CLAIJun 12, 2023

Large language models and (non-)linguistic recursion

arXiv:2306.07195v17 citationsh-index: 12
Originality Synthesis-oriented
AI Analysis

This addresses the problem of understanding the emergence of uniquely human cognitive properties in AI models, but it is incremental as it builds on prior work on metalinguistic abilities in LLMs.

The study investigated whether GPT-4 can exhibit recursive behavior, a key feature of human language, by designing prompts to elicit and analyze recursive structures. The results show that GPT-4 can produce and analyze recursive structures when explicitly prompted.

Recursion is one of the hallmarks of human language. While many design features of language have been shown to exist in animal communication systems, recursion has not. Previous research shows that GPT-4 is the first large language model (LLM) to exhibit metalinguistic abilities (Beguš, Dąbkowski, and Rhodes 2023). Here, we propose several prompt designs aimed at eliciting and analyzing recursive behavior in LLMs, both linguistic and non-linguistic. We demonstrate that when explicitly prompted, GPT-4 can both produce and analyze recursive structures. Thus, we present one of the first studies investigating whether meta-linguistic awareness of recursion -- a uniquely human cognitive property -- can emerge in transformers with a high number of parameters such as GPT-4.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes