Very interesting article by @melaniemitchell.bsky.social one of the most thoughtful critics of LLMs. Hallucination is built into LLMs. One urgent implication to me is that we should not use autonomous LLM based agents in any reasonably important scenario today, despite all the hype around them.
