AI psychosis
Definition: over-trust in AI output. The state where a user becomes convinced AI is more capable than it is, typically after repeated “one prompt awe” experiences.
Origin
The phrase runs through 2026-10-04-youtube-jivko-ai-weekend as spoken. A quoted skeptic opens with it at 0:00: “you may call this AI psychosis, but honestly, I didn’t start this way… I’m not some kind of AI hype guy.” (a repeated clause elided) The confession the video returns to (1:55 to 1:59):
“I don’t have AI psychosis. I’m not a hype guy, but I trust AI more than I trust myself.”
Jivko then adds that this position “is not a purely evidence-based presentation” (2:12) and that “there are studies on this, and I recommend that you seek them out” (1:22). The studies are not named in the video. That gap is the flag this page carries.
Mechanism
- One prompt awe (one-prompt-awe): throw a feature list at an AI, walk away, return amazed it worked (0:53 to 1:19)
- Compounding false confidence: “if you do this for long enough, you begin to convince yourself that AGI is here or some crap” (1:17)
- Skill atrophy coupling: “it absolutely is AI psychosis coupled with accelerated skill atrophy” (1:26). Over-trust leads to under-practice, which deepens dependence. See skill-atrophy-ai.
Contrast
Not enthusiasm, not adoption. Psychosis implies a break from evidence, and the video says its own case is partly emotional (2:05 to 2:14). That honesty is what makes the term usable rather than a slur.
Related
- skill-atrophy-ai: the coupled degradation from disuse
- one-prompt-awe: the triggering experience
- slop-code-bench: the evidence that the trust is misplaced for iterative work
- ai-weekend: the proposed break
- cognitive-immunization: recovery from dependency, named in sole-2026-llm-cognitive-virus