“When we sign such contracts without binding governance, abstract excuses such as ‘I’m just doing research’ start to ring hollow. At least this is the case for me. … Regardless of whether we eventually lead the A.G.I. race, my contributions helped improve our models, and I cannot stop wondering how these models will be used in the end, and what, if anything, will constrain them.”
—Andreas Kirsch, a research scientist at Google DeepMind, expressing his concerns in a recent X post over the company’s Pentagon contract.

And now for the main event…

Dr. Abdul El-Sayed, the Bernie Sanders–endorsed progressive Michigan Senate candidate, was no latecomer to that thorny political rose that has recently blossomed around A.I. data centers. In January, he published a “terms of engagement” for data centers, harvesting all the low-hanging fruit that’s had so many people in such an uproar: no rate hikes, community transparency requirements, job guarantees, energy-reliability guarantees, water protection, etcetera.

In the intervening months since, of course, A.I. has grown into one of the more salient issues of the midterms: Both incumbents and challengers have directly addressed broader concerns around the tech with bills and legislative proposals, even as Congress remains inactive. Last week, El-Sayed inserted himself further into the A.I. political discourse by publishing his “A.I. Under Democracy” plan—a quixotic proposal that, among other things, calls for all frontier A.I. labs to become public benefit corporations; demands Big Tech to divest from their A.I. businesses; proposes 50 percent public ownership of these companies, complete with annual dividends; and requires that more than half of each company’s board consist of publicly elected officials.

The plan, which ostensibly traces its roots to a fair bit of Bernie-style economic populism, goes further still in a few of its details. To wit: “A.I. Under Democracy” argues for a tax on corporations that automate work, a breadth of safety requirements, enforceable international cooperation, and a list of so-called “no-goes.” For instance, “A.I. may not deny medical care, autonomously fire weapons, conduct warrantless surveillance, replace human oversight in life-or-death decisions, or make hiring or firing decisions.”

According to recent polling, El-Sayed has a healthy but surmountable lead over his competition in the Democratic primary, Rep. Haley Stevens. (State Sen. Mallory McMorrow, whom I spoke to a few weeks ago about her own A.I. plan, dropped out of the race on Sunday.) The primary has attracted around $30 million in outside PAC spending, though it has yet to draw the interest of the major A.I. PACs. The following conversation has, as always, been lightly edited for clarity.

Ian Krietzberg: What have you been seeing and hearing that compelled you to respond with this plan?

Abdul El-Sayed: There’s not a campaign event that I do where the question of either data centers or A.I. itself doesn’t come up. From young people, it’s the existential risk of not being able to sell your intellectual capital for money on the labor market; for others, it’s what happens when Big Tech comes and pushes a data center onto their existence. Regardless, this is the biggest question facing humanity and our country in the coming decade, and you’re seeing startlingly little from the U.S. Senate at all, and I want to be able to demonstrate what leadership looks like.

Among voters, is there an issue within A.I. that feels most salient?

Future possibility of work, and then data centers. For folks who’ve been watching this, thinking about how fast this technology has developed over the past three years, there is an existential doom question, but that one feels further off than, I don’t know if I’ll be able to have a job in the future. I also hear a lot about A.I. for very particular uses—like, I don’t want to be rejected for a job via A.I.; I don’t want A.I. to tell me what healthcare I can’t have via prior authorization; I don’t want A.I. firing weapons. That’s kind of how it shows up.

How do you think this issue stacks up to the major issues that draw voters and swing elections?

A.I. is an affordability issue, full stop. If you’re worried that a data center is coming to town and there’s no safety around your utility rates, you’re worried your utility rates are going to skyrocket. If you’re worried that there’s no safety around how they use water, you’re worried that your water rates are going to skyrocket, or whether you can get clean water at all. If A.I. is coming for 50 percent of the white-collar jobs, as Anthropic’s C.E.O. suggests it might, then that’s going to be a really big problem for you. If you’re under 40 and you’re worried that you might never be able to afford a home, and now you can’t get a job—again, it’s a massive affordability issue.

It all really comes back to who the economy is stacked up for. A.I. is the expression of economic inequality in technological terms. It’s big billionaires being able to borrow from other billionaires to buy compute from other billionaires to create a technology that is going to take the rest of our jobs to accrue more capital for billionaires.

You have certain “no-goes” listed in the plan. I’m curious how much you’ve encountered a desire among ordinary people to find a way to turn A.I. off completely, rather than do it responsibly.

I hear a lot of that, and I honestly have a whole lot of empathy for it. For most of us, all we’ve ever seen with new technology is that it gets used to monetize against us, and A.I. feels like the endgame of that. Now, if folks saw responsible A.I.—i.e., we’re using A.I. to cure cancer, or to design the next generation of medications, or to take on the climate crisis, or to address our energy issues—I think a lot of people would feel differently about it.

But the experience of having A.I. in our lives now is you hear about a data center being moved into your backyard so you can get bombarded with slop videos, and maybe you might have access to a $20-a-month ChatGPT that hallucinates anyway, and then they tell you this thing’s going to take your job. I understand why people are pissed off. I understand the technology has real incredible use cases.

I also understand the incentives that are driving the technology into its worst-possible use cases, and I want to address the incentives. While A.I. is new technology, democracy is old technology, and for a long time we’ve trusted democracy to be able to govern the approach we take to technology and the ways we think about where we allocate our scarce resources. And so I want to put A.I. under democracy, so that we get the best outcomes and not the worst.

Regarding China and the A.G.I. race—

That race we’re being told we’re running.

Yeah, that race. It’s been leveraged by some technologists and politicians as a reason for not regulating A.I.—we’ll “fall behind” if we regulate, and we can’t afford to do so.

To what end, though? We have to feed all our jobs to A.I., and our security to A.I., and risk existential doom so that we don’t lose to China, so that China doesn’t take our jobs and yield existential doom?

How many times are we going to watch our public policy get eroded, broken? How much are we going to lose at the risk of losing some existential battle to China that they’ve been selling to us in some way or another for the past 70 years? I would much rather make sure that we have good jobs; that people who are graduating can get those jobs; that we don’t inadvertently create a technology that can create existential doom for humanity, for fear of losing some theoretical war to China that China doesn’t even appear to be fighting.

Are you worried that the A.I. PACs will enter the race?

If you look at the ad spend against me, I’m already at $30 million. How many more attack ads are you going to throw at me? I think people are sick and tired of being told that they ought to believe in the 30-second attack ad on television, and I would much rather be in a situation where I’m fighting for the truth. I want people to know what I want to do when I’m a U.S. senator. If I don’t succeed in becoming a U.S. senator, that will not have been the worst outcome. The worst outcome is you succeed in [becoming] a U.S. senator, and then you fail the responsibilities over which you were supposed to actually govern, and I am not willing to do that.

People’s lives are hard, and if you’re not addressing both the challenges that make their lives hard right now and the ones that are going to make their lives harder, then you’re missing the boat. Every leader needs to be super focused on this question of A.I. because it’s barreling down on us faster than we can even know.

That’s all for today. I’ll see you on Thursday.

Ian

Need help? Review our FAQ page or contact us for assistance. For brand partnerships, email ads@puck.news.

You received this email because you signed up to receive emails from Puck, or as part of your Puck account associated with {{customer.email}}. To stop receiving this newsletter and/or manage all your email preferences, click here.

Puck is published by Heat Media LLC. 107 Greenwich St., New York, NY 10006