Could AI chatbots undo the harms of social media? | FT Tech

Large language models elevate expert consensus and moderate views, in sharp contrast to social platforms with research shows AI useage tends to be depolarising, says John Burn-Murdoch You can enable subtitles (captions) in the video player Could AI chatbots undo the harms of social media? The past 15 years have been characterised by waves of populism, polarisation, and eroding trust in experts. Many factors are at play, but one is the dramatic change in our information environment brought about by social media. Every media revolution transforms who shares information and what form it takes. Some technologies are fundamentally democratising and polarising, widening the pool of publishers and views beyond a narrow elite and amplifying radical and anti-establishment voices. TikTok and the printing press arrived 600 years apart but share some of these characteristics. Others pushed the opposite way. Radio and television had high barriers to entry, creating a monopoly for establishment voices and moderate positions. As AI chatbots now take off, it's worth asking which category they fall into. And there's good reason to believe it's the latter. Social media firms make money from attention, which means rewarding sensationalism with little regard for truth. AI firms, by contrast, are competing to serve customers paying for accurate, reliable tools, often for business-critical purposes. When chatbots surface harmful content, they're on the hook in the way social platforms have avoided. That's the theory, but what about the evidence? Using detailed data on tens of thousands of responses to questions on policy and beliefs, I tested how the major AI chatbots shape conversations about politics and society. The results strongly support the theory of AI as depolarising. While different platforms behave in subtly different ways, all of them nudge people away from the most extreme positions amplified on social media and toward more moderate, expert-aligned stances. Even accounting for chatbots' sycophantic tendencies, conversations with AI still direct hardline partisans of both flanks away from extreme beliefs. They're also far less likely to surface or support conspiracy theories than the content on social platforms. Now, these are only preliminary findings, and AI usage and behaviour could still shift. But for now, there's cause for optimism that the next information revolution may take us in a less corrosive direction than the last.