How AI agents can actively scaffold critical thinking during paper reading

Characterize how AI agents, specifically large language model-enabled conversational agents, can actively scaffold users’ critical thinking throughout the process of academic paper reading, including the forms of guidance and interaction that effectively support higher-order thinking.

Background

While AI tools can aid reading efficiency and provide multiple perspectives, they also raise concerns about cognitive offloading and reduced independent thinking. The paper notes that, despite growing interest in AI-supported critical reading, the mechanisms by which AI agents should interact in situ to actively support critical thinking during paper reading are not well understood, motivating a need to determine effective scaffolding strategies.