Using cognitive psychology to understand GPT-3

TL;DR

This paper uses cognitive psychology tools to evaluate GPT-3's decision-making, reasoning, and learning abilities, revealing strengths in some tasks and notable weaknesses in others, to better understand its capabilities.

Contribution

It introduces a cognitive psychology framework to systematically assess GPT-3's cognitive-like behaviors and limitations, advancing understanding of large language models' decision processes.

Findings

GPT-3 performs well on vignette-based tasks and decision-making from descriptions.

GPT-3 outperforms humans in a multi-armed bandit task.

GPT-3 struggles with causal reasoning and is sensitive to small perturbations.

Abstract

We study GPT-3, a recent large language model, using tools from cognitive psychology. More specifically, we assess GPT-3's decision-making, information search, deliberation, and causal reasoning abilities on a battery of canonical experiments from the literature. We find that much of GPT-3's behavior is impressive: it solves vignette-based tasks similarly or better than human subjects, is able to make decent decisions from descriptions, outperforms humans in a multi-armed bandit task, and shows signatures of model-based reinforcement learning. Yet we also find that small perturbations to vignette-based tasks can lead GPT-3 vastly astray, that it shows no signatures of directed exploration, and that it fails miserably in a causal reasoning task. These results enrich our understanding of current large language models and pave the way for future investigations using tools from cognitive psychology to study increasingly capable and opaque artificial agents.