The future politics of AI

I’m sipping some robot coffee next to Teri Olle, the lady who lost the battle for AI safety in California. She looks tired, eyes crinkled and dark under her gray-streaked hair. I’m tired too; we chat about our tiredness like two people who are way too old to be working in tech. Then she turns to me and says something that will haunt me for the next six months. “Josh, this bill was just the beginning. Companies care a lot about AI,” she said. “But voters don’t. Nothing we do will matter in the face of that dramatic imbalance in the politics.”

Teri is one of authors of SB 1047, a widely-debated AI safety bill in California. A few days after we first meet, she sends me a more qualified message. Many legislators cared. “They often invoked the mistake of waiting on social media, and now trying to clean up the mess years on.” But the governor was much more susceptible to the tech point of view, and the tech lobby had focused on the governor. So SB 1047 failed.

AI safety is failing as a political movement. It lost one battle, and then another, and then another, and then another. Capabilities grow, and institutions stagnate. Now power is consolidating around a handful of CEOs and great-power politicians, all locked in a race to control the future.

But politics is not one battle, and AI is not one race.

Another start

In the last season of the Public AI Seminar, we heard about AI safety from Yoshua Bengio, perhaps its leading proponent. We heard about democratic AI from Divya Siddarth, open source AI from Irene Solaiman, community-powered AI from Jeni Tennison, and AI nationalisms from Sarah Myers-West and Amba Kak. And we heard from Aza Raskin, who in a wide-ranging conversation admitted that one of the most powerful stories of social media, 2010’s The Social Network, came out too late to matter.

We debated these narratives, each an achingly incomplete picture of the world, passed like soiled scraps of samizdat between intellectuals with no instinct or even desire for power. I see these movements rising and falling—this movement too, this public AI. Now is the time to grasp power, and I’m afraid that we are letting it go as the bare hand does with wind, touching but never possessed and never possessing.

That’s why, in this new season of the seminar, I want to talk with artists and storytellers: people who can help us descend from the comforting aloofness of tech policy to bodies and breaths. Or, to riff on Pinsky: to make the medium of AI “a human body, a column of air inside the chest”.

Over the next two months, I will post here a series of reflections on the seminar and on the future politics of AI. To the new and old members of the seminar, and to everyone in this movement: I’m glad you’re here. Let’s begin.