Exclusive: Guardian analysis reveals that dozens of policy submissions from across the political spectrum incorrectly summarise real research and invent or wrongly cite sources
Good article, they built an automated reference lookup to systematically identify submissions that contained hallucinated references, and also searched for submissions that brazenly included ChatGPT reference links.
Good article, they built an automated reference lookup to systematically identify submissions that contained hallucinated references, and also searched for submissions that brazenly included ChatGPT reference links.
The fact that we’ve allowed the industry to get “hallucination” into our language instead of “fabrication” or “bullshit” like it really is, is sad.
Beyond sad. It is a threat to the running of government, to policy making, and ultimately to citizens. Who/what surveills AI? More AI?