Modeling Information Problem Solving with Generative AI: A Framework for Higher Education

Description

In contemporary higher education, students are increasingly expected to manage and synthesize large volumes of online information. However, many struggle with effective digital information problem solving (IPS), a competence critical to academic success. Although generative AI tools offer new opportunities by streamlining information retrieval and content creation, they may undermine students' critical engagement with information. Existing IPS models have yet to fully address how AI transforms the IPS process. This theoretical gap limits our ability to guide students in developing IPS competence that balances effective AI use with maintaining control over the process. This study refined the IPS-I model by investigating how 14 novice and 12 expert teachers used generative AI during an authentic IPS task. Participants were instructed to use both generative AI and traditional internet sources to prepare a short presentation on healthy social media use. Data were collected via think-aloud protocols, screen recordings, and retrospective interviews. Data analysis combined deductive coding based on the IPS-I model with inductive coding for emerging AI-related behaviors. Results revealed that novices tended to over-rely on AI, bypassing critical processing and synthesizing phases, while experts maintain control by critically engaging with AI outputs and combining them with traditional internet sources. The study proposes the Digital IPS (DIPS) model, which decomposes the cognitive processes during IPS using traditional internet sources and generative AI. The DIPS model offers a theoretical framework for future research and the design of educational interventions for supporting AI-enhanced IPS.

Period	29 Aug 2025
Event title	EARLI 2025: Realising Potentials through Education: Shaping the Minds and Brains for the Future
Event type	Conference
Location	Graz, AustriaShow on map
Degree of Recognition	International