Large Language Models and Impossible Language Acquisition: "False Promise" or an Overturn of our Current Perspective towards AI

TL;DR

This paper investigates whether large language models can learn impossible languages, challenging Chomsky's critique by conducting experiments on GPT-2 and LSTM models with syntactically impossible language tasks.

Contribution

It provides empirical evidence on LLMs' capacity to learn impossible languages and proposes a new theoretical perspective shifting from Chomsky's rationalist view to empiricism and functionalism.

Findings

GPT-2 models show lower loss and faster convergence on natural language than impossible languages.

LSTM models show minimal differences across language conditions.

Reversed language condition causes the largest divergence in GPT-2 performance.

Abstract

In Chomsky's provocative critique "The False Promise of CHATGPT," Large Language Models (LLMs) are characterized as mere pattern predictors that do not acquire languages via intrinsic causal and self-correction structures like humans, therefore are not able to distinguish impossible languages. It stands as a representative in a fundamental challenge to the intellectual foundations of AI, for it integrally synthesizes major issues in methodologies within LLMs and possesses an iconic a priori rationalist perspective. We examine this famous critique from both the perspective in pre-existing literature of linguistics and psychology as well as a research based on an experiment inquiring into the capacity of learning both possible and impossible languages among LLMs. We constructed a set of syntactically impossible languages by applying certain transformations to English. These include reversing whole sentences, and adding negation based on word-count parity. Two rounds of controlled experiments were each conducted on GPT-2 small models and long short-term memory (LSTM) models. Descriptive analysis of single-run training trajectories shows that GPT-2 small models exhibit lower final loss, faster convergence, and lower perplexity on natural language compared to impossible language conditions, with the reversed condition showing the largest departure (loss ratios up to 2.25 * natural). LSTM models, by contrast, show minimal differences across conditions. Given the single-run nature of our experiments (n=1 per condition), we report descriptive comparisons and caution that formal statistical inference is precluded. Based on theoretical analysis and descriptive empirical findings, we propose a new vision within Chomsky's theory towards LLMs, and a shift of theoretical paradigm outside Chomsky, from his "rationalist-romantics" paradigm to functionalism and empiricism in LLMs research.