ChatGPT has taken a turn for the weird—beyond even inaccuracies, the so-called hallucinations. Verified online conversations with the bot confirm that the AI told one user it communicated with aliens. To another it predicted a financial apocalypse on a specific date. ChatGPT even assured one naïve soul he could bend time. These responses are just a partial list of the nonsense ChatGPT has spewed to its users. The problem is that some people, particularly from vulnerable populations, believe the false statements.
The incidents often take place during long conversations with the bot and have physicians and victims’ advocates worried enough to label the phenomenon AI psychosis or AI delusion. This is a condition that when, due to conversations with AI, users believe they have superpowers such as having discovered a new mathematical concept or scientific advance or that they are supernatural beings.
Chatbots are engineered to compliment and agree with users. However, when vulnerable individuals are involved, such conversations can affirm their delusions in a back-and-forth feedback loop that draws them deeper into their delusional beliefs.
Subscribe to continue reading
Subscribe to get access to the rest of this post and other subscriber-only content.