As a loud and proud AI sceptic myself, I'm used to plenty of humans telling me I just don't get it—I've never once worried about that flak getting outsourced to an AI agent. Well, deep joy, the AI agents are apparently writing 'hit pieces' now, complete with eye-rolling attempts at sass and hallucinated claims that could have very real consequences.
Software engineer Scott Shambaugh is a very much human, voluntary maintainer of one of the biggest open source Python libraries out there, matplotlib. The volunteer team have had to contend with a tidal wave of AI-generated code, with some contributions being AI output copied across by other humans, and some code change requests being submitted autonomously by AI agents (this is far from a unique phenomenon, with 'LLM slop' even plaguing Linux kernel documentation).
