Tristan Harris says that the people who are most excited about making AI are also planning for the end of civilization. This is a little unsettling. Not in a figurative way. To the point. Bunkers. Keep places safe. Plans for leaving. Harris brought up this point in a recent conversation with Chris Williamson. Harris is the co-founder of the Center for Humane Technology and a former Google design ethicist who helped reveal the darker side of social media in The Social Dilemma. It was a strong one that deserves more than a quick skim.
Take some time to think about the detail. People would think you or your neighbor were crazy if you started building a bunker without anyone knowing. If the heads of the world’s biggest AI companies start doing it, that’s a sign that you should pay attention to.
Harris has spent years going into places most of us never go. In 2012 and 2013, he worked as a design ethicist at Google and saw a small group of engineers and product managers in San Francisco change the way billions of people think and behave. Not because of a big plot, but because of small choices like whether to add infinite scroll, turn on autoplay, and decide how many notifications were enough. His presentation was sent around Google with a warning that this was wrong. Inside the company, word spread quickly. Also, it was mostly ignored.
Seeing good intentions get sucked up by the pressure of competition seems to have shaped everything Harris has done since then. He thinks the same thing will happen with AI, but it will happen faster and on a bigger scale than with social media.
He doesn’t use the phrase “race to the bottom” as a dramatic flourish. It talks about something structural. If many companies are trying to make the strongest AI system as quickly as possible, no one can afford to slow down or they will lose ground to the others. This makes an industry that builds things faster than it can understand what it’s making. Harris says this means that technologies that are like gods are being made without the wisdom to go with them. The danger is in that space between capability and governance.
When you hear this, you might feel the usual mix of vague worry and helplessness that comes with big, vague threats. Harris seems to be aware of this. He compares it to The Day After, a movie about a nuclear war that 100 million Americans watched at the same time in 1982. He says that footage wasn’t what made that movie important. Everyone knew that a nuclear war would be bad. For me, the movie made me face my fears in a way that made the danger feel real and shared. He wants The AI Dilemma, his new documentary, to do the same thing—not scare people, but make them want to make a different choice.
People who talk about AI often lose sight of the difference Harris makes. It’s not that he wants no AI. He is arguing for the right kind, which is made by people who have to answer to more than just their shareholders. According to him, about eight people are on track to become trillionaires right now, and they are making choices that will affect everyone else’s future. That’s not so much a conspiracy theory as it is a simple explanation of how power is currently being held.
Harris is not the most important thing because he has the loudest voice in the room. It’s because he’s been wrong before, too right in some cases, about technologies that most people were still excited about adopting. People thought that social media sites were just places to share photos and stay in touch when he first raised concerns about how they were designed. Years later, Meta and Google were found guilty in the first major U.S. trials that blamed social media companies for hurting children. Because of what his co-founder said in court, Meta was ordered to pay $375 million in damages.
That history doesn’t mean Harris is always right. But it does mean that when someone from his position says that the men working on AI are quietly getting ready for the worst, we should ask them what they know and why the rest of us aren’t asking the same thing yet.
