AI in Research: The Dilemma Game

Should you submit a manuscript that is fully ideated and written by an LLM? What should you do if you get a seemingly AI-generated review of your paper? Dilemma Game is a discussion-based card game (and mobile app) encouraging honest dialogue about scientific integrity. Recently, our RAI team had the opportunity to work with the National Technical Library (NTK) team on a special edition dedicated to novel academic dilemmas posed by AI.

The Academic Services team at NTK organizes regular events for PhD students and other early-career researchers. One of their recurring endeavors is local facilitation of the Dilemma Game, a game developed at Erasmus University Rotterdam that provides academics with realistic dilemmas and uneasy choices. The game deliberately selects academic scenarios independent of players’ backgrounds, touching on topics ranging from publishing ethics to funding issues and team dynamics, with the aim of sparking a lively discussion in which participants try to persuade their colleagues of their truth.

Since the original game predates the recent boom of generative AI, we were happy to be approached, through reference of our founding member Lucie Stewart, by Eliška Skládalová from NTK with the ask to brainstorm a possible “AI update” of the game: creating new dilemmata focusing on issues with AI generation, authorship, or chatbot dependence. The collaboration resulted in ten brand-new scenarios and was ready for a test run at the event on January 14. Below is a small teaser for what the newly designed scenarios look like, learn more about the outcome in Eliška’s post.

Mixed with some scenarios from the original deck, the AI dilemmata prompted debates on the tradeoffs between textbook choices and practical complications. Participants differed in their interpretations of the scenarios and their ultimate decisions, as well as in their willingness to use AI tools in their own research. Still, it should be noted that participants of the Dilemma Game events are likely subject to certain selection bias (through their decision of attending an ethics-based game), and we, at the Responsible AI initiative, believe that a broader “survey” on people’s sentiment about generative AI usage in research would be beneficial, especially in the computer science circles which are often portrayed as highly techno-optimistic.