TY  - RPRT
TI  - Neither Stochastic Parroting nor AGI: LLMs Solve Tasks through Context-Directed Extrapolation from Training Data Priors
AU  - Harish Tayyar Madabushi
AU  - Melissa Torgbi
AU  - Claire Bonial
PY  - 2025
UR  - https://arxiv.org/abs/2505.23323
ID  - 2505.23323
ER  -