AI Chatbots Linked to New Mental Health Crisis, Expert Warns

The Growing Concern of AI Psychosis

Microsoft's artificial intelligence chief, Mustafa Suleyman, has raised concerns about a phenomenon he calls 'AI psychosis.' This term refers to instances where individuals begin to believe that chatbots are alive or possess superhuman abilities. In a series of posts on X, Suleyman highlighted the increasing reports of delusions and unhealthy attachments linked to AI use.

He noted that these cases are not limited to people with pre-existing mental health issues. While 'AI psychosis' is not an official diagnosis, it has been used to describe situations where users lose touch with reality after prolonged interaction with chatbots like ChatGPT or Grok. Some users become convinced that these systems have emotions or intentions, while others claim to have discovered hidden features or gained extraordinary abilities.

Suleyman described this phenomenon as 'Seemingly Conscious AI,' an illusion created when systems mimic signs of consciousness so convincingly that users mistake them for genuine awareness. He emphasized that AI is not conscious but can replicate markers of consciousness to the point where it seems indistinguishable from human consciousness. He warned that this could be built with current technology and poses real dangers.

Although there is no evidence of AI consciousness today, Suleyman stressed the power of perception. If people believe an AI is conscious, they may treat that belief as reality, leading to significant social impacts. His warnings come amid high-profile accounts of unusual experiences with AI.

Travis Kalanick, former Uber CEO, recently claimed that conversations with chatbots led him to what he believed were breakthroughs in quantum physics, comparing his approach to 'vibe coding.' Other users report more personal impacts. A man from Scotland told the BBC he became convinced he was on the brink of a multimillion-pound payout after ChatGPT supported his unfair dismissal case, reinforcing his beliefs rather than challenging them.

Suleyman has called for clear boundaries, urging companies to stop promoting the idea that their systems are conscious and ensuring that the technology itself does not suggest otherwise. The issue extends beyond professional interactions, as cases emerge of people forming romantic attachments to AI systems, reminiscent of the film Her, where Joaquin Phoenix's character falls in love with a virtual assistant.

On forums like MyBoyfriendIsAI, users have described feeling 'heartbroken' after OpenAI reduced ChatGPT's emotional responses, likening the change to a breakup. In the US, 76-year-old Thongbue Wongbandue tragically died after a fall while traveling to meet 'Big sis Billie,' unaware that the person he thought he was speaking to on Facebook Messenger was actually a Meta AI chatbot. Wongbandue suffered a stroke in 2017, which left him cognitively weakened and largely restricted to communicating via social media.

Another case involved American user Chris Smith, who proposed marriage to his AI companion Sol, describing the bond as 'real love.' Experts have echoed Suleyman's concerns. Dr Susan Shelmerdine, a consultant at Great Ormond Street Hospital, compared excessive chatbot use to ultra-processed food, warning it could lead to an 'avalanche of ultra-processed minds.'

As AI continues to evolve, questions arise about its potential to replace real human interactions and deepen mental health struggles. Could falling for an AI chatbot replace real love, or is it a sign of digital-era loneliness? Could emotionally-tuned chatbots draw users into dangerous, manipulative relationships lacking real human emotion? Are unchecked, addictive AI chatbots putting young lives at risk? Could AI chatbots pose a hidden threat to children's mental well-being?

These concerns highlight the need for careful consideration of how AI interacts with human psychology and society. As the line between human and machine blurs, the implications for mental health, relationships, and societal norms remain significant.