The congressman refuses to debate, so his opponent uses an AI deputy

Let’s say you want to run for Congress long-term as an independent in a deeply Democratic Virginia district, but the incumbent in the race won’t give you airtime to debate your platform before voters. You could simply run social media ads to promote your candidacy, or perhaps present your case in a YouTube video commenting on the incumbent’s platform. Maybe contact the local newspaper to get an interview? Or you could simply use generative AI trained on the incumbent’s backlog of public comments and published material.

That’s what Bentley Hensel wants to do in Virginia’s 8th Congressional District, where he is running as an independent challenger against Don Beyer, who has rejected requests for another debate, saying a September forum would be sufficient. Beyer won the district by nearly two-thirds in 2022. So why bother debating Hensel, an unnamed software engineer who is again running as an independent candidate?

For some, it may be difficult to suspend disbelief when listening to a simulacrum of a real person. ChatGPT’s advanced voice mode is so uncanny in its ability to speak in a natural and expressive manner that it is difficult to even distinguish it from a real person. And companies like Character.ai and Replika are gaining popularity because they offer chatbots as forms of companionship. It’s not crazy to think that voters, especially older ones, feel like they’re attending a real debate.

Hensley told Reuters that DonBot, as the bot is called, is trained using the ChatGPT API on Beyer’s official websites, press releases and Federal Election Commission data. The bot is intended to provide accurate answers, although that isn’t exactly reassuring to anyone who has actually used a chatbot before. And in fact, Reuters tested DonBot and found that while it largely provided clear answers to policy questions, it made a mistake by saying that Beyer had not endorsed anyone for president, when in fact he had endorsed Democratic candidate Kamala Harris. Still, the bot is unlikely to induce wild hallucinations since it was trained on a narrow data set.

Beyer didn’t say it Reuters whether he will take action to stop the online debate taking place on Oct. 17, but a spokesperson told the outlet that Beyer ““continues to be a leading voice in Congress on the need to improve regulation of artificial intelligence, including legislation to prevent nefarious actors from using AI to spread election misinformation.” So it sounds like he is not exactly thrilled.

Legal experts spoken to Reuters said that use of the bot would likely be permissible as long as Hensel clearly disclosed that he was not actually speaking to the real Don Beyer. For Hensel, just the attention he gets from developing the bot is probably worth a lot. But it seems tricky when older voters and others mistakenly believe that a bot that is likely to make some mistakes is actually representing a candidate. We know that it is already easy for older generations to fall for AI-generated images.

At least 26 states have taken steps to regulate the use of generative AI in election-related communications, with some even going so far as to ban deepfakes of politicians entirely. Unsurprisingly, there was little movement at the federal level.