Placing AI as a partner in argument rather than a tutor — comparing individual-AI, group-AI and human-only configurations

1Moving where AI sits

Research on AI in education has concentrated on the roles of tutor, recommender and evaluator. Each places AI above the learner, guiding. What this project tries is moving that position, placing AI as a participant in the argument. How learners and AI co-construct knowledge through dialogue, the description notes, has scarcely been studied.

2Comparing three configurations

Setting the three side by side makes it possible to measure what adding AI changes against not adding it. What the project measures is not only learning outcomes: interaction patterns, students intellectual autonomy and trust in the AI are also objects of study. If understanding improves but the capacity to think independently does not develop, the result means something different.

3The constraint of a large class

The reasons collaborative argumentation is hard to run are given as unequal participation, diverse goals, prior knowledge gaps and lack of coordination — all worsening as numbers rise. The underlying constraint is that an instructor cannot circulate through every group discussion. The hope is that if generative AI can monitor and offer support just in time, the format can hold even at scale.

4Leaving design principles behind

The project uses a design-based research approach, iteratively developing two generative-AI-supported argumentation modules, piloting them in lab sections of an introductory biology course, and conducting field studies comparing configurations.

The intended outcome is not improvement of one course but generalizable design principles for future AI-augmented learning systems, along with a novel framework integrating distributed cognition, epistemic cognition and evolutionary thinking. Amounts are the obligated amount as of the check date and may change.