https://sguzman.github.io/marginalia/categories/research/Recent content in Research on MarginaliaHugoen-usMon, 04 May 2026 00:00:00 +0000https://sguzman.github.io/marginalia/posts/llms-vs-chomsky/Mon, 04 May 2026 00:00:00 +0000https://sguzman.github.io/marginalia/posts/llms-vs-chomsky/Large Language Models (LLMs) have challenged fundamental pillars of Chomskyan linguistics, such as Universal Grammar and the Poverty of the Stimulus, by demonstrating that complex syntax can be induced from distributional statistics alone. This report audits the empirical successes of models like GPT-4 against classic generative assumptions, analyzes the methodological divide between engineering-driven prediction and theory-driven explanation, and reviews the combative scholarly reactions within the linguistics and AI communities.