44 million teachers?

At a conference the other week there was a presentation by the Otermans Institute, who have this slogan:

Democratising Learning through AI Technology

They are building AI agents which can act as personal tutors. The idea is that these can be rolled out in places where there is a shortage of teachers, of which the presenter told us, there was currently a global shortage of 44 million.

We will come back to the question of whether this counts as 'democratising' - at first blush that seems a rather strange use of the word since democracy is not about having employability skills (for that is what Otermans are actually teaching) but having a say in the process of government under which you live. Tyrants also want their subjects to have good employability skills, after all.

Let's instead think about that number: 44 million. There is not a shortage of 44 million people, just 44 million teachers. What is a teacher? A teacher is a person who is willing and able to teach. Note the two conditions. So the shortage means we lack 44 million people who:

  1. Have the skills to teach
  2. Want to work as teachers

So one solution would be to provide a lot more training and incentivise 44 million people to take the training and work as teachers: by offering them good jobs (reasonable pay, good working conditions etc.).

That would be very expensive. But then so is AI.

We don't know the true cost of AI because those funding it like to keep that hidden, but the levels of investment would go a long way towards paying for those 44 million teachers. And we mustn't forget that AI companies get away with externalising all the environmental and social costs of their business. Add those in (after all, training teachers, unlike training AI, has very little environmental cost and a great deal of social benefit) and we could probably pay for all those 44 million teachers if we would only #JustStopAI.

So why are 'we' - as in the people who do it and the people who implicitly consent to it happening by going 'Wow! How cool! And what a great use of technology!' - developing AI teachers rather than real teachers?

The answer is that the issue is not about how much it costs but who pays and who profits.

If we are to train and employ 44 million more teachers globally, that will have to be done predominantly by governments and paid for by taxation. In some countries that will be wealth tax or other progressive taxation schemes, in others - like the sorry little island I live on - it is likely to be more regressive forms of taxation. But either way, that tax is predominantly spent on salaries for teachers and teachers of teachers. It is not spent on buying a service from a company which makes a profit for its owners.

However, if we train AI to do the work of 44 million teachers, then the companies that own that AI and have paid to develop it will sell it to the same governments. The governments will still have to raise taxes to pay for it, but now that tax will not be spent on salaries for their citizens (some of which is recycled back in taxation) but on contracts to supply services from global corporations looking for significant return on their investment: profit.

Which brings me back to democratising learning.

Setting aside my earlier cavils for a minute, which feels intuitively more 'democratic' to you?

  • Raising taxes to pay employ people in your own country to do work
  • Raising taxes to buy services from profit-seeking global corporations

I hope the answer is obvious. Even if those profit-seeking organisations hide their true motives by setting up not-for-profit organisations (or paying universities) to do the work.

And if we do care about real democracy, not just a skilled workforce but giving political power to citizens, then we should be terrified by AI teaching agents.

Real human teachers can care about truth and justice, can be role-models of democratic citizens who engage in political processes and challenge misuse of power, can challenge assumptions and prejudices, inspire originality, creativity and freedom. They can make good people by being good people. AI teaching agents might be able to do some of that, though I doubt it, but they can be trained to squash all thoughts of empowerment and emancipation in their pupils, and don't have pesky lives outside the classroom which might challenge conformism. While we create incentive structures around teaching which are pretty successful in selecting people for obedience (Chomsky), that is not a perfect mechanism. Teaching institutions always have a few rebels. AI solves that problem for the people who want education to be a training in conformity to consumerist - or other - ideals.

Let's be honest, the Otermans Institute and others engaged in this same AI Assisted Learning project have no right at all to be using the word 'democratic'. They are engaged in a project which will reinforce the power of capital over labour, reinforce inequality within and across societies by maintaining rent-seeking economic structures, and give an incredibly powerful tool of thought control to would be tyrants.


You'll only receive email when they publish something new.

More from Tom Stoneham
All posts