What "Human-Centred AI in Education" Actually Means (Beyond the Buzzword)
EDUtech Asia 2026 has a theme: "Human-centred education, powered by AI and tech" (EDUtech Asia, 2026). It's a good phrase. It's also everywhere: in conference programmes, vendor marketing decks, Ministry of Education policy documents, and institutional strategy statements across Southeast Asia.
The problem with phrases that appear everywhere is that they can mean everything and therefore mean nothing. "Human-centred AI" has become a kind of rhetorical shorthand, something organisations attach to their AI initiatives to signal good intentions, without necessarily committing to anything specific.
So what does it actually mean? And how do you tell the difference between AI in education that genuinely centres humans (learners, teachers, and communities) and AI that uses the phrase to make a product decision look like a values decision?
Three things it actually means
1. The educator's logic drives the AI — not the other way around
One version of AI in education is AI that decides what to teach, when to teach it, and how to sequence learning. The system determines the learning path based on what it calculates will be most efficient. The educator's role becomes supervisory at best.
That's not human-centred. It's AI-centred education with a human in the room.
Human-centred AI in education looks different: the educator defines the learning outcomes, the content, the sequence, and the pedagogical approach. The AI delivers within that structure, more consistently, more accessibly, at greater scale, but the teaching logic belongs to the teacher.
Noodle Factory describes their approach this way: "We didn't retrofit teaching onto AI. We built AI around teaching." That distinction matters. When educators upload their course content and define their outcomes, and the AI operates within those parameters, the human teaching logic is what's being scaled, not replaced. That's the kind of specificity that separates a genuine commitment from a talking point.
2. Learner agency is preserved, not engineered away
There's a version of "personalised learning" that is actually just optimised compliance — the system figures out the fastest path to a correct answer and routes each student down it. The student gets there, but may not develop the capacity to navigate ambiguity, evaluate conflicting information, or reason under uncertainty.
Human-centred learning systems are designed to build learner agency, not just improve learner outcomes on measurable metrics. That means leaving room for the student to struggle productively, to make choices, to encounter friction, because that's where durable learning happens.
The practical question to ask of any AI learning system: does it develop learners' ability to think independently, or does it make learners more dependent on AI to think for them?
3. Equitable access is a design requirement, not an afterthought
AI tools that require stable high-speed internet, expensive devices, or high levels of digital literacy to use effectively are not human-centred in the context of Southeast Asian education, where a significant portion of learners study in lower-resourced settings, on mobile devices, with inconsistent connectivity.
Human-centred AI in this region has to grapple with access seriously. That means designing for mobile-first, for offline or low-bandwidth scenarios, and for learners who are not already confident digital users. Vendors who describe their tools as human-centred without addressing these constraints are using the phrase to describe their values, not their product decisions.
How to evaluate the phrase when you encounter it
When a platform, policy, or programme describes itself as "human-centred," here are the questions worth asking:
Who defines the learning? If it's primarily the AI, be sceptical. If it's primarily the educator, with the AI as the delivery mechanism, that's a meaningful difference.
What data is collected, and who benefits from it? Student behavioural data collected to improve a vendor's model is a different thing from student progress data surfaced to help a teacher support a struggling learner. Both can coexist in the same platform. Which is primary?
Does it increase or decrease educator autonomy? Tools that expand what educators can do, giving them back time, extending their reach, and amplifying their judgment, are different from tools that standardise and constrain what educators do. The former is human-centred. The latter may be efficient, but it's not the same thing.
What happens when the AI is wrong? In a genuinely human-centred system, there is a human in the loop with the authority and the information to catch and correct errors. A system that routes students through AI-determined pathways with minimal educator visibility isn't centred on anything human. It's just automated.
The phrase is worth fighting for
None of this is an argument against AI in education. Quite the opposite. The most compelling case for AI in education is precisely that it can be human-centred: freeing educators from repetitive tasks, extending high-quality learning to more students, and surface the information teachers need to make better decisions.
But that only happens if the design choices behind the technology are genuinely oriented toward learners and teachers, not just efficient outcomes on measurable metrics.
"Human-centred AI in education" is a useful phrase when it describes something real. The job for educators, school leaders, and edtech buyers is to push past the phrase and into the specifics: what it means in practice for this platform, this policy, this classroom.
The answer will tell you a lot.
Sources
Noodle Factory builds AI tutoring tools for universities, polytechnics, and K-12 schools across Southeast Asia. Learn more at noodlefactory.ai.
