How AI revised this bill

There is an irony in using AI to write a climate bill. The data centers that run these models draw enormous amounts of electricity, and in the United States more of that new demand is being met by burning gas than by any other source. By Microsoft's own accounting, its emissions in fiscal 2024 were 23.4 percent above its 2020 baseline, and they rose another 25 percent in fiscal 2025, driven by the expansion of its data centers.¹ As Kate Aronoff of The New Republic puts it, whatever AI developers claim their technology will eventually do for the climate, "for now—and into the foreseeable future—they are using a lot of gas." AI is part of the problem this bill is written to solve.

AI's boosters promise something bigger: that it will become a genie, close to a god, that makes our hardest problems go away, climate among them. The genie is their own image. Sam Altman, OpenAI's chief executive, describes the marketing challenge for OpenAI and the industry this way: "we need to explain to the world. Number one, we are close to creating a genie that can grant any wish."² He lists "fixing the climate" among the triumphs he expects, and Google DeepMind has promised that AGI will "supercharge" our ability to tackle climate change.³ Bill Gates told reporters that data centers add only 2 to 2.5 percent to energy demand, and that AI will "certainly" cut emissions by more than that.⁴ I don't believe it, and neither does the linguist Emily M. Bender, who calls it "utter magical thinking": "We know what we need to do about the climate crisis." It is, she says, "a political problem and a social problem." That promise is likely empty, and it is dangerous even when it is sincere, because it gives everyone a reason to wait. Eric Schmidt, Google's former chief executive, said it out loud: "we're not going to hit the climate goals anyway because we're not organized to do it," and "I'd rather bet on AI solving the problem than constraining it and having the problem."⁵ As Brian Merchant puts it, it is the kind of promise that "saps the will to act." Every year spent waiting for a machine to save us is another year of extraction.

AI cannot solve climate change. The solutions already exist: the IPCC found that low-cost options are "already available," led by cheap solar, wind and batteries.⁶ What we lack are laws that take the profit out of extraction. What stands in the way is power, and only people can change that: voters, organizers, workers, and the representatives they elect. Writing about AI, the climate scientist Katharine Hayhoe puts it this way: "Tech companies won't regulate themselves. That role belongs to governments, and governments respond to citizens."⁷ Michael Mann, another climate scientist, says it plainly: "The obstacle isn't technology." Naomi Klein called waiting for machines to hand us a more palatable answer "not a cure for this crisis" but "one more symptom of it." Democratic power is what changes the rules, and changing the rules is what this crisis requires.

So this is how I used AI, and I think it is the only way we can rely on it to help: as a tool inside a democratic process, never a substitute for one. Even the International Energy Agency says AI "is not a silver bullet and does not remove the need for proactive policy." A person set the goal and made every decision. Everything the AI produced, including the research, the reviews, the competing drafts and the arguments between them, is published so anyone can check it. Kate Crawford, who maps AI's costs to the planet, says the conversation we need to have about it is "fundamentally a democratic one." And what came out is not a solution. It is a bill, a proposal for the people's representatives to debate, amend, and vote on. Bruce Schneier and Nathan Sanders, writing on AI and democracy, warn that AI magnifies whatever power wields it; that is why the power here stays with people.

A private citizen working with AI tools produced, in days, a bill that has been researched, reviewed from seven directions, argued over, and rebuilt — work that would once have taken a staff and a season. That does not make the bill right. It means the slow part is no longer the drafting.

The technology to end fossil fuel extraction exists and is cheap. The legislative text can now be drafted quickly. What is missing is the political will to pass it.