How LLMs Might Think
For philosophers and AI researchers, it offers a conceptual clarification about the nature of LLM cognition, but is incremental.
The paper critiques an argument that LLMs do not think, and proposes that if LLMs think, they likely engage in arational, associative thinking. No concrete numbers are provided.
Do large language models (LLMs) think? Daniel Stoljar and Zhihe Vincent Zhang have recently developed an argument from rationality for the claim that LLMs do not think. We contend, however, that the argument from rationality not only falters, but leaves open an intriguing possibility: that LLMs engage only in arational, associative forms of thinking, and have purely associative minds. Our positive claim is that if LLMs think at all, they likely think precisely in this manner.