What's this about?
Mollick skimmed the websites of his favourite hard science-fiction authors and found that most of them hate LLMs. The breakdown he reports is the interesting part: a surprising number because they think it is a useless stochastic parrot, some over intellectual property, and only a minority because of existential risk.
What is a "stochastic parrot"?
A 2021 coinage arguing that a language model is a system for stitching together plausible sequences it has seen, with no understanding — a parrot with a random number generator. It became the central sceptical position in AI. The reason Mollick's finding is surprising is that hard science fiction is the genre most associated with taking machine intelligence seriously, so you would expect existential concern rather than dismissal.
What's the community saying?
- The distribution is the finding: dismissal outnumbers doom among exactly the people who spent careers imagining machine minds.
- One reading is craft-based — a professional writer evaluates generated prose as prose and finds it bad, which is a different judgement from evaluating capability.
- The IP objection is the one with legal traction and the least discussed here.