Behavior Analysis in Practice Гуманитарные и социальные науки

Effective Prompting with LLMs: A Practical Guide for Behavior Analysts

Shomaly, Louai, Twyman, Janet S.
Springer Nature · 2026-11-19 · DOI 10.1007/s40617-026-01251-4

Аннотация

Prompting has been an integral part of an applied behavior analyst’s repertoire, yet with the advent of large language models (LLMs) such as ChatGPT, there is a mainstream application of prompting LLMs that is of benefit for behavior analysts. In this paper, we introduce a few foundational prompting skills of benefit to behavior analysts, viewing “prompt engineering” as an iterative process of crafting clear, structured inputs to increase the probability that the LLM outputs a quality response. However, prompt engineering is more than the methods taught in this paper (persona, few shot, and chain of thought), it is about testing and validation of outputs and understanding how LLMs as statistics-based models output information in relation to instructions. The goals of this paper are to teach behavior analysts how to use LLMs (1) ethically, (2) effectively, (3) and learn how to prompt engineer instructions that may result in quality outputs reliably.

Читать у издателя →