fAIr science — How AI can support a transparent biopsychological research cycle

Le résumé fourni par la source

Biological psychology links mental phenomena and physiological processes, advancing our understanding of the human condition and informing clinical applications. Unlocking this potential demands rigorous and transparent methodology. Artificial Intelligence (AI), encompassing generative (e.g., large language) models and machine-learning classifiers, promises to advance open science in this field through automated literature synthesis, experimental design assistance, and enhanced code documentation. Yet the same tools risk undermining transparency through algorithmic opacity, systematic biases, and non-reproducible outputs. In this paper, we critically examine how AI can support or hinder open science practices throughout the biopsychological research cycle. We discuss current AI applications at each research stage and explore their potential benefits (e.g., accelerated systematic reviews, stimulus generation, metadata formatting) alongside relevant concerns (e.g., premature study exclusion, confabulated citations, amplified publication bias). To provide practical guidance, we offer recommendations on how AI may be used to promote the adoption and implementation of open science practices at each step of the biopsychological research cycle. Alongside potential pitfalls and a future vision, these guidelines equip biopsychological researchers with actionable support for integrating AI-based methods into their work in line with open science principles.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.