In many ways, much of the dialogue around Generative AI is remarkably simple and, in it’s own way, comforting. There is a menu of polarised or provocative positions that one may choose from (a choice of starters from bias to untruths, a main course of hallucination, and for those with a sweet tooth, the death of the artist), as well as a general flavour of tech utopia or dystopia to flavour it with (AI will save or doom us, today or tomorrow, as soon as it becomes sentient or weaponised).
What’s not to enjoy? In a very real sense, all of these positions, topics, and ideas are true, or real and we are wise to be wary.
But these ‘truths’ are not definitive truths. They are features of a landscape, but not the landscape itself.
When we wrote ‘Engines of Engagement: A Curious Book About Generative AI’, we were keen to write the book that nobody else was writing and, hence, we wrote a book of uncertainty and space. We sought to inform the stage of debate, not lecture to it.
The important word is ‘curiosity’. The book is curious both in terms of our own outlook, and in terms of the artefact we produced. It is not a book of certainty.
One of the ideas we explored was the question of ‘intelligence’ itself. I was at a lecture recently where an eminent professor casually dropped in that they expected to see General Intelligence announced this year, probably by late summer. Others tend towards a rather more conservative view and say that we will never see it at. Our position: well, we remain unsure (although we are not cancelling our summer holidays to wait for an announcement), but the debate may be eclipsed by circumstance.
One could argue that semantics aside, if we treat a system as intelligent, then in a pragmatic sense it holds the card for such a quality. It’s kind of like Google Maps. Is it ‘trustworthy’? Well not in a quantifiable sense. But still i trust it. At least to get me to the cinema. At the point where we trust the output, where we treat the system as an intelligent one, then maybe it is?
This week has seen much excitement (at least in the Society of Nerds, of which we are enthusiastic members) in the notion of the Universal Teacher. Not a sequel to the Jean-Claude Vann Damme masterpiece, but rather the notion of a personal ‘teacher’ adaptive, unique, and expert at individual tuition. Perhaps the distinction is that this is not an ‘intelligent’ system that i can interrogate, but rather a system that can actively teach.
Are we about to meet the Universal Teacher? Or some semblance of such?
A technology that holds mastery (or again, maybe not a human conception of mastery, but which nonetheless is faultlessly masterful), that can react to individual performance, can structure developmental pathways and the formation of schema and knowledge, and which can adapt in approach according to individual performance. A system that can teach.
At first glance, this sounds obvious and (almost) easy. Right now i can utilise free to use tools to summarise books, synthesise ideas, answer my questions, explain their reasoning, produce step by step worked examples, or entertain me with stories, riddles and songs. That’s teaching right?
Way way back in the popular narrative of Generative AI (e.g. April last year) Donald Clark provided both a rounded exploration of what a Universal Teacher is (and exceedingly annoyingly made the Universal Soldier joke way before i came up with it), and posited that we may be headed towards it. Just over a year later, he indicates that it may now be here.
http://donaldclarkplanb.blogspot.com/2023/04/are-we-on-verge-of-having-universal.html
He also updates his definition with the rather offhand description (for such a monumental thing) that a Universal Teacher is “A free teacher who speaks, listens, remembers, tutors, using all media types, can read handwriting, provide personalised feedback, on any subject, anytime, anywhere, in any language”.
http://donaldclarkplanb.blogspot.com/2024/05/is-teaching-becoming-obsolete-with.html
There is little wriggle room here. Almost every word in that definition is built upon billions of dollars of investment, and decades of research. But the second word may be the most important.
When we talk about learning, about education, about teaching, we tend to conflate two things. Pedagogy and the Business of learning. There is an underlying science of learning, and there are many types of business built upon it (or in some cases, tangentially leaning upon it, but without much in the way of rigid foundations).
Generative AI and the layers of technologies that, together, are undoubtedly close to a notion of a Universal Teacher form a pedagogic revolution. And these same technologies form a tsunami that will break against much of the ‘business’ of learning.
I caught up with an old friend this morning, who talked about his daughter going through university. She’s studying art and he was decidedly underwhelmed by what she was being given. Little challenge, little space to experiment and explore, not much of a social experience (as half the students are remote and there is no structured online collaboration) and a hefty loan to pay for it all. This is the Business of learning. It’s just that at some point it may have forgotten about the learning part.
We spend years of our lives in formal education (if we are lucky enough to live in countries that can afford it, and where – dependent upon gender – we are allowed to access it). When we consider the idea of a Universal Teacher, it’s all too easy to focus on the details and minutia, the philosophy and pedantry, at the top of the debate. The reality is that for many people just a decent teacher would be a benefit. An imperfect but available one.
For me, the progression in these technologies brings risk, for sure, but overwhelmingly it’s a matter of hope. In the book we discuss how Generative AI makes high quality dialogue – once an expert and rare feature – commoditised. When we wrap that dialogic capability into a more universally capable engine – capable in that it’s trans-disciplinary, multi talented, not to mention able to ‘speak’ in many tongues – and as we see more radical development of the ability to truly adapt to individual cognitive differences and ability, we do see something truly incredible.
Because a Universal Teacher is a foundation of a model of learning that becomes delaminated from structural systems of education. It also paves the way for a re imagination of the structural landscape of learning. If we no longer need systems built upon industrial models of education, perhaps we can see more diverse entities emerge that support the social aspects. Instead of going to a University, maybe i subscribe to laboratory facilities, sense making communities, a Student Hall, and a book club.
Perhaps the reason why the latest progression towards a Universal Teacher has been somewhat ignored in more mainstream media – despite it’s radical potential – is that we don’t care that much about teaching. But to ignore it is to miss the point that sooner than later we will be considering our stance on the Universal Clinical system, the Universal Lawyer, the Universal Financial Advisor, and Fitness Coach.
My own school will soon celebrate 300 years of teaching. I suspect that in one years time the Universal Teacher will have undergone more change, development, and impact, than many of those 300 years.