AI is allowing for entirely new ways to increase civic involvement in policymaking which is reshaping the legislative process and reducing the impact of lobbying. Lawmakers in the US and Brazil have been using AI to draft parts of or entire pieces of legislation — with varying degrees of acceptance (see here).
In the US this year, someone running for mayor proposed to replace the town's human lawmakers and use AI for all decision making; in the UK, someone tried the same for national government. In both cases, they said they would appear in human form only where essential. The electoral rules they were under don’t allow for AI to stand for election. Other jurisdictions may one day.
Residents of Cheyenne, capital of the US state of Wyoming, had the option in their August primaries of voting for Vic, the Virtual Integrated Citizen chatbot, based on ChatGPT, as mayor. The bot was trained and managed by librarian Victor Miller, whose name had to be on the ballot as a registered voter and who would be the AI’s human interface (or “meat avatar” in Miller’s words) where needed.
Miller promised an AI tool that would make 100 percent of decisions as mayor, itself a tricky concept for generative AI, which is probabilistic rather than deterministic. But he managed to yield yes/no answers on simulated votes.
Miller is upfront about the practicalities of the daily work of lawmakers. He believes it simply isn’t feasible for them to process the amount of documentation they’re expected to absorb each day. Vic, meanwhile, can process and understand the texts.
“Lawmakers already do not understand legislation,” Miller told MLex. “It is understood in America that Congress does not read the bills. They receive 8,000 pages in the morning and vote on it in the afternoon.”
Miller said that legislators are already using techniques analogous to AI by “using an army of 40,000 congressional staffers, who are the sharpest young people this country has, and this is how they have been dealing with the complexity. We now have an even better way, and it is available to all of us now.”
Vic ended up getting just 327 votes — though this was still more than two other human candidates.
Miller now plans to stand for city councilman in Cheyenne in 2026. He’s set up the Rational Governance Alliance to develop open-source tools for governance and decision-making to train many more potential candidates to act as interfaces for AI lawmakers around the country.
He’s signed an “income share agreement” and would be the first “rationally bound delegate,” whereby he promises to use AI for all decision making and give 10 percent of his gross salary from an elected position to his alliance to further fund it.
— AI Steve —
In the UK, a new political party attempted to get AI into power at the national level.
For the snap parliamentary election in July, the startup SmarterUK party put forward AI Steve, a chatbot and opinion-gathering service that would be represented where necessary — in Parliament, for example — by the human Steve Endacott, whose day job is running a business creating generative AI voice assistants for companies.
He described wanting to use AI to increase public engagement. The party hoped to use AI to research and draft policies, but the “key aspect is that we are liaising with more human voters to create policies so the AI is just an enabler,” Endacott told MLex.
Constituents would be able to chat with the AI tool and suggest policies. The ideas would be collated and passed to a panel of human policy makers, and then another human panel would validate the ideas.
Endacott only got 179 votes, and pinpointed low awareness as one of the main issues. His polling found that only 9 percent of his constituency, in the southern English city of Brighton, had heard of him, despite the AI bid making headlines around the world.
“The key weakness in most democracies is that voters are disengaged and don’t think they can influence the policy makers,” Endacott said, referring to 2023 polling by the Office for National Statistics that showed that almost two thirds of people feel they have little or no confidence that they have a say in what the government does. “In our opinion, [this] is the biggest issue that needs dealing with.”
— AI government —
Taiwan has an even more enveloping AI-enabled solution: a shadow government helping deliver new legislation.
Since a bout of social unrest in 2014 and the emergence of the flourishing civic technology community known as "g0v" (pronounced "gov-zero"), the Taiwanese government has been developing a new approach to civic engagement and digital literacy which has brought a shadow government called vTaiwan, run by the country’s Public Digital Information Service.
It uses online platforms to bring people together, with groups then taken through AI-facilitated workshops on an issue, with real-time analysis of the groups’ stance. Further AI tools process consultation responses and turn them into new legislation. Communications platforms use algorithms designed to bring people together — the opposite goal to that of social media platforms.
So far, the approach has delivered new regulations including creating a regulatory sandbox system that has been used for fintech, 5G spectrum and self-driving cars, a new strategy for tackling misinformation in elections and a way to regulate Uber. vTaiwan also contributed to the country’s draft AI regulation, the AI Basic Act, published in July.
Taiwan, intensely civic-minded as a legacy of its complicated path to democracy, is upheld as arguably the leading example of how technology, including AI, can be used to improve public engagement as countries and local authorities copy elements of its approach.
Its lawmakers are not legally bound to accept the consultation process and would no doubt still be subject to the influence of lobbyists. This runs the risk of the vTaiwan process being seen as “openwashing” — only pretending to be transparent. It wouldn’t, for example, be uncompromising enough for the likes of Wyoming’s Miller.
— AI to replace or assist? —
“I believe replacing politicians with AI is critical,” Victor Miller told MLex. “My goal is to get rid of politicians completely. I think this needs to be done at speed, to avoid more wars and to bring prosperity to those who are governed.”
Endacott’s SmarterUK party, meanwhile, pitches itself as “policies for the people, by the people,” and its website calls for a different type of lawmaker rather than none at all: “No Career Politicians. SmarterUK is staffed by sensible businesspeople.”
“I don’t think that voters would be happy to leave policy generation to pure AI routes,” Endacott told MLex. Should it ever replace human lawmakers, though? “Absolutely not. We want to use AI to empower politicians and make them more efficient, but we want to keep the human element.”
For Miller, though, “any attempt to take a lukewarm position — à la AI Steve — is a trap that will leave us with simply more politicians.”
While existing lawmakers in the US are using AI in ever more aspects of their jobs, those who have spoken to MLex don’t foresee it replacing them (see here).
Arizona state representative Alexander Kolodin, who has used ChatGPT in drafting a bill on deepfakes, told MLex that “AI doesn't replace any of the humans in the process. Nor could it ever, because the legislative process is one that is fundamentally about human judgment.”
Please e-mail editors@mlex.com to contact the editorial staff regarding this story, or to submit the names of lawyers and advisers.