By: Irfan Ahmed, Dean at Sohar University, Oman
AI is no longer a futuristic add-on to education. It has quietly become infrastructural. From adaptive learning platforms and analytics dashboards to generative lesson-planning tools and automated feedback systems, AI now shapes how teaching happens on an everyday basis.
Much of the public and academic discussion has focused on what AI can do for students: personalize learning, close gaps, accelerate progress. Far less attention has been paid to what AI is doing to teachers, not in terms of job loss, but in terms of cognition, judgment, and professional growth.
Our recent qualitative study set out to explore this neglected question. Drawing on interviews with 23 high school teachers across five countries who regularly use AI-enabled instructional tools (the UAE, Egypt, Bahrain, the UK and USA), we asked a deceptively simple question: How does sustained engagement with AI reshape teachers’ professional thinking?
Teaching is cognitive work, not just task execution
Teaching has always been cognitively demanding. Long before AI entered classrooms, teachers were already engaged in continuous diagnostic reasoning: deciding what to teach next, interpreting student confusion, sequencing instruction, and adjusting explanations in real time. These are not administrative tasks; they are the cognitive heart of professional expertise.
Cognitive Load Theory (CLT), a framework typically used to study how students learn, offers a useful lens here. CLT distinguishes between:
- Intrinsic load (task complexity),
- Extraneous load (avoidable effort caused by poor design), and
- Germane load (the productive effort that builds understanding and expertise).
Traditionally, teachers have been seen as managers of students’ cognitive load. Our study extends this logic to teachers themselves. Our theory is that, when AI systems take over instructional decisions, they do not merely reduce workload for teachers, they redistribute cognitive responsibility to computers.
Efficiency gains, cognitive losses
Across platforms and contexts, teachers we interviewed described a consistent pattern: AI tools were undeniably efficient. Lesson plans appeared in seconds. Dashboards identified learning gaps instantly. Feedback was generated automatically. Many teachers spoke of relief, even gratitude.
“When I open the dashboard, I see every student’s gaps, but I no longer must find those gaps myself. The system shows me, and I just follow what it suggests.”
But alongside this relief came a quieter realization.
As AI systems increasingly handled planning, sequencing, diagnosis, and feedback, teachers found themselves thinking less deeply about those processes. Diagnostic reasoning became monitoring. Lesson design became template acceptance. Assessment became confirmation rather than interpretation.
From a CLT perspective, this is not just reduced workload. It is a shift of germane cognitive load away from teachers and toward algorithms. The very mental effort through which teachers refine judgment and build expertise is gradually displaced.
Importantly, this did not feel like sudden deskilling. Teachers did not “lose” competence overnight. Instead, they described a slow fading of cognitive challenge. One teacher captured it starkly:
“I don’t feel tired anymore — but I don’t feel like I’m growing either.”
When data starts to feel smarter than you
Another recurring theme was algorithmic authority, whereby teachers felt that algorithms were more likely to be right than them. Dashboards, performance indicators, and AI-generated recommendations carried an aura of objectivity. When teachers disagreed with the system, many described doubting themselves rather than the algorithm.
This matters because professional expertise depends on judgment under uncertainty. When AI presents learning as clean, visible, and optimized, it can subtly retrain expectations, away from ambiguity, experimentation, and error. In science classrooms, for example, teachers worried that perfect simulations replaced the messy failures through which both students and teachers learn.
AI did not remove teacher agency, but it made it conditional and effortful. Teachers who resisted automation did so deliberately: treating AI outputs as starting points for discussion rather than instructions to follow.
“AI (Century Tech) gives me the numbers, but not the story. I use the numbers to start conversations. I ask students why they think the system rated them that way. That’s where my teaching still lives.”
Yet these acts of resistance often ran against institutional cultures that prioritize speed, metrics, and visible data.
Ethics as a cognitive burden, not abstract policy
Discussions of AI ethics in education often remain abstract: transparency, bias, governance. Our findings suggest these issues are experienced far more intimately.
Teachers described the strain of justifying decisions they did not fully understand, defending algorithmic outcomes to students and administrators, and working within systems whose logic remained opaque.
“It (Brisk) gives me comments like ‘good use of evidence’ or ‘expand your argument.’ The problem is, I don’t always know why it wrote that, and students think the comment is mine.”
These pressures were not philosophical, they were cognitive and emotional loads layered onto everyday teaching.
In this sense, ethical concerns are not external to pedagogy. They directly shape the mental conditions under which teachers work.
AI as infrastructure, not a tool
One of the most important insights from the study is that AI should not be understood as a discrete classroom tool. It functions as infrastructure, quietly redefining what counts as legitimate work, valuable expertise, and professional success.
Across very different platforms, the same pattern emerged: automation of core pedagogical judgment led to similar cognitive consequences. This suggests the issue is not poor implementation or individual tools, but a deeper shift in how teaching itself is being reorganized.
If efficiency becomes the dominant logic, the profession risks losing the intellectual struggle that makes teaching a form of expert practice rather than procedural management.
Rethinking AI integration around teachers, not just learners
The question, then, is not whether AI should be used in education. It already is. The more urgent question is what kind of teachers AI is helping to produce.
If AI removes extraneous burden while preserving teachers’ engagement with uncertainty, interpretation, and judgment, it can support professional growth. But if it systematically offloads those cognitively generative practices, efficiency gains may come at the cost of expertise.
Reframing AI integration through teacher cognition, and extending frameworks like CLT beyond students, offers a way to see these trade-offs more clearly.
In the end, the challenge is not technological. It is pedagogical and professional. The future of AI in education will depend less on what algorithms can do, and more on whether educational systems still make room for teachers to think.
