Are Language Models Models?
This work addresses a foundational issue in AI and cognitive science by challenging the interpretation of language models, which is important for researchers and practitioners to avoid overestimation of their theoretical value.
The paper critiques the claim that language models serve as cognitive models, arguing that this assertion is unsupported at Marr's three levels of analysis and that labeling them as such overstates their capabilities and contributes to hype.
Futrell and Mahowald claim LMs "serve as model systems", but an assessment at each of Marr's three levels suggests the claim is clearly not true at the implementation level, poorly motivated at the algorithmic-representational level, and problematic at the computational theory level. LMs are good candidates as tools; calling them cognitive models overstates the case and unnecessarily feeds LLM hype.