In a new viral video, Sen. Bernie Sanders tried to expose how the AI industry is a threat to Americans’ privacy, but ended up demonstrating how AI chatbots’ tendency to agree with and flatter users can make them less tools of discovery themselves and more like mirrors of users’ own beliefs.
We have seen this problem before with the rise of people suffering from “AI psychosis” where AI chatbots reinforce the irrational thoughts and beliefs of mentally unstable people. In some cases, this dark pattern even led users to take their own lives, several lawsuits allege.
In Sanders’ case, the AI pandering manifested as an AI chatbot that shaped its answers to suit the politician.
It’s worth noting that the interview begins with Sanders introducing himself to Claude (whom he mistakenly refers to as an AI “agent”). This is a move that could impact chatbot responses.
Sanders then asked about AI companies’ data collection practices and other privacy concerns, and Claude responded by agreeing with what politicians want to hear. Part of the reason is the way Sanders frames the question: “What would surprise the American people in terms of knowing how that information is collected?” “How can we trust AI companies to protect our privacy when they use people’s personal information to make money?” These guiding questions encourage the chatbot to accept the premise of the question and come up with an appropriate response. That’s how these things work.
And when Claude’s answers suggested that a topic was more complex or nuanced than Sanders had framed, Sanders disagreed, prompting the chatbot to acknowledge with some AI self-deprecation that the senator was “absolutely right.”
Considering chatbots to be a source of universal truth rather than a tool subject to user influence, the flattering nature of AI can lead people down a dangerous path.
It’s unclear whether Mr. Sanders knows this to be true and simply doesn’t care (after all, it’s just an ad!) or whether he really thinks he tricked Mr. Claude into becoming a whistleblower in the AI industry.
And, of course, given that this was a staged “interview,” there’s also the question of whether Sanders’ team coaxed the chatbot to respond in a certain way.
There are real concerns when it comes to data collection and privacy, but things aren’t as black and white as the AI response in this video suggests.
We already live in a world where companies collect and sell the data of online users at scale, and have been for years. We know that social media giants like Meta are turning personalized ads into billion-dollar cash printing machines. And thanks to regular transparency reports from big tech companies, we know that governments around the world regularly request access to user data for their own purposes.
While AI may represent a new medium for lawmakers to potentially regulate, personal data has long powered the digital economy. (Ironically, Anthropic is an AI company that has pledged not to make money from personalized ads, despite what its response to Sanders suggested.)
The whole conversation between Sanders and Claude is beside the point for people who understand how AI chatbots work, but at least it definitely gave us a great new meme.
