I've enjoyed AI as a gimmick, but due to multiple long term engagements with explicit requirements not to use AI, I've never used it on a job meaningfully - though I do appreciate it can do wonderful things with minor tasks. The case for Skepticism Quality of work - AI work requires review. We've all seen the shit it pulls, from hallucinations, to rambling sentences, to just a general inability to genuinely and intuitively analyse data provided. Anecdotal evidence here, but I find I'm much more effective at reviewing a piece of work if I've written it from scratch, than if the whole thing is created by somebody/something else. Reviewing work is arguably the harder skill than writing it, and AI makes it so our jobs become more the latter than the former. Long term model collapse / Death of Innovation - in general terms, LLMs create content based on an existing database of works, mapping commonalities based on prompt to generate 'new' material.…