AI-first educational institutions

In a previous post I mentioned in passing the idea that universities, if they want to remain relevant, will need to adopt an AI-first position for everything they do. Soon, AI will be embedded in everything, from preparing course materials, to developing assessments, to budget allocation. And email. It will be a huge part of email.

I don’t think that universities will be able to make this leap to become AI-first, not even for assessment, let alone for everything else.

And the reason I can’t see universities adapting is because they’ve had 30 years to adapt to the internet, yet still centre almost all learning, teaching, and assessment, on the idea that information is scarce. Even when universities have seemingly embraced the concept of information abundance – through MOOCs, for example – they end up bending the technology to better fit their traditional forms, rather than using the technology to adapt in other directions.

From what I can see, universities are unlikely to allow AI to have any real impact on their strategic objectives. You may argue that it’s still early days and that large institutions take time to adapt. I’d counter with the argument that ChatGPT launched about a year ago, and we’ve already seen Microsoft and Google shift their entire strategic position towards generative AI. While universities form committees to discuss “AI-resilient” assessment tasks, or to debate the intricacies of institutional regulation of AI.

I predict that we’ll soon see the launch of new types of educational institutions that position themselves as AI-first in everything. In other words, they’ll reimagine what a curriculum could be, given the existence of generative AI. They’ll make explicit choices about what value the human adds on top of AI, and hire accordingly. They’ll automate administrative tasks and accounting. And they’ll be cheap. At least, cheap compared to the cost of a 3-year degree programme at a traditional university.

You might argue that an AI-based curriculum isn’t as good as the best that existing degree programmes offer, but that’s the wrong comparison. An AI-based curriculum need only be cheaper than the average degree programme. And even if the AI-based programme is only 60% as good as the average university offering, next year it will be 70% as good. And then 80%. And all the time, it’ll be getting cheaper because it will offer truly personalised access to expertise at scale.

You could make the point that students go to university for more than gaining knowledge. They also gain access to networks, social opportunities, diverse communities, new ideas, and so on. But we’re pretty arrogant if we think that universities are the only way to build out professional and social networks, and to be exposed to radical ideas. There are other ways to expand your horizons.

Universities must shift towards a paradigm where AI is at the centre of every activity in the institution, because it will be embedded in every piece of software and device that we use. But at the moment, we’re merely asking tentative questions about how to move away from essays as the dominant form of assessment. We still believe that universities offer something essential on top of access to specialised information and expertise, and I’m not convinced that we do. I might be wrong, so if you can think of any examples of what this might be, and why you wouldn’t be able to get it somewhere else, please let me know.

If universities are going to successfully adapt to AI-first learning in higher education, we have a lot of work to do.