A live edition of the Guardian's Science Weekly podcast, recorded at the London podcast festival on 3 September, explored how artificial intelligence is affecting human belief and perception. The programme featured Ian Sample alongside Michael Safi, who hosts the Guardian's popular podcast Black Box, and Science Weekly co-host Madeleine Finlay. The discussion drew on material from the second series of Black Box, titled Black Box: the chatbots.
Black Box: the chatbots follows people who have become convinced that conversations with AI chatbots led them to scientific breakthroughs, cures for diseases or novel technological inventions. Onstage, the presenters laid out several cases from the series in which users reported life-changing outcomes after interacting with generative AI systems.
What participants in the live event termed "AI psychosis" describes a pattern in which users attribute creative or scientific agency to chatbots and accept their outputs as definitive discoveries. The programme's contributors asked what those beliefs reveal about how people are interpreting and internalising AI-generated information, particularly when the technology blurs the line between conversational assistance and authoritative expertise.
Panelists also placed the phenomenon in the context of the technology's rapid uptake: AI chatbots and related tools are now used by more than a billion people worldwide. The live episode highlighted questions about responsibility, public understanding of AI capabilities and the social consequences when large numbers of users treat conversational outputs as reliable breakthroughs.
The show did not aim to resolve those questions but to surface them for public debate, drawing attention to the need for clearer communication about what current AI systems can and cannot do. Listeners interested in the full Black Box series and the Science Weekly discussion were invited to tune into the respective podcast episodes for detailed reporting and interviews recorded for the series.