Osnabrück University researchers develop AI-supported automated documentation of experimental designs in Osnabrück, Germany; reduces documentation workload and boosts open science

Osnabrück University researchers develop AI-supported automated documentation of experimental designs in Osnabrück, Germany; reduces documentation workload and boosts open science

Better research thanks to “SweetPea”: How AI makes scientific experiments more transparent: Uni Osnabrück

Better research thanks to “SweetPea”: How AI makes scientific experiments more transparent

With “SweetPea”, Osnabrück University is working on new AI-supported approaches to make scientific experiments easier to plan, understand and reproduce.

How can scientific experiments be made more transparent, efficient and reproducible? This question is the focus of the international research project “SweetPea: Automating the Implementation and Documentation of Unbiased Experimental Designs”, in which Osnabrück University is participating. The project is funded by the US National Science Foundation (NSF).Junior Professor Sebastian Musslick from the Institute of Cognitive Science at Osnabrück University will implement the project together with researchers from Princeton University, the University of Utah and Brown University.Experimental designs form the basis of empirical research. In many scientific disciplines, variables are manipulated in order to investigate cause-and-effect relationships. “However, the increasing complexity of modern experiments also increases the difficulty of designing experimental procedures in such a way that unintentional bias or so-called confounding is avoided,” explains Musslick. Experimental sequences are often still created manually or simply put together randomly - procedures that are either time-consuming or could promote hidden biases.

The open source system “SweetPea” aims to solve these challenges through the automated generation, validation and documentation of experimental designs. The aim of the project is to provide researchers with tools that make experiments more efficient, transparent and reproducible. The methods developed are to be used in particular in large-scale behavioral experiments and, in the future, also in the development and evaluation of AI systems.

Osnabrück University’s contribution focuses on AI-supported methods for the automated documentation of experimental designs. This should reduce the workload for researchers and facilitate the comprehensible documentation of experiments in the spirit of the open science movement.

“The documentation of complex experiments represents a considerable additional effort for many researchers. Our aim is to largely automate this process and thus facilitate open and reproducible research,” explains Sebastian Musslick.The project ties in with the “AI for Science” initiative of the"Data Science"Research Center. This involves developing AI methods that can support and partially automate scientific processes - from planning experiments to generating knowledge.

Further information for the media:Prof. Dr. Sebastian Musslick, Osnabrück UniversityInstitute of Cognitive Science