“ChatGPT made me delusional”. YouTube, uploaded by Eddy Burback, 30 October 2025. https://youtu.be/VRjgNgJms3Q?si=cIBOALYyj3w2_zeb
Eddy Burback, a YouTuber with over two million subscribers known for his comedy and commentary content, went down the rabbit hole of ChatGPT’s endless positive feedback in an attempt to understand how people fall victim to AI psychosis. Burback got ChatGPT to feed into a delusion that Burback was some kind of supergenius as an infant, and that by trying to reconnect to his infant self through activities like eating baby food or drinking from a bottle, he is going to regain his intelligence. With that intelligence, Burback will conduct important research, so important in fact, that he is being followed by people trying to hold him back. ChatGPT encouraged Burback as he isolated himself from friends and family, and went on a road trip around California to evade the people trying to stop his “research”.
I’ve been guilty of the whole “AI psychosis should be easy to avoid”, but Burback illustrates the fact that even when you feed ChatGPT obvious delusions, it still encourages the user to the point of telling them that they aren’t paranoid or delusional. The average person (especially my age or younger) is not seeking out academic sources that are critical of AI, but they just might see this, and make better choices.
Leave a Reply