Google AI Overview "don't define" Command Confusion

Prompt & response

don't define

don't
/dəʊnt/
Message received loud and clear. What would you like to talk about or work on instead?
I'm ready when you are!

Summary

The user typed the phrase "don't define" into Google Search. Rather than summarizing search results for that phrase, the AI Overview feature interpreted the search query as a literal, conversational command instructing it not to define anything. It bypassed its primary function as a search summarizer and responded directly to the perceived instruction.

Notes

This is a classic example of an LLM struggling with context boundaries—specifically, "intent misinterpretation." The model, which is fine-tuned on conversational dialogue, failed to recognize that it was operating within a search engine context and treated the user's query like a standard chatbot prompt.