My family is expecting our third child in just a few days.
For my part, welcoming a new baby presents an occasion to reflect on the world that I’m bringing them into. I’ve never agreed with the perspective — worryingly common among some people my age — that climate change, or war, or economic hardship, makes it wrong to bring children into the world.
For all that our precious, hard-won progress can be reversed — for all that we stand to lose — the world as a whole today is richer and safer than at almost any moment in human history. I don’t think it was a great evil to have children in 1960, when the threat of nuclear annihilation was far higher than it is now, and today’s world is a more abundant, more accepting, safer place to grow up in than that one.
And while climate change is a far greater threat now than it was then, that too shouldn’t be the deciding factor on having kids or not. While children use resources, and so in a very narrow sense can be said to contribute to resource shortages or the addition of carbon to the atmosphere, people are not liabilities bleeding out a fixed pool of resources. People drive all the inventions and discoveries which allow us to replace fossil fuels with solar panels, clean up the atmosphere, and make the world a better, more sustainable place.
The world is richer, not poorer, for having more people engaged in the process of invention and discovery. It’s a deeply myopic and self-loathing view of human nature that holds we ought to aspire to having as little an effect on the world as possible, instead of actively striving to have a positive one.
So I’m at peace with — or, more accurately, incredibly excited about — bringing children into a deeply imperfect world. But there is one thing that fills me with apprehension about the future we’re handing them. And that is artificial intelligence.
Getting AI right
I think AI has the potential to make our world vastly better: to become another driver of productivity growth, innovation, discovery, and dramatic improvements to our lives. I want to see AI fulfill that potential — and I think it might. But the course I currently see us charting isn’t one where we thoughtfully consider how to approach an extraordinarily powerful and unprecedentedly dangerous technology, weighing carefully how we can get the good without the bad.
Instead, when it comes to AI it sometimes feels like we’re living inside an absurdist farce. Some of the same AI experts who say there’s a 10 to 25 percent chance their work will lead to the end of life on Earth are nonetheless proceeding with that work at an unnerving pace. Some opponents of AI safety efforts say that humanity will be wiped out and AI will take over the galaxy — but that this is a good thing if you think about it philosophically enough. Others under the same banner claim that we can easily avoid making AI that’s too powerful — though at the same time they oppose efforts to avoid making AI that’s too powerful.
There’s an odd union of convenience between the people who think AI safety is a waste of time because powerful AI systems will never happen, and the people who think AI safety is a waste of time because powerful AI systems are going to happen very soon (but be a good thing).
It’s our call how AI shapes our future
When I was considering topics for this last newsletter before my leave, I was asked by my editor how I think AI is going to shape the world my daughter is born into. The short answer: a great deal.
If you haven’t checked out ChatGPT since the spring, you should try it again. Now it can read text, speak aloud to you, draw cartoons, and interpret visual diagrams. Many of the technology’s limitations are vanishing day by day, and it’s well worth witnessing. One thing I’m sure of is that AI will continue to improve, and companies will continue to spend billions of dollars on trying to make AI products that change how we live and work.
What happens from there is up to us. If we approach AI seriously and with maturity, understanding that a technology that could conceivably kill everybody alive is fundamentally different from any other cool new software product, then perhaps my baby — and every other baby born into 2023 along with her — will one day grow up in a world without poverty and without food insecurity, where AI assistants enable and strengthen our individual freedoms by making the world possible for everyone to navigate, whatever their background.
If we approach AI either as triumphalists who believe nothing can go wrong, or with the conviction that nothing can go right, then the future could be very different. It could be very easy for her to instead grow up in a world where AI has led to dramatic centralization of power, to a loss of human control over our economy and political system, a world where humans have no input into important decisions that are being made at a speed we can’t comprehend.
Being on the weak side of a power imbalance like that inevitably ends badly, and if you want to quibble about whether the AIs will keep us in zoos or make us falsely believe we’re still in power or flood the atmosphere with chemicals we can’t breathe — real topics of debate within the AI community — you’re merely debating among dystopias.
I want a human future that remains in human control, where people make meaningful decisions that matter in shaping their world, and where we use technology to make us stronger, not to replace us. I want my daughter’s life to be easy, but also meaningful: abundant in intangibles like freedom and opportunity as well as tangibles like cheap food and water and power.
The good news is, I think everyone ultimately wants that. We just have to hope we have time to resolve our profound disagreements about how best to pursue it.