Maike: a Socratic Chatbot

The challenge

Students are increasingly exposed to AI tools that offer instant answers and polished outputs, encouraging passive consumption rather than active engagement. Such dependency can lead to cognitive offloading, reduced agency, compromised academic integrity, diminished self-efficacy, and normalized surveillance and privacy violations. If we aim to cultivate critical thinkers, we must critically evaluate the tools we use and how we use them.

Our position: scaffold, do not substitute

Before building anything, we take a position: AI in education should be judged by whether it develops human reasoning rather than replaces it. We set this out as a guiding principle for human-centered AI in a position paper at the ACM AI Leadership Summit 2026, From Substitution to Scaffolding: Breaking the Self-Reinforcing Harm Cycle of AI in Education (and Beyond). Two companion papers develop the argument: Do AI Tutors Empower or Enslave Learners?, the opening talk of the GenAIHEd workshop at AIED 2025, and AI in Education Beyond Learning Outcomes: Cognition, Agency, Emotion, and Ethics, which proposes evaluating educational AI on four dimensions rather than on test scores alone.

Our response: Maike, a Socratic chatbot

In response to these challenges, we are developing Maike, a Socratic chatbot designed to support, not supplant, student thinking. Unlike traditional chatbots that provide quick answers, Maike is guided by pedagogical principles that emphasize inquiry over instruction, dialogue over delivery. Drawing from the tradition of Socratic questioning, Maike prompts learners to reflect, articulate, and revise their reasoning through meaningful conversation.

The chatbot is designed to foster critical thinking, self-regulated learning, and a more profound sense of agency. By asking the right questions rather than offering immediate solutions, Maike encourages students to engage deeply with complex ideas, enhancing both their cognitive and metacognitive skills. Our goal is to design an AI-powered learning companion that empowers rather than diminishes the learner.

Maike is open source and runs on small language models, which keeps it inexpensive, transparent, and deployable in a real school rather than only in a lab. Under the pedagogy sits a concrete mechanism: Maike reads a student's argumentative essay through argument mining, identifying and assessing the claims and reasons it contains, and uses that analysis to generate critical questions adapted to what the student actually wrote.

The classroom study

We have now run an ethics-approved classroom study with secondary-school students enrolled in the International Baccalaureate Theory of Knowledge course (a curriculum known for its emphasis on critical thinking and epistemological inquiry).

• Objective: to examine how students interact with Maike, and what that interaction does to their written argumentation and critical thinking, compared with a conventional chatbot.
• Design: between-subjects, with around 60 students working with either Maike or a conventional chatbot while revising a Theory of Knowledge essay.
• Measures: pre/post argumentative essays, and the full dialogue logs of every conversation, for interaction analysis.
• Timeline: conducted in June and July 2026. Analysis is ongoing.

Looking ahead

Maike represents an interdisciplinary effort to align educational AI with core pedagogical values. Rather than letting technology shape learning in unchecked ways, we propose a reflective and transparent approach; one that prioritizes the development of independent thinkers in a digital age. If we are to trust AI in the classroom, it must first earn that trust by supporting the learner's growth, not replacing it.

The study is one step in a longer program: empirical research into how AI systems affect human cognition, agency, emotion and ethics, at classroom scale and with real participants, rather than benchmark scores alone.

Our scientific publications