It’s been a dizzying few weeks of “is AI reasoning and / or alive” discourse, with the attendant pushback that these systems are just fancy autocomplete. But back in March, Margaret Mitchell, one of the authors of the famous Stochastic Parrots paper, took to Medium to clarify something: AI, as a whole, is not a stochastic parrot — LLMs are. Critics are missing a fundamental difference between the two technologies, even though they’re related. The post has been going back around, and it’s worth a read.
There is a vast array of technology called “AI” that is not reducible to LLMs, and many current AI systems that utilize LLMs also leverage a variety of other technologies, including hand-written-rules, deterministic (non-stochastic) programs, various algorithms, and non-language models. This means that “AI”, broadly, is not equivalent to a large language model; it is not “just a stochastic parrot”.
[Margaret Mitchell]
