By Lea Laursen Pasgaard, AAU Communication and Public Affairs. Translated by LeeAnn Iovanni, AAU Communication and Public Affairs.-

On the same day the article was published, it was mentioned in a US congressional hearing as an example of the caveats policymakers need to observe when using artificial intelligence for decision support. Thus, it is also an example of how research can have an impact, both as a scientific contribution and as input to the political agenda, says David Budtz Pedersen, Professor in the Department of Communication and Psychology at Aalborg University (AAU).

Calls for caution

With the rise of artificial intelligence, decision makers and governments around the world have access to a powerful new instrument that can gather knowledge and evidence that then serves as the knowledge base for policymaking. But when the knowledge base is rooted in algorithms and artificial intelligence, it also calls for great caution. Can democratic decision makers ascertain the quality of the information that artificial intelligence collects? How do we identify gaps in information composed by large 'generative language models'? 

These are some of the questions that David Budtz Pedersen raises in a new article just published in the renowned scientific journal Nature. He wrote the article in collaboration with colleagues from Cambridge University, University College London, Georgetown University and the European Commission.

Compilation of cross-cutting recommendations and observations

- A unique aspect of the article is that representatives of the European Commission and a large AI-powered business are among the authors. The article itself is thus an expression of the interdisciplinary interest and expertise that has arisen around artificial intelligence, the professor says. 

The research underpinning the article draws on the project Algorithms, Data and Democracy (ADD) which is funded by VELUX FONDEN and VILLUM FONDEN. Like the article, the project aims to generate debate about the use of artificial intelligence in democratic institutions. The research is relevant for researchers, decision makers, and knowledge persons in the public sector.

The idea for the article arose when the authors were invited earlier this year to contribute to a seminar in Washington D.C. organized by the American Association for the Advancement of Science. In the months following the seminar, David Budtz Pedersen and his colleagues decided to write their recommendations and observations for the article now published in Nature.

Read the article here