Prompting as Scientific Inquiry

TL;DR

This paper argues that prompting is a scientific method for studying and controlling large language models, positioning it as behavioral science rather than mere art or workaround.

Contribution

It reframes prompting as a scientific approach, emphasizing its role in understanding and guiding LLM behavior as a form of behavioral science.

Findings

Prompting unlocks major LLM capabilities like few-shot learning and chain-of-thought.

Treating LLMs as trained, opaque organisms shifts prompting from art to science.

Prompting provides insights into model behavior through native language interface.

Abstract

Prompting is the primary method by which we study and control large language models. It is also one of the most powerful: nearly every major capability attributed to LLMs-few-shot learning, chain-of-thought, constitutional AI-was first unlocked through prompting. Yet prompting is rarely treated as science and is frequently frowned upon as alchemy. We argue that this is a category error. If we treat LLMs as a new kind of complex and opaque organism that is trained rather than programmed, then prompting is not a workaround: it is behavioral science. Mechanistic interpretability peers into the neural substrate, prompting probes the model in its native interface: language. We contend that prompting is not inferior, but rather a key component in the science of LLMs.