At most, <em>Murthy</em> seems to involve a few isolated incidents where government officials — most likely because they were under extraordinary pressure to vaccinate hundreds of millions of Americans against Covid-19 as fast as possible — lost their tempers with corporate officials who shared the same broad goals as the <a href="https://www.vox.com/joe-biden" data-source="encore">Biden administration</a>.</p>
<p id="AIpWuX">As the Justice Department explains in its <a href="https://www.supremecourt.gov/DocketPDF/23/23-411/293780/20231219192259919_23-411ts%20Murthy.pdf"><em>Murthy</em> brief</a>, the federal government routinely speaks with social media companies about their editorial decisions. The Department of Homeland Security often briefs social media companies on how to “‘recognize and react to violent extremist content’ posted by terrorist groups”; the FBI often notifies social media providers that one of their users is promoting terrorism; the Cybersecurity and Infrastructure Security Agency (CISA) flags social media content that contains election-related disinformation, “such as false statements about the time, place, and manner of elections”; the White House sometimes asks social media companies to remove accounts that falsely impersonate a member of the president’s family; and government officials also ask social media companies to remove harmful medical advice that could injure people who follow it.</p>
<p id="f6CvXI">And so long as the government merely asks the platforms to remove content, rather than trying to threaten or coerce them into doing so, government officials may lawfully make such requests.</p>
<p id="FI482m">Social media platforms, moreover, are <a href="https://www.supremecourt.gov/DocketPDF/23/23-411/293780/20231219192259919_23-411ts%20Murthy.pdf">often eager participants in these conversations</a> because they share the government’s desire to, say, suppress terrorism or protect people from quack medical advice. During the pandemic, for example, these platforms “regularly reached out” to the CDC “to ensure that the information the social media companies chose to promote on their platforms remained consistent with the latest CDC guidance.”</p>
<p id="bO8HXo">Companies like <a href="https://www.vox.com/facebook" data-source="encore">Facebook</a> and <a href="https://www.vox.com/youtube" data-source="encore">YouTube</a>, in other words, decided on their own that they did not want to publish content that could lead to more Covid-related deaths. And they often proactively reached out to the government for advice on what sort of content should be removed.</p>
<p id="CghzFN">There are quite a few cases, however, where social media companies disagreed with the government’s opinion that a particular social media post should be taken down. And the plaintiffs in the <em>Murthy </em>case — two red states, plus a handful of individuals who are upset that some of their content was removed by social media companies — highlight a few examples of when these disagreements escalated into forceful rhetoric.</p>
<p id="8pTOfu">The <a href="https://www.supremecourt.gov/DocketPDF/23/23-411/299644/20240202144405984_2024-02-02%20-%20Murthy%20v.%20Missouri%20-%20Brief%20of%20Respondents%20-%20Final%20with%20Tables.pdf">plaintiffs’ brief</a>, for example, points to a communication between a White House official and <a href="https://www.vox.com/twitter" data-source="encore">Twitter</a> where the official asked Twitter to “get moving on the process for having” an anti-vaccine tweet “removed ASAP.” And they flag a few examples where government officials grew frustrated with the social media companies’ slow answers and responded with a commanding tone (“I want an answer on what happened here and I want it today”). </p>
<p id="dvhQK3">The plaintiffs also point to some isolated statements, made by White House officials who were responding to questions from reporters, where the Biden administration endorsed policy changes that social media companies would likely find objectionable, such as a May 2021 statement by the White House press secretary that President Biden supports <a href="https://www.vox.com/antitrust" data-source="encore">antitrust</a> reforms.</p>
<p id="7kFb7Q">Some of these statements were probably unwise. In a world where a hostile, highly partisan judiciary looms over every Democratic administration, Biden administration officials would be smart not to speak to employees at media companies as if they were their boss lest some judge latch onto these statements as an excuse to charge the administration with constitutionally impermissible coercion.</p>
<p id="Jz0Kmv">But the Fifth Circuit’s <em>Murthy</em> opinion, which spends about 14 pages describing incidents where various federal officials asked social media companies to remove content, does not identify a single example of any official taking coercive action against a platform that refused such a request. Nor does it identify anything vaguely resembling the implicit threat in the <em>NRA</em> case. </p>
<p id="Klbi6W">If anything, the Fifth Circuit’s own evidence suggests that the platforms felt perfectly free to ignore the government’s requests whenever they wanted to. Among other things, the Fifth Circuit admitted that, about half the time, the platforms <a href="https://www.vox.com/scotus/2023/9/22/23883888/supreme-court-social-media-first-amendment-netchoice-paxton-murthy-missouri-twitter-facebook">rejected the FBI’s requests to pull down content</a> and nothing happened to the platforms as a result.</p>
<p id="98WsUY">And yet, despite its inability to identify any actual coercion by government officials, the Fifth Circuit invented a new rule that would allow it to shut down more benign communications between these officials and social media companies. </p>
<p id="OrOe5E">And then it issued a sweeping injunction that is simultaneously too vague for Biden administration officials to figure out how to comply with it and so broad that it prohibits communications that clearly do not violate the First Amendment.</p>
<p id="J479Nt">If this injunction, which has been temporarily blocked by the Supreme Court, were to go into effect, that would mean that no one in the FBI would know what it is allowed to do if it discovers that the Russian government is flooding social media with content intended to incite an insurrection in the United States. And no one in the CDC would know if they are allowed to respond to a request from Facebook asking if the drug ivermectin cures Covid (<a href="https://www.vox.com/future-perfect/22663127/ivermectin-covid-treatments-vaccines-evidence">it does not</a>).</p>
<h3 id="eNR4YF">So what did the Fifth Circuit’s opinion actually say?</h3>
<p id="W1f1r7">Briefly, the Fifth Circuit invented a new legal standard that, it claims, should govern cases where the government asks a media company to voluntarily remove content. Under the Fifth Circuit’s approach, huge swaths of the federal government’s communications with social media companies are illegal because the government “<a href="https://www.vox.com/scotus/2023/9/22/23883888/supreme-court-social-media-first-amendment-netchoice-paxton-murthy-missouri-twitter-facebook">entangled themselves in the platforms’ decision-making processes</a>.”</p>
<p id="4ZPEcM">The Fifth Circuit never defines the word “entangled” beyond using other, equally vague adjectives to describe what the government is not allowed to do. The opinion, for example, faults the government for having “consistent and consequential” communications with social media platforms.</p>
<p id="AkJedy">Then, having articulated this imprecise legal rule, the Fifth Circuit issued a broad injunction prohibiting the Biden administration from ... well, it’s not at all clear what the Biden administration can’t do. The federal government seems to have been ordered not to have “consistent and consequential” communications with social media companies — whatever the hell that means.</p>
<p id="AoG9CJ">This is, to say the least, not normal behavior from a federal court. Typically, when a court enjoins any party from taking any action, it defines the scope of that injunction clearly enough that it’s possible to figure out what the enjoined party is not allowed to do. But the Fifth Circuit’s injunction is so ill-defined that no one in the government can possibly figure out whether they violate it if they speak to a social media company. So government officials are likely to cut off contact altogether, lest they be held in contempt of court.</p>
<p id="UROY5m">In any event, if the Supreme Court wants to follow existing law — always an <a href="https://www.vox.com/scotus/2023/7/8/23784320/supreme-court-2022-term-affirmative-action-religion-voting-rights-abortion">uncertain proposition with this Court</a> — there is an easy way for the justices to toss out the Fifth Circuit’s decision without blessing some of the more imprudent communications between the social media platforms and government officials. </p>
<p id="Qh7shr">Any plaintiff who brings a federal lawsuit must show that they’ve been injured in some way by the defendant they are suing — a <a href="https://www.vox.com/2021/6/17/22538462/supreme-court-obamacare-california-texas-stephen-breyer-standing-individual-mandate-constitution">requirement known as “standing.”</a> As the Supreme Court said in <a href="https://supreme.justia.com/cases/federal/us/504/555/"><em>Lujan v. Defenders of Wildlife</em></a> (1992), moreover, “there must be a causal connection between the injury and the conduct complained of.” </p>
<p id="YSQ5Pa">So a plaintiff alleging that a White House official was too demanding in one of their conversations with a social media platform cannot seek relief in federal court unless they can show that this conversation actually caused a social media company to pull down a specific piece of content posted by that plaintiff.</p>
<p id="mhIugR">And even if one of the <em>Murthy</em> plaintiffs can make such a demonstration, which is unlikely, the Supreme Court’s precedents impose an even higher barrier on plaintiffs asking a federal court to issue an injunction (a court order forbidding the defendant from acting in a particular way). The controlling case is <a href="https://supreme.justia.com/cases/federal/us/461/95/"><em>City of Los Angeles v. Lyons</em></a> (1983), which held that a plaintiff who has been injured in the past by the government cannot seek an injunction banning that activity unless the plaintiff “was likely to suffer future injury” similar to what they experienced in the past.</p>
<p id="7POKYs">The <em>Murthy</em> plaintiffs, in other words, must show 1) that a government official had an unconstitutional conversation with a social media platform, 2) that this conversation caused that platform to remove some of the plaintiffs’ content that the platform wouldn’t have removed anyway, and 3) that a similarly unconstitutional conversation is likely to happen in the future that would lead to the same outcome for the same plaintiff.</p>
<p id="mgZ1DG">That’s an extraordinary burden. And it is exceedingly unlikely that any plaintiff can meet it in this case.</p>
<p id="fW8dnU">A majority of the justices already disagreed with the Fifth Circuit once, when the Court <a href="https://www.supremecourt.gov/opinions/23pdf/23a243_7l48.pdf">temporarily blocked the Fifth Circuit’s decision last October</a>. So the Biden administration has good reason to be optimistic that <em>Murthy</em> won’t end with another sweeping court order that seems to even forbid the FBI from warning Twitter that some of its users are using the platform to coordinate criminal activity.</p>
<p id="2a5csA">Nevertheless, Democrats in government should take the same lesson from <em>Murthy</em> that they need to take from <em>NRA</em>. No one even tangentially related to the <em>Murthy</em> case did anything approaching the egregious constitutional violation that occurred in <em>NRA</em>, but that didn’t stop Republicans on the Fifth Circuit from combing through the Biden administration’s communications with social media companies, looking for a reason to issue a broad and unworkable injunction.</p>
<p id="6A5Oiy">One of the Fifth Circuit’s many errors is it seemed to assume that government actions have no value — that the public’s interest in not having terrorists find recruits on YouTube pales before some anti-vaxxer’s interest in being able to falsely tweet that Covid vaccines are unsafe. Democratic officials need to understand that judges who hold this worldview are pervasive throughout the judiciary and that they will eagerly seize upon any mistake made by a public official to sabotage whole swaths of the government.</p>
https://www.vox.com/scotus/2024/3/11/24092227/supreme-court-murthy-missouri-nra-vullo-first-amendment-social-mediaIan Millhiser2024-03-07T07:30:00-05:002024-03-07T07:30:00-05:00Researchers use online opt-in surveys all the time. Should they?
<figure>
<img alt="A man working on a laptop in a home office, sitting in front of a window in the daytime." src="https://cdn.vox-cdn.com/thumbor/m-Cz-9KVIZwGQGyQq2j7qiVWr4k=/555x0:4928x3280/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73188514/1229160021.0.jpg" />
<figcaption>Kira Hofmann/picture alliance via Getty Images</figcaption>
</figure>
<p>Survey sites recruit respondents with the promise of a reward, which may lead to bogus answers. That doesn’t mean the data is unusable. </p> <p id="bgUFdN">Search around for ways to make a little extra money online, and you might find yourself at one of many sites that offer to pay you to take surveys.</p>
<p id="UuffpK">There’s Swagbucks, SurveyJunkie, InboxDollars, and KashKick, for instance. On each of these sites, users are paid small amounts of money for completing surveys, playing games, or making purchases. </p>
<p id="4Ew7xM">The surveys on these sites are “opt-in” surveys, meaning that participants are actively choosing to take them, rather than researchers pulling a random sample of a population to poll, as professional pollsters do.</p>
<p id="yN9vK6">Unsurprisingly, opt-in surveys can lead to some skewed results: earlier this week <a href="https://www.pewresearch.org/short-reads/2024/03/05/online-opt-in-polls-can-produce-misleading-results-especially-for-young-people-and-hispanic-adults/">Pew Research Center wrote</a> about their analysis of one such opt-in survey that found 20 percent of US adults under 30 believe that “The Holocaust is a myth.” Pew’s attempt to replicate this result via a random sampling of Americans found that just 3 percent of Americans under 30 agreed with an identically worded statement about the Holocaust — a percentage that was more or less the same across <em>all</em> age groups. </p>
<p id="94XHYl">The analysis also included this incredible tidbit: </p>
<p id="RMwwe0"><em>“In a February 2022 survey experiment, we asked opt-in respondents if they were</em><a href="https://washstat.org/hansen/2022Kennedy.pdf"><em> licensed to operate a class SSGN (nuclear) submarine</em></a><em>. In the opt-in survey, 12% of adults under 30 claimed this qualification, significantly higher than the share among older respondents. In reality, the share of Americans with this type of submarine license rounds to 0%.” </em></p>
<p id="Dv85Lw">Oof, right? </p>
<p id="qvzrTn">The <a href="https://www.vox.com/google" data-source="encore">Google</a> results for survey sites are filled with reviews from people who are mainly concerned with whether these sites are “legitimate” or scams. But the Pew analysis points to another question: just how good is the data collected for a survey when its participants are incentivized to speed through as many as possible in order to earn cash? </p>
<h3 id="V7gdfX">The problems with opt-in surveys, explained</h3>
<p id="9L2F6v">I dug around and, surprise! It’s complicated. </p>
<p id="nRNNTr">“Errors are introduced (and remediated) in the survey process at every step,” noted David Rothschild, an economist at Microsoft Research. The fact that a survey was conducted online for a small reward isn’t necessarily enough information to analyze data quality in a meaningful way. </p>
<p id="74eQZ7">As Pew noted in its analysis, the Holocaust denial survey used an agree/disagree format that can lead to “<a href="https://www.qualtrics.com/blog/avoiding-the-yes-bias/">acquiescence bias</a>” — a tendency for respondents to give an affirmative reply. This means that while the survey collection method might have been part of the problem, the question itself may have also led to inaccurate results. </p>
<p id="OdNPT6">“There are many types of opt-in online audiences; some have strong vetting to ensure the respondents are who they say they are and produce high quality responses, while others just accept whomever without any pre-response quality control,” Rothschild added. </p>
<p id="ZorGSu">Here’s what you need to know. </p>
<h3 id="r32Epx">How do online survey sites work? </h3>
<p id="5ztSay">Although there are a couple different models, the online survey sites we are talking about offer small rewards in exchange for survey participation. Most say they try to “match” users to relevant surveys based on the data they collect about their users, and generally speaking, you only get paid if you qualify to take the survey and complete each required question. </p>
<p id="QCbwd4">Typically, these sites pay users in points, which translate to small dollar amounts per survey, if they pass a set of screening questions and complete the entire survey. These points often do not translate to very much money: I created an account on Swagbucks and checked a list of available surveys. They included a 20-minute survey for 119 “Swagbucks,” which translates to … $1.19. </p>
<p id="13cgMY">Longer surveys may offer more, while some surveys with a 10-minute time estimate offer less than a dollar. These are similar to the rates I saw on SurveyJunkie. On <a href="https://www.vox.com/amazon" data-source="encore">Amazon</a>’s Mechanical Turk, a marketplace for work that includes survey taking, <a href="https://www.nytimes.com/interactive/2019/11/15/nyregion/amazon-mechanical-turk.html">a survey might pay less than 10 cents</a>. </p>
<h3 id="9qdNKR">Why would pollsters and researchers use sites like these to collect responses? </h3>
<p id="j9R9Fk">In some applications like election polls, as Pew noted, opt-in surveys can perform similarly to random probability-based surveys. Which is great, because they are generally much cheaper to conduct. </p>
<p id="4c87Ah">“Lower cost survey pools are great for exploration” and when you don’t need a very precise outcome, said Rothschild. The results are generally faster, cheaper, and more convenient. </p>
<p id="wJEUCW">“Especially for research that’s being done on a close-to-shoestring budget, opt-in online surveys are a natural choice for scholars trying to study diverse aspects of social behavior,” added Thomas Gift, an associate professor of political science at University College London. </p>
<p id="S3tNox">Gift and another researcher studied the potential of fraudulent responses in online opt-in studies after using an opt-in study themselves to study a separate question. “It was only during the fielding of the experiment that large cohorts of respondents seemed to be giving suspicious answers about their backgrounds,” he said. “So we investigated further.” </p>
<h3 id="Sps2OG">Why, and when, are online surveys prone to bogus respondents? </h3>
<p id="jzPxDd">Researchers can use a lot of tools, including screening questions, to weed out bad responses and end up with a set of usable data. But there are some instances, such as obscure beliefs or surveys where you need really precise data, where opt-in online surveys are going to be a problem. </p>
<p id="rnYynQ">Pew noted a few considerations here: based on their research over the years, online opt-in polls have a tendency to overestimate fringe beliefs (they gave the example of belief in conspiracy theories). That overrepresentation is more severe among younger respondents, and among Hispanic adults, they noted. </p>
<p id="g5lOSb">Gift and his research partner hired a “nationally-recognized” marketing firm — which they left unnamed in their paper for liability reasons — to conduct a survey for them that collected respondents with experience in the Army. This firm, they said, distributed the survey to a number of sub-vendors that provided financial incentives for responses (these sub-vendors were also left anonymous). </p>
<p id="uQri3I">In order to detect whether respondents really did have experience in the Army or not, Gift used screening questions embedded in the survey. Respondents were asked about saluting protocol, and for specific information on their military background.</p>
<p id="VXR2Xo"><a href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3957523">Based on their analysis of those screeners</a>, nearly 82 percent of respondents may have pretended to be associated with the Army in order to take the survey and get paid for it. About 36 percent of those respondents passed the knowledge screening test, but were identified as probably misrepresenting themselves based on their answers to the survey questions themselves. </p>
<p id="5LcBRx">And there was also evidence in the survey results that some respondents were taking the survey a bunch of times, giving nearly identical answers and tweaking their demographic data enough to pass as different people, presumably to get paid multiple times for the same survey. </p>
<h3 id="lR3mLa">How can researchers minimize bogus responses and end up with useful data from an online survey? </h3>
<p id="iRK3VN">Essentially, by testing the respondent. Online surveys use attention checks, IP tracking, anti-bot software, and monitoring the time it takes for someone to complete a survey in order to try to mitigate fraud. Asking respondents questions like the one Pew flagged about having a license to drive a submarine is a pretty good way to tell whether someone is just cruising through and answering questions as quickly as possible, or if they’re actually reading the questions. </p>
<p id="Nq3HGe">Nothing is going to catch every single bogus response, and, as Rothschild noted, some low-quality responses will slip through attention checks. </p>
<p id="PbmY18">There are also other models for collecting data online, Gift noted. Opt-in volunteer surveys “aren’t without their limitations,” but they create a different set of incentives for participants that don’t rely on a financial reward. Gift highlighted the work of the <a href="https://dlabss.harvard.edu/">Harvard Digital Lab for Social Sciences</a>, an online platform that allows people to volunteer to participate in social science research. </p>
<p id="ZJ5cyt">While researchers might not be able to catch every single bad response, they can be transparent about how they collected their data, Rothschild noted. And it’s worth looking for that information the next time you see a shocking headline about a shocking belief held by The Youth. </p>
https://www.vox.com/technology/2024/3/7/24092604/online-opt-in-survey-quality-explained-swagbucks-surveyjunkieA.W. Ohlheiser2024-03-07T07:00:00-05:002024-03-07T07:00:00-05:00How Nvidia beat everyone else in the AI race
<figure>
<img alt="A close-up of an Nvidia-made computer chip with its name and logo printed on it." src="https://cdn.vox-cdn.com/thumbor/idto3H4hU1r9hS-Dg-Awi5h3bNc=/198x0:3347x2362/1310x983/cdn.vox-cdn.com/uploads/chorus_image/image/73188457/GettyImages_1251437220.0.jpg" />
<figcaption>Nvidia, founded in California in 1993, originally made chips mainly used for gaming. | Jakub Porzycki/NurPhoto via Getty Images</figcaption>
</figure>
<p>The company you might not have heard of is now worth $2 trillion — more than Google or Amazon.</p> <p id="69v4Cs">Only four companies in the world are worth over $2 trillion. <a href="https://www.vox.com/apple" data-source="encore">Apple</a>, <a href="https://www.vox.com/microsoft" data-source="encore">Microsoft</a>, the oil company Saudi Aramco — and, as of 2024, Nvidia. It’s understandable if <a href="https://www.bloomberg.com/news/newsletters/2023-08-25/how-do-you-say-nvidia-despite-its-growth-pronunciation-bedevils-people">the name doesn’t ring a bell</a>. The company doesn’t exactly make a shiny product attached to your hand all day, every day, as Apple does. Nvidia designs a chip hidden deep inside the complicated innards of a computer, a seemingly niche product more are relying on every day. </p>
<p id="TbvtWN">Rewind the clock back to 2019, and Nvidia’s market value was hovering around <a href="https://www.sec.gov/Archives/edgar/data/1045810/000104581020000010/nvda-2020x10k.htm">$100 billion</a>. Its incredible speedrun to 20 times that already enviable size was really enabled by one thing — the <a href="https://www.vox.com/2023/4/28/23702644/artificial-intelligence-machine-learning-technology" data-source="encore">AI</a> craze. Nvidia is arguably the biggest winner in the AI industry. ChatGPT-maker OpenAI, which catapulted this obsession into the mainstream, is currently worth around <a href="https://www.nytimes.com/2024/02/16/technology/openai-artificial-intelligence-deal-valuation.html">$80 billion</a>, and according to market research firm <a href="https://www.grandviewresearch.com/industry-analysis/artificial-intelligence-ai-market">Grand View Research</a>, the entire global AI market was worth a bit under $200 billion in 2023. Both are just a paltry fraction of Nvidia’s value. With all eyes on the company’s jaw-dropping evolution, the real question now is whether Nvidia can hold on to its lofty perch — but here’s how the company got to this level.</p>
<h3 id="hJ6CY4">From games to crypto mining to AI</h3>
<p id="XojKWO">In 1993, long before uncanny AI-generated art and amusing AI chatbot convos took over our social media feeds, three Silicon Valley electrical engineers launched a startup that would focus on an exciting, fast-growing segment of personal computing: <a href="https://www.nytimes.com/1993/09/16/nyregion/video-violence-it-s-hot-it-s-mortal-it-s-kombat-teen-agers-eagerly-await.html">video games</a>.</p>
<p id="4lW14T">Nvidia was founded to design a specific kind of chip called a graphics card — also commonly called a GPU (graphics processing unit) — that enables the output of fancy 3D visuals on the computer screen. The better the graphics card, the more quickly high-quality visuals can be rendered, which is important for things like playing games and video editing. In the <a href="https://d18rn0p25nwr6d.cloudfront.net/CIK-0001045810/e198721c-58f6-4772-8633-cb95ca474654.pdf">prospectus</a> filed ahead of its initial public offering in 1999, Nvidia noted that its future success would depend on the continued growth of computer applications relying on 3D graphics. For most of Nvidia’s existence, game graphics were Nvidia’s raison d’etre.</p>
<p id="PL5ODt">Ben Bajarin, CEO and principal analyst at the tech industry research firm Creative Strategies, acknowledged that Nvidia had been “relatively isolated to a niche part of computing in the market” until recently.</p>
<p id="QuUoYV">Nvidia became a powerhouse selling cards for video games — now an entertainment industry juggernaut making <a href="https://www.latimes.com/entertainment-arts/business/story/2023-08-21/global-gaming-industry-report-2023-league-of-legends-activision-electronic-arts-nintendo-mobile-gaming-mario-bros">over $180 billion in revenue</a> last year — but it realized it would be smart to branch out from just making graphics cards for games. Not all its experiments panned out. Over a decade ago, Nvidia made a failed gambit to become a major player in the <a href="https://www.reuters.com/article/idUSN30427294/">mobile chip market</a>, but today <a href="https://www.counterpointresearch.com/insights/global-smartphone-ap-market-share/">Android phones use a range of non-Nvidia chips</a>, while iPhones use Apple-designed ones.</p>
<p id="5hpBUk">Another play, though, not only paid off, it became the reason we’re all talking about Nvidia today. In 2006, the company released a programming language called CUDA that, in short, unleashed the power of its graphics cards for more general computing processes. Its chips could now do a lot of heavy lifting for tasks unrelated to pumping out pretty game graphics, and it turned out that graphics cards could multitask even better than the CPU (central processing unit), what’s often called the central “brain” of a computer. This made Nvidia’s GPUs great for calculation-heavy tasks like machine learning (and, crypto mining). 2006 was the same year <a href="https://www.vox.com/amazon" data-source="encore">Amazon</a> launched its cloud computing business; Nvidia’s push into general computing was coming at a time when massive data centers were popping up around the world. </p>
<div class="c-float-right"><aside id="0vucfE"><q>Nvidia has joined the ranks of tech titans designated the “Magnificent Seven”</q></aside></div>
<p id="oHvnLq">That Nvidia is a powerhouse today is especially notable because for most of Silicon Valley’s history, there already was a chip-making goliath: Intel. Intel makes both CPUs and GPUs, as well as other products, and it manufactures its own semiconductors — but after a <a href="https://www.vox.com/2016/4/20/11463818/intel-iphone-mobile-revolution">series of missteps</a>, including not investing into the development of AI chips soon enough, the rival chipmaker’s preeminence has somewhat faded. In 2019, when Nvidia’s market value was just over the $100 billion mark, <a href="https://www.intc.com/filings-reports/annual-reports/content/0000050863-20-000011/0000050863-20-000011.pdf">Intel’s value</a> was double that; now Nvidia has joined the ranks of tech titans designated <a href="https://www.cnbc.com/2024/02/19/magnificent-7-profits-now-exceed-almost-every-country-in-the-world-should-we-be-worried.html">the “Magnificent Seven”</a>, a cabal of tech <a href="https://www.vox.com/stock-market" data-source="encore">stocks</a> with a combined value that exceeds the entire stock market of many rich G20 countries.</p>
<p id="u9Wp8x">“Their competitors were asleep at the wheel,” says Gil Luria, a senior analyst at the financial firm D.A. Davidson Companies. “Nvidia has long talked about the fact that GPUs are a superior technology for handling accelerated computing.”</p>
<p id="OZzGSL">Today, Nvidia’s four main markets are gaming, professional visualization (like 3D design), data centers, and the automotive industry, as it provides chips that train self-driving technology. A few years ago, its gaming market was still the biggest chunk of revenue at about <a href="https://www.sec.gov/Archives/edgar/data/1045810/000104581020000010/nvda-2020x10k.htm">$5.5 billion</a>, compared to its data center segment, which raked in about $2.9 billion. Then the pandemic broke out. People were spending a lot more time at home, and demand for computer parts, including GPUs, shot up — gaming revenue for the company in fiscal year 2021 jumped a whopping 41 percent. But there were already signs of the coming AI wave, too, as Nvidia’s data center revenue soared by an <a href="https://www.sec.gov/Archives/edgar/data/1045810/000104581021000010/nvda-20210131.htm">even more impressive 124 percent</a>. In 2023, its revenue was <a href="https://www.cnbc.com/2024/02/21/nvidias-data-center-business-is-booming-up-over-400percent-since-last-year-to-18point4-billion-in-fourth-quarter-sales.html">400 percent</a> higher than the year before. In a clear display of how quickly the AI race ramped up, data centers have overtaken games, even in a gaming boom. </p>
<p id="thZ1GJ">When it went public in 1999, Nvidia had <a href="https://www.wsj.com/articles/SB917224371262043000">250 employees</a>. Now it has over <a href="https://media.iprsoftware.com/219/files/202311/corporate-nvidia-in-brief-pdf-december-3056300-r2-2.pdf?Signature=j%2FonafsYJugHfH6cXe%2FdwO6g%2Bns%3D&Expires=1709842242&AWSAccessKeyId=AKIAJX7XEOOELCYGIVDQ&versionId=Kg1wCZP.LuqFtUOJi3dAakvjMU.DGF27&response-content-disposition=attachment">27,000</a>. Jensen Huang, Nvidia’s CEO and one of its founders, has a personal net worth that currently hovers around $70 billion, an over 1,700 percent increase <a href="https://www.forbes.com/profile/jensen-huang-1/?sh=5edc5583a6c4">since 2019</a>.</p>
<p id="iG3rOo">It’s likely you’ve already brushed up against Nvidia’s products, even if you don’t know it. Older gaming consoles like the <a href="https://www.cnet.com/tech/tech-industry/nvidia-to-work-on-playstation-3-chip/">PlayStation 3</a> and the <a href="https://www.techpowerup.com/gpu-specs/xbox-gpu.c1866">original Xbox</a> had Nvidia chips, and the current Nintendo Switch uses an Nvidia mobile chip. Many mid- to high-range laptops come packed up with an Nvidia graphics card as well. </p>
<p id="8KgNkY">But with the AI bull rush, the company promises to become more central to the tech people use every day. <a href="https://finance.yahoo.com/news/tesla-ceo-elon-musk-were-using-a-lot-of-nvidia-hardware-174620935.html">Tesla cars’ self-driving feature</a> utilizes Nvidia chips, as do practically all major tech companies’ cloud computing services. These services serve as a backbone for so much of our daily internet routines, whether it’s streaming content on <a href="https://www.vox.com/netflix" data-source="encore">Netflix</a> or using office and productivity apps. To train ChatGPT, OpenAI harnessed tens of thousands of <a href="https://www.bloomberg.com/news/articles/2023-03-13/microsoft-built-an-expensive-supercomputer-to-power-openai-s-chatgpt">Nvidia’s AI chips</a> together. People underestimate <a href="https://www.youtube.com/watch?v=YsZ-lx_3eoM">how much they use AI</a> on a daily basis, because we don’t realize that some of the automated tasks we rely on have been boosted by AI. Popular apps and social media platforms are adding new AI features seemingly every day: TikTok, <a href="https://www.vox.com/instagram-news" data-source="encore">Instagram</a>, X (formerly <a href="https://www.vox.com/twitter" data-source="encore">Twitter</a>), even <a href="https://www.vox.com/pinterest" data-source="encore">Pinterest</a> all boast some kind of AI functionality to toy with. <a href="https://www.vox.com/slack" data-source="encore">Slack</a>, a messaging platform that many workplaces use, recently <a href="https://slack.com/blog/transformation/how-we-designed-slack-ai">rolled out the ability to use AI</a> to generate thread summaries and recaps of Slack channels.</p>
<h3 id="gQOgG9">Nvidia’s chips continue to sell out — for now</h3>
<p id="UwIaQl">For Nvidia’s customers, the problem with sizzling demand is that the company can charge eye-wateringly high prices. The chips used for AI data centers cost tens of thousands of dollars, with the top-of-the-line product sometimes <a href="https://www.cnbc.com/2023/04/14/nvidias-h100-ai-chips-selling-for-more-than-40000-on-ebay.html">selling for over $40,000</a> on sites like Amazon and eBay. Last year, some clients clamoring for Nvidia’s AI chips were <a href="https://www.trendforce.com/news/2024/02/28/news-nvidias-h100-ai-chip-no-longer-out-of-reach-inventory-pressure-reportedly-forces-customers-to-resell/">waiting as much as 11 months</a>.</p>
<p id="zYTeOq">Just think of Nvidia as the <a href="https://www.cnn.com/style/hermes-birkin-bag-origins-cost/index.html">Birkin bag</a> of AI chips. A comparable offering from another chipmaker, AMD, is reportedly being sold to customers like Microsoft for <a href="https://seekingalpha.com/news/4061277-amds-35b-ai-guidance-is-sandbagged-citi-say">about $10,000 to $15,000</a>, just a fraction of what <a href="https://www.reuters.com/technology/nvidia-chases-30-billion-custom-chip-market-with-new-unit-sources-2024-02-09/">Nvidia charges</a>. It’s not just the AI chips, either. Nvidia’s gaming business continues to boom, and the price gap between its high-end gaming card and a similarly performing one from AMD has been <a href="https://www.theverge.com/2022/11/16/23462949/nvidia-amd-rtx-4080-rdna-3-7900-xt-price-size">growing wider</a>. In its last financial quarter, Nvidia reported a gross margin of <a href="https://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-fourth-quarter-and-fiscal-2024">76 percent</a>. As in, it cost them just 24 cents to make a dollar in sales. AMD’s most recent gross margin was only <a href="https://ir.amd.com/news-events/press-releases/detail/1180/amd-reports-fourth-quarter-and-full-year-2023-financial">47 percent</a>.</p>
<p id="mGOOdp">Nvidia’s fans argue that its yawning lead was earned by making an early bet that AI would take over the world — its chips are worth the price because of its superior software, and because so much of AI infrastructure has already been built around Nvidia’s products. But Erik Peinert, a research manager and editor at the American Economic Liberties Project who helped put together a <a href="https://www.economicliberties.us/our-work/reshoring-and-restoring-chips-implementation-for-a-competitive-semiconductor-industry/">recent report</a> on competition within the chip industry, notes that Nvidia has gotten a price boost because TSMC, the biggest semiconductor maker in the world, has <a href="https://www.reuters.com/technology/tsmc-q4-profit-falls-19-beats-market-expectations-2024-01-18/">struggled for years to keep up</a> with demand. A recent <a href="https://www.wsj.com/tech/ai/nvidia-ceo-jensen-huang-vision-company-f05db212">Wall Street Journal</a> report also suggested that the company may be throwing its weight around to maintain dominance; the CEO of an AI chip startup called Groq claimed that customers were scared Nvidia would punish them with order delays if it got wind they were meeting with other chip makers. </p>
<div class="c-float-right"><aside id="V9cxx2"><q>“The biggest challenge for Nvidia is that their customers want to compete with them”</q></aside></div>
<p id="SMweuQ">It’s undeniable that Nvidia put in the investment into courting the AI industry well before others started paying attention, but its grip on the market isn’t unshakable. An army of competitors are on the march, ranging from <a href="https://www.nytimes.com/2024/01/29/technology/ai-chips-nvidia-amazon-google-microsoft-meta.html">smaller startups to deep-pocketed opponents</a>, including Amazon, <a href="https://www.vox.com/meta" data-source="encore">Meta</a>, Microsoft, and <a href="https://www.vox.com/google" data-source="encore">Google</a>, all of which currently use Nvidia chips. “The biggest challenge for Nvidia is that their customers want to compete with them,” says Luria. </p>
<p id="bBU1HQ">It’s not just that their customers want to make some of the money that Nvidia has been raking in — it’s that they can’t afford to keep paying so much. Microsoft “went from spending less than 10 percent of their capital expenditure on Nvidia to spending nearly 40 percent,” Luria says. “That’s not sustainable.”</p>
<p id="B4iOOx">The fact that <a href="https://fortune.com/2024/02/22/what-does-nvidia-do-chips-ai-jensen-huang/">over 70 percent of AI chips</a> are bought from Nvidia is also cause for concern for <a href="https://www.vox.com/antitrust" data-source="encore">antitrust</a> regulators around the world — the <a href="https://www.vox.com/european-union" data-source="encore">EU</a> recently started <a href="https://www.bloomberg.com/news/articles/2023-09-29/eu-begins-early-stage-probe-into-ai-chip-market-abuses-that-nvidia-dominates">looking into the industry</a> for potential antitrust abuses. When Nvidia <a href="https://nvidianews.nvidia.com/news/nvidia-to-acquire-arm-for-40-billion-creating-worlds-premier-computing-company-for-the-age-of-ai">announced in late 2020</a> that it wanted to spend an eye-popping $40 billion to buy Arm Limited, a company that designs a chip architecture that most modern smartphones and newer Apple computers use<a href="https://nvidianews.nvidia.com/news/nvidia-to-acquire-arm-for-40-billion-creating-worlds-premier-computing-company-for-the-age-of-ai">,</a> <a href="https://www.ftc.gov/legal-library/browse/cases-proceedings/2110015-nvidiaarm-matter">the FTC blocked the deal</a>. “That acquisition was pretty clearly intended to get control over a software architecture that most of the industry relied on,” says Peinert. “The fact that they have so much pricing power, and that they’re not facing any effective competition, is a real concern.”</p>
<h3 id="kdWWJf">Will the AI love affair cool off?</h3>
<p id="asFFEh">Whether Nvidia will sustain itself as a $2 trillion company — or rise to even greater heights — depends, fundamentally, on whether both consumer and investor attention on AI can be sustained. Silicon Valley is awash with newly founded AI companies, but what <a href="https://www.wsj.com/tech/ai/ais-costly-buildup-could-make-early-products-a-hard-sell-bdd29b9f">percentage of them will take off</a>, and how long will funders keep pouring money into them?</p>
<p id="HcM7WR">Widespread AI awareness came about because ChatGPT was an easy-to-use — or at least easy-to-show-off-on-social-media — novelty for the general public to get excited about. But a lot of AI work is still focusing on AI training rather than what’s called AI inferencing, which involves using trained AI models to solve a task, like the way that ChatGPT answers a user’s query or facial recognition tech identifies people. Though the AI inference market is growing (and maybe <a href="https://www.wsj.com/tech/ai/how-a-shifting-ai-chip-market-will-shape-nvidias-future-f0c256b1">growing faster than expected</a>), much of the sector is still going to be spending a lot more time — and money — on training. For training, Nvidia’s first-class chips will likely remain the most coveted, at least for a while. But once AI inferencing explodes, there will be less of a need for such high-performance chips, and Nvidia’s primacy could slip.</p>
<p id="DPmGs1">Some financial analysts and industry experts have expressed wariness over <a href="https://apolloacademy.com/the-current-ai-bubble-is-bigger-than-the-1990s-tech-bubble/">Nvidia’s stratospheric valuation</a>, suspecting that AI enthusiasm will slow down and that there may already be <a href="https://www.ft.com/content/dc47c5f3-9bd4-4da0-a5cb-c795efd14c9c">too much money going toward</a> making AI chips. Traffic to <a href="https://www.reuters.com/technology/chatgpt-traffic-slips-again-third-month-row-2023-09-07/">ChatGPT has dropped off</a> since last May and some investors are <a href="https://www.theinformation.com/articles/some-ai-startups-find-the-moneys-no-longer-so-easy">slowing down the money hose</a>.</p>
<p id="Ar2FGy">“Every big technology goes through an adoption cycle,” says Luria. “As it comes into consciousness, you build this huge hype. Then at some point, the hype gets too big, and then you get past it and get into the trough of disillusionment.” He expects to see that soon with AI — though that doesn’t mean it’s a bubble. </p>
<p id="7LAFXe">Nvidia’s revenue last year was about <a href="https://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-fourth-quarter-and-fiscal-2024">$60 billion</a>, which was a 126 percent increase from the prior year. Its high valuation and stock price is based not just on that revenue, though, but for its predicted continued growth — for comparison, Amazon currently has a lower market value than Nvidia yet made almost <a href="https://ir.aboutamazon.com/news-release/news-release-details/2024/Amazon.com-Announces-Fourth-Quarter-Results/">$575 billion in sales</a> last year. The path to Nvidia booking large enough profits to justify the <a href="https://fortune.com/2024/02/22/nvidia-stock-price-earnings-investors-outlook/">$2 trillion valuation looks steep</a> to some experts, especially knowing that the competition is kicking into high gear.</p>
<p id="E9k1es">There’s also the possibility that Nvidia could be stymied by how fast microchip technology can advance. It has moved at a blistering pace in the last several decades, but there are signs that the pace at which more transistors can be fitted onto a microchip — making them smaller and more powerful — is <a href="https://www.wsj.com/articles/tech-progress-is-slowing-down-b7fcaee0">slowing down</a>. Whether Nvidia can keep offering meaningful hardware and software improvements that convince its customers to buy its latest AI chips could be a challenge, says Bajarin.</p>
<p id="OB4I2I">Yet, for all these possible obstacles, if one were to bet whether Nvidia will soon become as familiar a tech company as Apple and Google, the safe answer is yes. AI fever is why Nvidia is in the rarefied club of trillion-dollar companies — but it may be just as true to say that AI is so big because of Nvidia.</p>
<p id="F7Cwwa"></p>
https://www.vox.com/money/2024/3/7/24092309/nvidia-stock-earnings-valuation-ai-explainerWhizy Kim