In short

Despite a century of predictions, AI won't replace teachers. Humans are biologically adapted to learn from other humans through a social 'dance' that drives attention and motivation, and schools build social skills the labour market demands. Schools also run on undocumented, tacit knowledge that can't easily be captured for automation. Following the Acemoglu–Autor task framework, AI will automate specific tasks while supporting human teachers, potentially helping address UNESCO's projected shortfall of 44 million teachers by 2030.

Predictions that machines will replace teachers are over a century old, Edison thought film would do it, and they keep being wrong. This essay explains why, and what AI will actually change.

The argument rests on three things automation struggles with. First, humans are biologically built to learn from other humans: the teacher-student relationship awakens attention and motivation in a way a screen doesn’t. Second, school is where we acquire the social skills the labour market increasingly rewards. Third, and most practically, the real work of schools lives in tacit, undocumented knowledge, the stuff in teachers’ heads that was never written down for an AI to learn from.

None of this means AI is irrelevant. Following the now-standard task framework, the essay argues AI will automate tasks, marking, planning, admin, and in doing so could help with a looming global shortage of millions of teachers. Augmentation, not replacement.

Key takeaways

  • Humans are evolutionarily wired to learn from humans; no machine can replace the social teacher-student dance that drives attention and motivation.
  • Education is fundamentally social; the labour market rewards social skills developed through human interaction.
  • Schools run on undocumented, tacit knowledge held in people's heads, making comprehensive automation impractical.
  • Tasks, not jobs, get automated (Acemoglu and Autor); teachers already use AI to save time.
  • AI can augment teachers, potentially helping meet UNESCO's projected need for 44 million more teachers by 2030.

Read the full piece

This is a summary. Read the complete essay, with all the sources and argument, on Substack.

Read on Substack →

Frequently asked questions

Will AI replace teachers?
No. Teaching depends on a biologically human social dynamic, builds social skills, and relies on undocumented tacit knowledge, so AI will automate specific tasks rather than replace teachers.
Why is teaching hard to automate?
Most of what matters in schools, lessons, interactions, body language, meetings, is never recorded and lives only in people's heads, making comprehensive automation impractical.
How will AI actually be used in education?
To automate discrete teaching tasks and support teachers and assistants, potentially helping address UNESCO's projected shortfall of 44 million teachers by 2030.

People & ideas in this piece

Stanislas DehaeneDavid DemingDaron AcemogluDavid AutorSal KhanB.F. SkinnerUNESCONatural pedagogyTacit knowledgeTask-based automation

Topics: AI, Work & the Economy