AI and Education

When I was at school in the late 1980s, our school got a computer. Just one, in the library. I remember using it for some document I wrote, I can’t remember what, on a floppy (yes, actually “floppy”) disk. At university we had computer labs, but I still wrote my undergraduate dissertation on an electronic typewriter. My MA was written on the computer and during my PhD I published two articles on an “electronic journal”. But computers had not had a significant impact on education.

Fast forward to now, and most schools have Google accounts for their students, even at primary school. During lockdown lessons were delivered on Microsoft Teams. Online applications are not only used for teaching Computer Studies, learning Python, HTML and CSS. They’re also used for maths homework and times tables, as well as pointing to videos on BBC website and others.

But computers and mobile devices have just replaced text books and written homework. They have changed where learning happens, not how learning happens. But I think AI should create a monumental change in the steps through education. The question is how long it takes academics to realise it.

Historic Focus on AI in Education

Up until now the basic perception of AI impact on education is “cheating” or getting AI to “write” your homework for you. The focus has been on how teachers can identify AI-written work and prevent the cheating.

The problem is this doesn’t pay attention to how AI is being used in business and how that could impact and benefit teaching. The obvious impacted subject is Computer Studies.

Computer Studies

The obvious sphere of influence for AI is in Computer Studies. Compare how it’s radically impacted IT and you can see some obvious ways it should change teaching. I remember stories of university teaching where students were required to write code in NotePad. I don’t know if they were apocryphal, I hope so because it’s pointless - no programmer in business writes code without an IDE, there was little point in teaching it. I don’t agree it’s like teaching arithmetic without a computer, I’d say it!s more like teaching woodworking by giving the students a blunt penknife!

Teaching coding using coding agents has to be the future. Get the students to understand the code by asking questions of the coding agent, create and understand what constitutes good documentation, work out how to troubleshoot and avoid naive AI approaches. The AI-native programmer / developer / engineer is key for employers over the next five years. It’s about being able to steer an AI agent to get the right intent.

Incidentally, I think AI is already having an impact on how existing IT professionals learn, which will have (or maybe already is having) a big impact for those who have built careers writing coding training courses.

The Progression of Learning

But AI should have a much wider impact, across all subjects. To understand how and why, it’s important to look at how levels of learning change as children grow, and how that relates to how AI is used in business. Students go through a number of steps from nursery schools (kindergartens) through to post-graduate education.

1. First comes learning to speak, read and write. This is foundational and obviously is still required. Text-to-speech and speech-to-text tools are powerful, but reading and writing is critical to critical to AI usage. In fact I would argue they become even more important. Instructions and questions need to be phrased clearly and unambiguously. Responses need good comprehension skills, which requires strong reading skills, to identify assumptions and next steps.
2. Reading and writing evolve during primary school as vocabulary complexity evolves and quantity increases. Again, this won’t change and in fact will become more important.
3. The big change comes in secondary / high school. The kind of learning required at this part of the education journey seems unchanged since the middle of the last century at least. It’s about reading more, gathering more learning, cementing that learning. And it’s fundamentally about trusting what you’re told. I think that is why so many people are struggling to see through populist politics - they either do not have the skills to challenge what they’re told or want to spread a particular biased narrative for their own purposes.
4. Education in the mid-to-late teens starts to focus on absorbing larger quantities of material and a broader range of opinions, to start to aggregate a lot of information into key points. This is continued through undergraduate studies.
5. Some undergraduate dissertations may begin to formulate original thinking or challenge opinions. But it’s usually during postgraduate education that this becomes a key element. Dissertations have to be longer, which means absorbing large quantities of material and generating large quantities of content.

Degree-level jobs usually expect the ability to employ this ability to absorb and generate structured content, as well as be able to work independently. Junior employees do this work while learning the judgement to make decisions themselves, which is what starts career progression.

The problem is AI is transforming what junior employees will need to do. AI allows you to do the research, to aggregate large amounts of data, and even to generate content. What junior employees need is the willingness to dig into sources and verify accuracy - this is not a skill it’s a discipline. Then they need the judgement for what makes good output, as well as the ability to filter information for what needs to be remembered, and the ability to retrieve it. These are not currently something AI can achieve, but that may change.

This all means using AI in schools. You can’t learn the discipline of drilling down into sources if you’re not using content generated by AI. And more importantly you don’t need to learn how to gather large amounts of information by reading if jobs use AI to do it for you. It’s a waste of time, focusing on something you’ll never do while not teaching what you’ll need to do.

When Will Education Evolve?

This is a radical change in education and won’t happen quickly. There’s a period of time that technology needs to catch up.

Per-token pricing is already having an impact on AI usage, even though the hyperscalers are trying to adapt to support increased demand. It will take time and impact more widespread business adoption by months and maybe more than a year, regardless of whether your priority is sovereign or hybrid AI.

Hosted or internal AI has its own challenges, and supporting throughput seems the biggest one. Technical advances like Turboquant or Rotorquant look like they will have a big impact, but they are still in their infancy. Distributed AI has challenges as well, we’ll see if they can be overcome.

On-device AI will have a major impact, not lest because they mitigate the power consumption concerns. Modern Macs have been able to run smaller models for a while, the challenge is memory. The recent announcement about NVIDIA RTX Spark and Microsoft bring Windows into the game as well. The Grace CPU and Blackwell GPU handle the VRAM limitation.

On the model side, NVIDIA already have the Nemotron family of models. Cost looks like it will be quite high for a start, but I’m sure that will change. Google are also targeting on-device inference, both for mobile and tablet as well as laptop, with their Gemma 4 models. Notably, Google are also targeting the consumer market with AI mode in their search engine and their deal with Apple. Ternary models also address the model size problem by taking a different approach than quantization.

All this means I think we’re in the AI equivalent of the early iPhone years. And I think it will be as pervasive in business within five years, mirroring the growth of the iPhone, which moved from widespread consumer adoption to business pervasiveness in a similar time period.

Why Education Needs to Evolve

This means education has maybe ten years to evolve. It’s not a given that it will, particularly because it’s such a major change.

The problem is that those who have AI-native skills will be highly sought, those without the skills may become a forgotten generation. It could also have a massive impact on mid-career and older employees, who just don’t have the skills to adapt to AI.

It’s not that the jobs will disappear (although some may), it’s that the job requirements will change, decision-making skills and responsibility will need to move down the career ladder, and the AI-native generation will be better qualified.