Alright, fine, I’ll admit it: the thought of the AI super tutor has kept me up some nights over the last few years. My dreams have been haunted by that sinister, feathered monster Duo from Duolingo. I’ve even wondered what I would retrain in; my wife may well disdain my practical skills at home, but plumbing remains the most distinct fantasy.
So perhaps you’ll forgive us just a little schadenfreude, when we see that Sal Khan – who had evangelised the AI future of education in a 2023 Ted Talk – has recently described his company’s own AI tutor, Khanmingo, as a “non-event” for students. I’ve even seen some well informed, articulate commentators online suggesting this is part of wider trend of the failure of ed-tech to fulfil its AI promises.
Whilst I long to write something really polemical (to garner the attention that this newsletter makes clear that I crave), I am afraid that AI’s role in education broadly and tutoring specifically will be something more balanced.
False promises and marketing spiel
I think part of the problem – and the reason for the antipathy from a lot of educators – is the marketing spiel from these ed-tech pioneers, who try and seduce parents with glossy phrases like ‘mastery based learning’ and ‘personalised learning pathways’.
Mastery based learning simply means that a student must score a certain percentage on a topic before advancing to the next topic. Tutoring already does this to a point (the limit here is that you are sometimes best leaving a topic for short while if, after multiple sessions, the student is still struggling, as it might be bad for morale).
Personalised learning pathways means that your next lesson is based around weaknesses you showed in the last session. Again, tutoring does this already and on a deeper level, as tutors can glean whether a mistake is conceptual, down to poor technique or down to lack of knowledge.
Furthermore, this sort of tech has existed for some time now. My father, when he wanted to learn Russian, collaborated with that aforementioned feathered creep Duolingo. His language learning was personalised and I can tell you that … он ни слова не говорит !
And if you can’t understand that then your Russian is no worse than his.
How AI can help tutoring
Already, we are seeing enormous benefits from AI in education. There are three main elements that spring to mind: ease and speed with creating lesson plans, speed at creating bespoke revision resources for students (though these must be checked again and again for accuracy) and automated marking.
The less time that tutors have to put into slightly mundane tasks, the more time they can put into thinking about how to improve and engage their student.
What AI can’t do and what tutoring should lean in to
AI tutors will improve, of that there can be no doubt, but most children don’t want endless screen time and most children would rather interact with a human than a bot.
There is nuance in the way children learn that AI is lightyears from being able to pick up on: attitude, motivation, organisation, drive, confidence. This is where tutoring must look to, an area Clarendon already excels in and is working hard at to improve our mentoring and coaching offering.