MEPSAI: Enhancing Meta-Research in Psychology by Generative Al

This project addresses the critical bottleneck of traditional manual meta-analysis research by leveraging generative artificial intelligence and Large Language Models to revolutionize how scientific evidence is synthesized. By developing modular prompting pipelines for automated data extraction, optimizing research designs, and ensuring automated quality assurance, the project aims to significantly enhance the scalability, credibility, and integrity of meta-research.

Meta-research – the systematic study of the research process itself – is fundamental to advancing the credibility and integrity of psychological science. However, the exponential growth in the volume of research output has rendered traditional meta-research methods, such as manual meta-analysis, increasingly unfeasible due to their time-­consuming and nature, labor-intensity, and susceptibility to human error. To address this critical challenge, this project proposes a potentially paradigm-shifting solution: leveraging the capabilities of generative artificial intelligence (AI) to revolutionize meta-research practices in psychology and related fields.  By harnessing the power of Large Language Models (LLMs), we aim to overcome the limitations of manual data extraction and enhance overall research efficiency in meta-research processes.

Specifically, this project will:

Through a systematic evaluation of pipeline performance against established benchmarks, and their integration into state-of-the-art scientific workflows, we will help mitigate bias and enhance the scalability of meta-studies, thereby increasing the comprehensiveness and credibility of research findings. Finally, by engaging the community in building an ecosystem for Al integration in meta-research, we will ensure the long-term utility and sustainability of these efforts.

Partners: