Review of Kamalov, F., Santandreu Calonge, D., & Gurrib, I. (2023). New Era of Artificial Intelligence in Education: Towards a Sustainable Multifaceted Revolution.

In an age where the pedagogical pendulum swings ever more decisively toward automation and AI-augmented instruction, the work of Kamalov et al., New Era of Artificial Intelligence in Education: Towards a Sustainable Multifaceted Revolution, arrives as a meticulously scaffolded provocation. This is not a paper merely about tools or platforms—it is a meditation on an epochal shift, one where artificial intelligence is recasting the very conditions under which learning unfolds.

If there is a central animating concern here, it is not technological novelty, but human adaptability. As the authors observe, “AI is revolutionizing the educational landscape, bridging gaps, and encouraging a more inclusive and effective learning environment”. Yet such transformations are not merely procedural. They compel a re-evaluation of what it means to teach—and to know—in a classroom increasingly populated by non-human agents.

The document’s structure is resolutely tripartite: applications, benefits, and challenges. Within this framework, four thematic pillars emerge—personalized learning, intelligent tutoring systems, assessment automation, and teacher–student collaboration. These are not simply administrative conveniences. They are expressions of a deeper epistemological realignment, wherein knowledge is no longer dispensed uniformly but instead choreographed algorithmically, tailored in real-time to the learner’s evolving cognitive profile.

In particular, the notion of personalisation—so often invoked with evangelical fervour—acquires new complexity here. The authors articulate a vision of “adaptive learning systems driven by AI” that can “analyse students’ performance, strengths, and shortcomings in order to offer tailored learning courses”. Yet this is not merely a function of student data. It is also a question of task complexity, of how AI might respond not only to who a learner is, but to what a learner is doing, in a particular moment, within a specific cognitive terrain.

This insight finds poignant resonance when set against the pedagogical narrative advanced in the companion piece, “Not All Help Helps”. There, we meet Yusuf and Ava—two students, one expert, one novice—grappling with the same task but reacting very differently to instructional support. “Yusuf rolls his eyes,” we are told, as scaffolding becomes interference rather than aid. “It’s holding him back.” This is the Expertise Reversal Effect (ERE): a phenomenon where instructional guidance, once helpful, becomes obstructive as the learner’s competence increases.

The juxtaposition of Kamalov’s macro-analytic lens with the micro-classroom dynamics explored in the ERE paper creates a striking dialectic. One text moves with sweeping ambition, the other with granular precision. Yet they converge on a critical point: support must be contextually responsive, not blindly generous. As Kamalov et al. affirm, “While the applications and benefits of AI in education can paint an alluring picture, it is important to be aware of potential hazards”. Chief among these hazards, we might now say, is well-intended overreach—whether human or artificial.

This confluence of ideas compels a rethinking of AI not as an omniscient tutor but as a co-navigator in the learning journey. In practice, this means designing intelligent systems not merely to detect gaps in knowledge, but to calibrate when and how much to intervene. The ERE literature reminds us that instructional timing is as critical as content. As Ava and Yusuf show, “adaptive teaching isn’t just about who the student is—it’s about the interaction between the student and the task at this moment in time”.

Indeed, one might say that both works are, at their core, arguments for calibrated responsiveness. Kamalov’s paper speaks of AI’s ability to “provide real-time analytics and insights… allowing educators to adjust their teaching strategies accordingly”. But such insights must be modulated with pedagogical tact—what Not All Help Helps aptly terms “the art of just enough.”

What emerges, then, is not a celebration of AI as a pedagogical panacea, but a plea for discernment. There is, as Kamalov’s authors caution, a tension “between the benefits of AI and its potential dangers”—bias, inequity, depersonalisation—lurking at the edges of algorithmic decision-making. To navigate this, we require systems that do not merely process data, but understand its pedagogical valence.

Both documents gesture toward a new model of teaching: one where the expertise of the educator lies not in providing constant answers, but in designing environments where the right help is available at the right time. Sometimes that help is a prompt. Sometimes it is silence. And sometimes, it is knowing when to step aside and let the learner wrestle with uncertainty.

In the final reckoning, this is not about AI in education, but about education with AI—education that recognises intelligence, whether artificial or human, not as fixed or hierarchical, but as deeply situated, contingent, and responsive. It is an education that begins, as all good teaching does, with a question: what does this learner need, right now? And—crucially—is wise enough to know that sometimes, the answer is less.