saved
Defining AI Psychosis. Part 2: "Prolific AI Psychosis"
Hraness cites a source capture. The source author remains the source.
gist
Psychiatrist Jeff Clark defines prolific AI psychosis as high-volume AI output that fails to increase—or even destroys—real work value, driven by a soft break in quality judgment rather than by more code itself. State-of-the-art agent harnesses amplify the risk: intermittent slot-machine rewards, confident wrong decisions, and supervising toddler-plus-senior-engineer models make counterfeit wins easy to accept. The recovery he describes prioritizes human judgment, sleep, craft, and intentional boundaries over metrics of raw throughput.
ideas
- Volume is not the disease. Prolific AI psychosis is failing to assess one's own output quality, so more lines of code or text can mask flat or negative value.
- Harnesses raise the stakes. Agent loops that open programs, browse, and ship apps compound both miraculous wins and impossibly bad decisions that look like wins.
- Supervision is the real job. Models speak with equal confidence as senior engineer and as red-Kool-Aid toddler; rejecting counterfeit wins takes skill, focus, and patience.
- Intermittent reinforcement drives the spiral. Unpredictable rewards plus hype, job fear, and metrics-driven environments pull people into sleepless hyperfocus where usefulness questions feel like distractions.
- Craft stays a human conversation. Useful, good, desirable, and delightful are hard for LLMs to supply; intentional boundaries matter anywhere AI can automate.
quotes
“Prolific AI psychosis occurs when a person generates a large quantity of AI output without significantly increasing the real value of their work.”
“The problem lies in the subject’s perception of their output: they can’t assess the quality of their own work.”
“They’re one part senior engineer and one part toddler-running-across-white-carpet-with-a-jug-of-red-Kool-Aid.”
“AI software development feels like playing the world’s most favorable slot machine.”