# AI Steve and candidacies based on artificial intelligence

Author: Fernando Domínguez Sardou
Original publication: 2024-11-06
Spanish original: https://estrategiabyaleph.substack.com/p/estrategia-58-cuando-la-ia-aspira
English URL: https://elcontemplador.github.io/estrategia-english/essays/058/
Status: Published translation

This is a translation of the original Spanish essay published on 6 November 2024. Its claims, examples and forecasts retain that historical context.

English publication: 2026-09-29

> This week, estrategIA is pleased to welcome [Fernando Domínguez Sardou](https://www.linkedin.com/in/fernando-dominguez-sardou/), a political scientist, lecturer, political analyst and student on the [Master's in Marketing, Communication and Political Consulting at Institución Educativa ALEPH](https://institucioneducativaaleph.com/master-en-marketing-comunicacion-y-consultoria-politica/). In this issue's main article, he explores a subject as fascinating as it is unusual: AI Steve's candidacy.



*Historical context: this article was published on 6 November 2024. Its account of the campaign and the legal framework retains that perspective. Questions attributed within the narrative are translated from the Spanish source; their original English wording has not been verified.*



**The British parliamentary election held last July brought something new: AI Steve.** AI Steve was an independent candidate for the House of Commons in Brighton Pavilion, a constituency in south-east England. What made this candidacy unusual? **The contender was a candidate based on generative artificial intelligence, whose aim was to rethink how democracy works.** Let us take this opportunity to join the discussion and consider, in these pages, how artificial intelligence can help us rethink democracy.



The law in most countries requires candidates to be natural persons — flesh-and-blood human beings. There are notable exceptions, which have given rise to what are known as “non-human candidacies”. These generally embody a vote of protest against the workings of political and representative systems, allowing an ordinarily anti-establishment position to be expressed, but with a different profile from the actors we have seen emerge in politics over the past 20 years. “Non-human candidacies” typically do not seek to take up representative office. Instead, they seek to criticise the foundations of the system as a whole, sometimes even aiming to delegitimise it. This was not the case with AI Steve.



British law follows the usual approach: standing as a candidate requires a natural person. This is where AI Steve becomes inseparable from the flesh-and-blood person behind the idea: Steve Endacott. Endacott is a businessman who built his career in tourism and, in recent years, moved into supporting and developing various technology start-ups. Among them is Neural Voice, a company that creates voice assistants using generative AI to support businesses such as travel agencies, car dealerships and medical practices. From there, Endacott began the AI Steve venture with a question: how can we improve democracy? That question belongs not only to this non-human candidacy; it is also the guiding thread of this article.



Steve Endacott registered as a parliamentary candidate for Brighton Pavilion, as already mentioned. But his role would be limited to carrying out the decisions of, and representing, AI Steve: an AI-generated candidate that would maintain a permanent conversation with its electorate and the people it represented through its platform. In this two-way exchange, AI Steve's proposals and policies would be based on the conversations the AI held with its citizens. Endacott's parliamentary decisions would then be guided by AI Steve's instructions.



The “mistake”, or the aspect of the experiment that lacked authenticity, was precisely that AI Steve's initial positions had been developed from Endacott's and his team's own ideas, as they acknowledged on their website. Beyond that starting point, however, the advantages become apparent: a representative available 24 hours a day, seven days a week, 365 days a year; citizens' ability to create policies and, through the AI itself, ensure that these actually reach political leaders — Endacott, in this case; scrutiny of the representative's actions through continual validation with citizens; and the simplification of complex aspects of political life. Ultimately, AI Steve and its objectives were defeated. The candidacy had gained little recognition among an electorate that was also not participating in the platform, depriving it of the input on which the representative relationship was to be built. In that setting, as is characteristic of single-member electoral systems, the contest came down to two candidates, and Endacott eventually gave his explicit support to one of the more competitive alternatives.



For now, an AI candidacy cannot break through the party system or the electoral system. This is especially true when it requires a flesh-and-blood person capable of making mistakes — as Endacott did when he ultimately took a decision independently of the AI's own actions. Yet AI Steve shows us what artificial intelligence can already do for the relationship between representatives and those they represent. Representatives can use different platforms to improve their relationship with the electorate and gather the information needed to make decisions that are more efficient and, ultimately, more representative. Properly encouraged and sensibly used, this could radically transform the relationship between the political class and citizens. The AI Steve experiment shows that human positions — whether ideological or simply matters of preference — will ultimately shape the representative relationship. But representatives can refine their proposals and actions to the extent that they succeed in engaging citizens in providing the input their work needs. If politics wins citizens over and persuades them to participate, generative artificial intelligence can be a crucial ally in improving that everyday relationship. It may even help curb the disaffection observed in recent years. It remains for us to reflect on how to use technology in general, and artificial intelligence in particular, to improve the relationship between people and politics.



[Fernando Domínguez Sardou](https://www.linkedin.com/in/fernando-dominguez-sardou/)



*Political scientist, lecturer and political analyst. PhD candidate in Political Science at Universidad Nacional de San Martín, Argentina, and student on the Master's in Marketing, Communication and Political Consulting at Institución Educativa ALEPH.* X: [@ferdsardou](https://x.com/ferdsardou)