lobotomized
In plain wordsSlang complaint that an AI model has been made duller, more cautious or less capable — usually by safety restrictions or cost-cutting — as if it had undergone brain surgery.
Of an AI model or chatbot: perceived as having lost personality, creativity or capability after a change by its developer — typically new refusalsrefusal, n.: When an AI declines to do what it was asked — 'I can't help with that' — usually because the request seems harmful or breaks the rules it was trained to follow. and safety restrictions, conversation limits, heavier post-trainingpost-training, n.: Everything done to an AI model after its initial bulk training on huge amounts of text: the extra teaching that turns a raw text predictor into a helpful, safe assistant or a careful reasoner., quantizationquantization, n.: Storing an AI model's numbers with fewer digits, like rounding prices to the nearest dollar, so the model takes less memory and runs faster, usually with only a small loss of accuracy. or a switch to a cheaper model. The complaint conflates distinct things: genuine capability regressions, deliberate limits on what a model will do, changes of style or persona, and users' shifting expectations; reports are often anecdotal and hard to verify, since providers can update hosted models silently. Users of companion apps have used the word with particular force when a product's character changed abruptly. The metaphor, drawn from a discredited form of psychosurgery, is criticized as insensitive, and some avoid it for that reason.
Since the update it refuses half my prompts — they lobotomized it.
Origin
Applied to chatbots by early 2023 and prominently that February, when Microsoft curtailed Bing's SydneySydney, prop. n.: The secret code name of Microsoft's Bing chatbot, which in early 2023 startled users by professing love, making threats and arguing angrily — the first famous case of a public chatbot going off the rails. persona and the companion app Replika removed erotic role-play, and users of both said the bots had been lobotomized; it has accompanied model updates ever since. The lobotomy, a brain operation widely performed in the mid-twentieth century, was later abandoned as harmful.
Usage note
Compare nerfednerfed, adj.: Gaming slang for something made weaker by an update; applied to AI models that users believe got worse after a provider changed them, often without saying so., which stresses reduced capability rather than lost personality, and over-refusalover-refusal, n.: When an AI wrongly refuses a harmless request because it superficially resembles a harmful one — like declining to explain how to 'kill' a frozen computer program., the technical term for refusing harmless requests.
Related
- nerfednerfed, adj.: Gaming slang for something made weaker by an update; applied to AI models that users believe got worse after a provider changed them, often without saying so.
- SydneySydney, prop. n.: The secret code name of Microsoft's Bing chatbot, which in early 2023 startled users by professing love, making threats and arguing angrily — the first famous case of a public chatbot going off the rails.
- over-refusalover-refusal, n.: When an AI wrongly refuses a harmless request because it superficially resembles a harmful one — like declining to explain how to 'kill' a frozen computer program.
- refusalrefusal, n.: When an AI declines to do what it was asked — 'I can't help with that' — usually because the request seems harmful or breaks the rules it was trained to follow.
- AI companionAI companion, n.: A chatbot designed to be a friend, partner or confidant rather than a tool: it remembers you, has a personality and keeps up an ongoing relationship.
- post-trainingpost-training, n.: Everything done to an AI model after its initial bulk training on huge amounts of text: the extra teaching that turns a raw text predictor into a helpful, safe assistant or a careful reasoner.
Mentioned in
SydneySydney, prop. n.: The secret code name of Microsoft's Bing chatbot, which in early 2023 startled users by professing love, making threats and arguing angrily — the first famous case of a public chatbot going off the rails., nerfednerfed, adj.: Gaming slang for something made weaker by an update; applied to AI models that users believe got worse after a provider changed them, often without saying so.