Towards Synergistic Teacher-AI Interactions with Generative Artificial Intelligence

Synthesis: Drawing on a systematic literature review, this UCL chapter proposes a five-level framework of teacher-AI teaming—transactional, situational, operational, praxical, and synergistic—to capture how GenAI interactions may replace, complement, or augment teacher competence. The framework moves beyond task division toward collaborative decision-making where teachers and AI engage in negotiation, constructive challenge, and co-reasoning.

Five Levels of Teacher-AI Teaming

The framework defines a progression from minimal to maximal teacher-AI collaboration:

1. Transactional — AI executes a concrete task on request with no shared context. Teacher sends prompt, AI returns output. Example: generating a quiz from a topic list. Risk: cognitive offloading without professional growth.
2. Situational — AI gathers data from the teaching/learning context and reports it back. A shared state exists for the task. Example: AI summarizes classroom discussion data. Teacher remains the primary decision-maker.
3. Operational — Teacher sets goals; AI not only reports context but also proposes actions to achieve those goals. AI moves beyond task execution into action planning. Example: AI recommends pedagogical interventions based on student performance data.
4. Praxical — AI incorporates learning patterns from prior interactions to adapt its behavior over time. The system learns from teacher feedback and refines its models. Teacher and AI develop shared practice.
5. Synergistic — Both agents engage in negotiation, constructive challenge, and co-reasoning that enhance each other’s capabilities. Outcomes emerge that neither could realize independently. This is the highest level of hybrid intelligence.

Key Insights

• Beyond task automation: Most current GenAI use in education operates at transactional or situational levels. Moving toward operational, praxical, and synergistic teaming requires deliberate design of the interaction paradigm, not just better models.
• Teacher agency is central: The framework argues that GenAI integration should enhance—not diminish—teacher agency. Higher teaming levels require teachers to actively engage in decision-making rather than passively accept AI outputs.
• Cognitive atrophy risk: Fully offloading teaching tasks to AI at lower teaming levels risks diminishing teachers' cognitive engagement and professional growth (deprofessionalization).
• Socio-technical factors matter: Beyond technological design, organizational culture, professional development, ethical guidelines, and institutional support shape whether teacher-AI interactions reach synergistic levels.
• Hybrid intelligence: The vision is teachers and AI as complementary agents—each bringing distinct capabilities—rather than AI as a replacement.

replacement.

What this means for practice

• Instructors. Place your current GenAI use on the five levels before changing tools — most existing use sits at transactional or situational teaming — then decide deliberately what would move a task upward, since higher levels require redesign of the interaction, not just a better model.
• Instructors. Keep professional judgment in the loop at the lower levels: the framework positions the teacher as the primary decision-maker in transactional and situational teaming and warns that full task offloading risks cognitive atrophy and deprofessionalization.
• Faculty developers. Align professional development with an AI literacy framework (the chapter points to UNESCO's AI competency framework for teachers) so teachers build the metacognitive and critical skills that praxical and synergistic teaming demand.
• Administrators. Treat implementation as staged capacity building: audit existing teacher–AI interactions to find the baseline, then upgrade transactional tools toward situational or operational teaming by connecting classroom data streams and goal-setting interfaces.
• Designers. Co-design prompts, interaction protocols, and decision-support flows with teachers, and run evaluation cycles on process analytics and teacher reflection rather than deployment metrics alone.

Limitations

• This is a conceptual framework paper with no empirical test of its own: the five levels are proposed from literature and illustrated by a review, and the authors state the field needs future empirical investigation.
• The supporting review searched a single database (Web of Science, 2010–2025); more than 7,000 records were screened by title and abstract by two coders (Cohen's K = 0.82), and data-extraction agreement was lower (Krippendorff's alpha = 0.69).
• Full extraction yielded 103 studies, and GenAI-powered tools were only 39% of them, so claims about the higher teaming levels rest on a minority of the evidence base.
• The review records which levels appear in the literature; it does not test whether reaching a higher level improves teaching quality or student outcomes.

Citation

Cukurova, M., Suraworachet, W., Zhou, Q., & Bulathwela, S. (2025). Towards Synergistic Teacher-AI Interactions with Generative Artificial Intelligence. arXiv preprint.