Michael Rowe

Trying to get better at getting better

What if the endpoint for teachers is to serve AI?

The posts I read about AI in education tend to fall into two main categories:

  1. AI will never replace teachers because teaching is about connecting with learners at a human level, and AI will never be human.
  2. AI will replace some of the tasks that teachers are currently responsible for, freeing them up to do more of the work they really want to do (typically, spending more time with learners).

I don’t find either of these arguments compelling but will only respond to the second one in this post.

We have a long history of incorrect assumptions that technology will free us from the mundane tasks that no-one wants to do anyway. An obvious example are the dishwashers and washing machines that were meant to remove the drudgery of housework. But I don’t know of anyone sitting around enjoying all this free time. This is the one of the paradoxes of technology; it’s supposed to save us time but we’re busier than ever.

We assume that the time saved by AI assistants will remain free, but I think that, as teachers have more tasks taken over by AI, they’ll be given more tasks to replace them. What kinds of tasks, I hear you ask? Well, what if we have to use our new free time to improve the algorithms that have replaced many of the tasks that we used to do? Maybe the role of the teacher will be to train the AI so that it can be a better teacher. Maybe the ultimate role of the teacher will be to serve the AI, which in turn will serve the student. And maybe that’s not a bad thing.

Also, we won’t need many teachers to train the AI (because it’s distributed). We’ll only need the best ones. And as AI gets better at teaching, fewer and fewer humans will meet this requirement.

Oh, and by the way, the same argument works for clinicians. Anyone saying that AI will only be used to augment and support clinicians rather than replace them hasn’t followed the argument through to its logical conclusion. Instead of spending all that free time to connect with patients, clinicians might have to spend it doing data analysis and refining the algorithms providing the real care to patients.

I suspect that many will find this conclusion unpalatable, but that’s no reason to disregard it. If you can see a different outcome, please let me know.


Share this


Discover more from Michael Rowe

Subscribe to get the latest posts to your email.