Deconstructing sentence disambiguation by joint latent modeling of reading paradigms: LLM surprisal is not enough

Shared on Bluesky by 2 AI experts

• T Tal Linzen @tallinzen.bsky.social amplified

  @dariopaape.eurosky.social

  Updated preprint with @tallinzen.bsky.social and @shravanvasishth.bsky.social : "LLM surprisal is necessary but not sufficient to capture English garden-path effects: Evidence from joint latent modeling of reading paradi…

  View on Bluesky →
• C Cognitive Science Society @cogscisociety.bsky.social amplified

  @dariopaape.eurosky.social

  Who's attending #CogSci2026 in Rio next week? 🇧🇷 I will be presenting joint work with @shravanvasishth.bsky.social and @tallinzen.bsky.social on LLM surprisal in garden-path sentences on Friday! arxiv.org/abs/2602.04489

  View on Bluesky →