Visionologies
EDUCATION•4 min read

The Student Already Has GPT. The Question Is What School Is Now For.

The framing "AI will change education" misses the speed. AI changed it. A first-year university student in 2026 has casual access to a tutor that explains derivatives at any level, drafts essays from rough notes, and runs Socratic dialogue on demand. The institution's traditional value propositions — knowledge transfer, scaffolded practice, evaluation — are now partially commoditised by a tool that fits in a phone.

Three institutional responses are emerging, at very different speeds.

The first is assessment that is robust to AI use: oral defences, project-based work, open-book exams with novel data. This is right but slow — accreditation and faculty workload structures resist it. The second is explicit AI-fluency curriculum (prompt design, model evaluation, agentic workflows). Politécnico do Porto, ISEG and a few others have early modules; most institutions have not started. The third is a renewed argument for in-person, cohort-based, mentorship-heavy learning where the value is the relationship and the project, not the lecture content. This is where some of the best programmes will end up — and probably the most expensive.

Teachers' jobs are not disappearing, but the skill mix is shifting: less content delivery, more facilitation, more evaluation design, more personalised feedback at scale (which AI itself helps deliver). The risk is widening inequality. Students at well-resourced schools learn to use AI as a thinking partner, asking it to challenge their reasoning. Students at under-resourced schools learn to use it as a homework substitute, copying its outputs without engagement. The educational system that does not notice this distinction will fail its students twice — once on the basics, and once on the shift.

arrow_back All insights