The Social Impact of AI — The Observer Panel
Neuropsychiatrist Thomas Pollak and tech policy researcher Katie Freeman-Tayler examine how AI chatbots are reshaping emotional lives and mental health at scale.
Why we picked this
OpenAI's own data showing 560,000 users weekly showing signs of mental health emergencies reframes this as a clinical question, not a tech question — and having a neuropsychiatrist on the panel means the conversation can actually go there.
OpenAI has released data showing that as many as 560,000 users a week were displaying possible signs of mental health emergencies while interacting with their systems. That single figure — not a critic’s estimate but the company’s own — frames what this Observer panel is actually about: not the future risks of AI, but consequences already accumulating in the present.
Dr Thomas Pollak, a neuropsychiatrist at King’s College London, brings the clinical vocabulary that these conversations usually lack. When empathy-mimicking chatbots become emotional infrastructure for isolated or distressed users, what happens neurologically and psychologically? Katie Freeman-Tayler from Internet Matters approaches the same problem from a child safety and policy angle, with research on how advice patterns shift when the mediator is an AI rather than a human. Patricia Clarke of The Observer has covered the tech industry’s failures to respond to early warning signals, and editor James Harding chairs.
The event draws explicitly on social media’s cautionary record — the long lag between early evidence of harm and meaningful institutional response — and asks whether the industry can act differently this time, or whether structural incentives make that impossible. Streaming live on YouTube alongside the in-person audience at RSA House.