Rethinking the Human in the Loop

Why testing the “we’ll always need teachers” claim helps us design better learning

The conference theme “Humanity in the Intelligent Age” echoes the common refrain these days that there must always be a human in the loop. In education, that becomes “we’ll always need teachers.” It’s a comforting claim. But the real value isn’t in defending it; it’s in testing it, critically, collaboratively, and in public in order that we design better systems whatever the future holds.

This isn’t apocalypse; it’s design discipline

Acting as our own Devil’s Advocate is a standard scientific method: trying to break your favourite hypothesis. If it survives, it’s stronger; if not, you’ve learned where to improve. We can apply that spirit to the teacher question, not to dismiss teachers, but to clarify which human roles truly add irreplaceable value, and where we might safely reduce reliance on traditional teaching roles.

And no, this is not an economic inevitability sermon. It’s pragmatic. Systems are under fiscal pressure; leaders expect more learning for the same or fewer resources. That doesn’t have to mean “fewer teachers.” It can mean teachers focusing on fewer, higher-impact tasks, perhaps the parts where human judgement, care and culture matter most.

Name the human value—so we can protect it

Even in a “teacher-light” system, humans are not necessarily absent. They may shape identity, community and values; read context and culture; celebrate achievement with genuine pride; spot when a learner needs encouragement rather than another quiz. They almost definitely set purpose, guardrails and governance.

Compare to the typical, not the ideal

We often compare technology to an idealised classroom with an expert teacher, a small class, and abundant time. Most learners experience something more constrained. A fairer question might be: could an AI-supported model deliver learning that matches, or sometimes exceeds the typical classroom, rather than the ideal, while expanding access and affordability?

A thought experiment: minimising direct teacher involvement

What might a credible “teacher-light” pathway look like at scale?

• AI as diagnostic guide: continuously maps each learner’s prior knowledge, misconceptions and progress; recommends next steps.
• Core knowledge & recall: high-quality media and readings, with well-designed checks and spaced practice for durable memory.
• Deeper understanding: structured AI dialogue that probes reasoning, asks for examples and counter-examples, and surfaces gaps.
• Collaborative learning: AI-facilitated peer discussions, with transparent rubrics and analytics to evidence individual understanding.
• Human oversight where it counts: mentors and moderators for wellbeing, values, culture, integrity, and authentic celebration of progress.

This may not capture the full richness of human-led teaching, but it could be good enough (and in some contexts better) for many learners, especially those currently excluded by cost, geography or capacity constraints.

Let’s not “virtue-signal”—let’s build

We don’t need to proclaim that we care about ethics, belonging and teacher professionalism; we can just design learning to maximise participation and impact even if that means dealing with limited resources. Institutions and companies will always care about competitiveness and financial sustainability. Fortunately, building trust, fairness and learner wellbeing is not charity. It’s also smart and can lead to better engagement, better outcomes, better reputation.

Pace matters: transitions are gradual

None of this happens overnight. Technology shifts create inertia and adaptation, not instant replacement. Remember the anxiety about self-driving trucks and how long that has taken to materialise at scale. Labour markets adjust; people move into roles with higher demand (we can’t get enough nurses, doctors, plumbers, carpenters). Expect evolution, not a cliff edge.

Why this matters for Berlin

This is not armchair speculation. It mirrors what educators will do at OEB: design, test and imagine AI-supported learning structures together. The spirit matches OEB’s “EdHack: A Participatory Design Sprint for Advancing Educator AI Literacy”, not replacing human intelligence, but enhancing it through critical, collaborative re-imagining.

Our workshop takes the same stance: we’ll try, in good faith, to design a credible learning experience that minimises direct teacher involvement. If it breaks, we’ll learn where and why the human role is essential. If it holds, we’ll have a blueprint for expanding access while reserving teachers for the highest-impact work.

An open invitation

If you’re open-minded, and willing to critique your own assumptions, then join us. Perhaps a teacher-light pathway won’t be what you prefer. It may still be useful: more accessible, more affordable, and, in some contexts, better than today’s reality. We may not be able to stop these shifts but perhaps we can shape them. Let’s do that work together, thoughtfully, humanely, and with design rigour.

Brian Mulligan, previously Head of Online Learning Innovation at Atlantic Technological University, Ireland is now a consultant with Universal Learning Systems.

Paul Bacsich is a UK-based consultant in online learning and e-learning with over 30 years’ experience, known for his pioneering work on virtual universities, open educational resources, and benchmarking digital education in both higher education and schools sectors.

Paul and Brian will be facilitating a workshop at OEB 2025 titled: “Higher Education Without Teachers? Re-imagining Learning in an AI-Supported, Low-Cost Future”


Written for OEB 2025 by Paul Bacsich and Brian Mulligan.


Join Paul and Brian for their Learning Café at OEB25.