We’re not going to elect AIs as political leaders anytime soon. This is not something that voters, political parties, or the legal establishment are seeking—or will accept. Such a development would require major changes to the law. But as we experience an increased level of AI automation and assistance, our human politicians will lean more on AI. This will largely happen invisibly, and we will mostly be okay with it—or, at least, unaware and apathetic. It’s just another step along a path we’ve trodden for a long time.
When a president makes a heartfelt public speech, we might resonate with it emotionally, but deep down we suspect that a speechwriter wrote it. When a legislator proposes a new bill, we might support it enthusiastically, but we assume that they didn’t write its text. And when we receive a holiday card in the mail from any of these people, it may make us smile, even if it was signed with an autopen. Those things are so much a part of politics today that we don’t even think about them.
Modern politicians are already collective and cyborg-ish. A politician today is really a complex socio-technical system with a single person as the public face. In presidential executive systems like that of the US, we like to pretend we’re just voting for one person, but we’re really voting for the whole package: a cabinet of allies, a coterie of advisors, a party’s policy platform. A good legislator manages a team of experts, takes advantage of research agencies, and learns from written summaries of events and issues. They prompt human speechwriters to write their speeches and human lawyers to draft their bills.
Gathering a good combination of people and tools is a major political skill, one that makes a huge difference in a legislator’s ability to govern. There is a singular figurehead: we vote for the individual, inspired by their ideology, charisma, or shared identity, then we hope that the system they bring along with them matches. Integrating AI tools well will become part of this skill.
In 2024, Victor Miller, a mayoral candidate in Cheyenne, Wyoming, pledged to defer all of his decisions to an AI. Standing in front of a campaign sign reading “AI FOR MAYOR,” he let an AI take questions at press conferences. That AI responded in real time in a synthesized voice. Miller came in a distant fourth in the race, with 3% of the vote. British businessman Steve Endacott tried to do much the same thing in running as an independent candidate for Parliament in 2024, campaigning on the idea that he was “controlled” by an AI avatar (which his company built) that would solicit feedback from thousands of residents on a weekly basis. He finished last, with 0.3% of the vote.
These examples are extreme, but this is an approach we expect to see others try. It is not unusual for politicians and parties to expressly rely on outside forces to guide their judgment and decision-making. Leaders worldwide tout their deference to religion, or science, or political philosophies. When sincere, that deference is a way of signaling to voters that they will adhere to a certain worldview—one shared by their voters—that provides prescriptions and justifications for their policy choices. We can think of an AI-led political party or politician as something like that: a way to predictably encode a policy platform through our vote. Even voters who have no interest in electing AIs may be attracted to AI as a tool to more reliably constrain their elected officials to adhere to their political interests.
We expect politicians to be early adopters of AI because they are constantly seeking an advantage. Their choice to use AI does not depend on whether their voters trust AI or think political applications are a good idea. For example, candidates will use chatbots to prepare for debates: a nonhuman persuader could serve as a foil to a human rebutter. It will be able to answer questions and spar with their opponents, mimicking their actual political rival’s voice, talking points, and speech patterns with a far better impersonation than a typical human stand-in. This has already happened in front of voters: when incumbent representative Don Beyer of Virginia’s eighth district declined to participate in one of several candidate forums, two independent challengers debated an AI representing him and trained on his writings.
In the future, we’ll accept that politicians rely on AI tools to draft and revise messages, to simulate and understand their electorate, and to help make campaign and policy decisions. We’ll soon regard it as normal that most of their communications are written by AI. None of this will necessarily be bad or change the nature and ideology of political leadership. But it will change the skills necessary to be a politician, devaluing leadership and, even more than today, benefiting charismatic individuals who are basically actors on the political stage.
More generally, an AI may be able to direct or even found a political party. The first example of this to emerge was the Danish Synthetic Party. In 2022, an artist collective in Denmark created an AI chatbot to interact with human members of its community, exploring political ideology through conversation and analysis of historical party platforms in the country. The party failed to receive enough signatures to earn a spot on the ballot.
Future AI-led party formation initiatives may yet succeed. The Synthetic Party demonstrated that an AI can formulate a set of policy positions targeted to build support among people of a specific demographic. An AI could also be designed to attempt to create an effective consensus platform capable of attracting broad-based support. Particularly in a European-style multiparty system, we can imagine a new party with a compelling narrative—and an AI at its core—winning attention, signatures, and votes.
Acceptance of AI use may grow only if there is adequate transparency and disclosure. If you’re a voter whose first language is different from most other people in your country, you would probably welcome the use of AI to translate candidates’ speech so you can understand their message more easily. You might already use an app on your phone to do this. But if they tried to trick you into thinking they speak your language themselves, they would likely lose your trust.
Like so many other issues, AI will attract political polarization in unexpected ways and by unpredictable events. Today, leftists decry corporate bosses using AI to crush labor, hawks champion AI development in an arms race against foreign foes, conservatives claim AI is biased against them. All this will change over time. Political actors will take up the mantle of high technology and AI for unexpected reasons, and those parties will be more likely to accept and push on the boundaries of AI-assisted politicking.
Our guiding principle here is to welcome uses of AI that distribute power and to resist those that concentrate power. If an opposition candidate running for local office gets help to bootstrap their campaign from an AI speechwriter, that’s a good thing. That candidate should still be held fully responsible for the words they speak and the ideas they represent, but they should not be discredited for using the help of machines any more than a better-funded candidate would be criticized for hiring human advisors. Candidates that propose to turn their policy platform over to AI models, however, should be criticized with all the same force that would be due to a candidate that promises to defer decisions to another private interest, like a billionaire or corporation.
Excerpted from Rewiring Democracy: How AI Will Transform Our Politics, Government, and Citizenship by Bruce Schneier and Nathan E. Sanders. Reprinted with permission from The MIT Press. Copyright 2025.
