Dispatches 004: Shape of Learning in a Changing World

Knowledge is a kind of ethereal liquid gold. The ultimate fungible resource. It opens doors. It changes how you can approach problems and the perspectives you have access to. It changes how you interact with the world, and therefore holds the ability to change your life.

That’s what I have always believed.

We live in a complex, globalised world. To make informed choices, you have to understand its shape. You have to possess the right cards to open the right doors. That’s learning in a nutshell, for me. The world & the keys.

Learning as such exists in many, many forms. Formation, skill-building, understanding a domain, developing ‘life skills’, learning about yourself, learning about learning. As a result, talking about “the future of education“ is like talking about everything and nothing at once.

After much reflection, I think it’s useful to separate out:

1. What are learning and education for to begin with? What kinds of learning do we need in society?
2. How do we build institutions, formats, and approaches to most effectively facilitate that?
3. How do we approach leaning outside of structured formats, i.e. lifelong / career learning?

You’re welcome to disagree with my delineation; I just find it handy.

On a very fundamental level, it comes down to open-ended and closed education: one set that serves specific goals (I.e. skills, domain proficiency, etc) and one that is exploratory by nature (I.e. engaging with complex topics such as history, culture, politics — among many others). Convergent vs Divergent thinking.

Both, of course, almost always intersect.

Institutions, by virtue of their sheer size and taking up entire dedicated portions of life, have deeply blended mandates. A university does many things beyond just teaching you skills and helping you to understand the world.

Other forms of education I have interacted with tend to skew strongly towards one or the other: open-ended, exploratory residencies on one end, and something like purely skills-driven MOOCs and training on the other.

Most things exist somewhere on that map. You could probably also split them by structure level, to get something like this:

By now you can see the picture. Under most current structures, there are several institutionalised learning structures (School, university) that are designed to equip you with core knowledge + some level of domain specialisation (That is inherently tied to a career later on).

The world, as it happens, is changing.

AI is changing both education itself and the purpose education serves. If job openings in major areas such as computer science and business take a hit, that feeds right back to degree demand and utility. Whether this will happen and how fast is up for debate, but I think it is abundantly clear that at least the skills portion of degrees will have to change massively.

As a result most, of the conversation around the future of education has focused on how universities and schools are under strain due to shifts in the job market and the economy as a whole. My take on the discourse is… well, yeah. Institutions are structures, designed to serve a specific purpose, with many complex organisational and legal layers around that purpose. Asking a university to reform in response to AI is like asking to rebuild the foundations of a house.

Compartmentalised, program-level learning is much faster to adjust — and this is why it will likely be filling many gaps over the next 5-10 years, as institutions catch up on reform. Given the scale of the change, we may need new institutions, period. For now, our best bet is rapid adaptation that addresses the most critical gaps while laying the groundwork for major reform.

This is also where we are likely to see the most innovation in EdTech. To jump ahead of myself, we will likely see a broader bifurcation. Highly adaptive, AI-augmented education tech powering unstructured, lifelong learning — accompanied by downright Socratic discussions in amphitheaters.

Universities moving firmly towards open-ended, social discourse and structured exploration, while new forms of EdTech provide supercharged adaptive learning elsewhere.

That’s my guess, anyway.

Institutions and Their Discontents

If you had a black slate, how would you design a university?

I have often wondered that myself. Because once you get into it, it’s devilish task.

Universities are of course institutions. The system depends on the country, but they universally produce a public good: knowledgeable young people who will go on to join the workforce, fill jobs, or even create them as they start their own companies. As such, universities are fundamentally shaped to fulfil that function. The entire degree system with its attached credentials acts as proof that Person X possesses the right domain skills, taught at the level of the associated institution. That’s the minimum promise of a university.

For the people enrolling, university life promises many other things. Meeting people from all walks of life. Exploring dozens or hundreds of societies and communities. Building all manner of life skills beyond just the degree.

In short, it’s a shaping structure. Spend a couple years of life here and we’ll equip you with all the basic kit you need to build a life.

Schools fulfil a somewhat similar function lower down the line, though naturally far more focused on formation than specific skills. The basic role of a school is to provide fundamental knowledge about the world, meet others, build social skills and learn how to learn. It’s a space to build your initial world model and catch yourself up on what humanity has figured out in our charming 12,000 years of post-nomadic existence or so.

Another shaping structure. Just even more foundational, and with even longer-lasting impact on a person.

This tells me several things:

1. Schools likely won’t be hit as strongly in terms of pure mandate. The goals do not change, whatever the economy. The structure is still useful. Where the trouble may arise is things like attention spans: how do you tell a 12 year old to engage with the class when they have a great shiny phone and 7 second reels on Instagram?
2. On another angle, schools traditionally act as a pipeline to university education. So if universities change, that has an impact earlier in the chain. What kind, I’m not so sure. Perhaps the progression will just become less clear, less focused on getting into the “best“ university and on test scores. It’s really a question about what will future universities be testing for to begin with.

On universities:

1. The jagged frontier finds us here again. Read: uneven impact. Elite universities will likely be able to coast on sheer prestige and status; signalling isn’t going anywhere. 2nd and 3rd tier universities, however? It will get dicy. Most likely the impact on degrees will closely mirror the job market, to say nothing of AI disrupting the entire assessment system. Personally, I think the whole cheating with AI thing is a major but interim problem that will get plugged one way or the other. The harder conversation is, do we adapt universities to still be useful for future careers, or do we shift function of these institutions?
2. Degree streams might generalise. This is extrapolating from my thinking about the shape of roles — that we will see reversal of specialisation. You just don’t need dedicated SEO marketers or web designers when you can have Product Managers and GTM Engineers. Ignore the buzzwords for a second. What is really happening here is that AI is subsuming lower abstraction levels like delegation, effectively moving many professionals higher up the chain. They still need to see the full picture and occasionally intervene. But that massively changes where the focus goes in domain-specific education and skills.

My takeaway?

Minor crisis for schools, but overall shape and mandate intact. Far larger challenges for universities down the line. Some will respond by adapting to the job market directly. Very few might try full scale reform, or even creating new institutions. As always, most likely we will see the middle road: new colleges within existing institutions to pioneer models while splicing and dicing degree offerings around the board.

Salons & Second Brains

Stepping outside of institutional education, what’s happening right now is a very interesting dual trend.

AI has finally surpassed the limits of previous static software. Agents are still in their early days, but Claude or ChatGPT can already design you a personal learning curriculum. I’ve built several myself, using a master-CV and general portfolio as foundations for the AI to infer my existing expertise. After including deep dives on specific projects, the results are not half-bad.

Simultaneously, we’ve seen an explosion of IRL events. Salons, firesides, workshops, hackathons. It’s a quirk of the startup scene of course, and London especially, but it’s not just that. Many novel programs and fellowships have emerged as a response to the changing job market and increasing pace of technological advancement. These range from Build Club to FR8 or Arrayah, or programs like ASES or 180FLP for young leaders I was myself a part of. Peer-to-peer, project based learning, heavily focused on learning by doing, with some degree of structured scaffolding to guide you.

It’s a bifurcation.

Why? Well, trend and countertrend would be my guess.

AI is finally delivering us AI tutors. Lo, Aristotle in a Bottle. Previous generation EdTech was ultimately stuck on MOOCs, video courses, and perhaps dedicated gamified apps for language learning. Advanced in AI are driving something of a revolution.

It’s not just hype, either. I’ve worked on this before when we tried to create a dynamic AI skills platform for businesses in 2023. Guess what? The tech didn’t exist. We simply could not make adaptive learning work short of prerecording massive quantities of material and strapping that to an adaptive engine that would structure it for you. Now, though? You could upload your entire company’s project management history coupled with personal CVs and statements into a secure environment, and have an agent match everyone’s experience with company priorities to produce a dynamic upskill plan. And this is the comparatively simple stuff.

At the same time, dedicated programs are front-loading the effort by designing a structure for you. Here is a space, reputable professionals to talk to, a core structure, other carefully selected peers. Off you go. Many of these are cropping up in tech, but the fellowship / residency model seems to be expanding to writers, creatives, and more. It’s a solid model, personally — a kind of concentrated upskill & exploration sprint, with meeting great people as a perk.

Altogether, this is likely the future of lifelong learning. Or just learning, really. That first part will become the norm soon enough.

Agentic AI systems can act a responsive copilot, working directly with your skillset and professional experience to find opportunities to extend yourself, learning something new, challenge your view, and more. As one example, I’ve given Claude my entire lifetime library CSV and asked it to screen for adjacencies & blind spots. So far it has recommended me a dozen books from countries I don’t have on my shelf, plus topics I haven’t engaged with much before.

I’m trying to do something similar with skills — looking for “negatives“ and adjacencies within my capabilities map. The knowledge mapping process is not very viable right now outside of compiling your CV, but this will be overcome eventually.

Simultaneously, events and programs are plugging the gap on social connection & experiential, project-based learning. There will always be demand for these for the simple reason that not everyone wants to spend time designing one themselves, me included. It’s very simple, reliable format to come together with 20-50 people, discuss changes in our field, and learn something in a dedicated environment.

Both of these are already getting picked up in industry. Companies like Sana developing learning agents on one end, and internal hackathons being hosted by tech companies on the other.

So… All Is Great Under the Sun?

I wish.

I believe that AI has the potential to revolutionise learning. I really do. Both by massively accelerating skill-based learning, and by shifting education towards open exploration, discussion spaces and more formation learning.

Where I see major challenges is with the transition.

If the job market moves too fast, institutions will struggle to adapt with similar speed. We have to keep in mind that this is an entire chain — kindergarten, school, university, work — playing out over some 20 years. The ripple effects have long waves.

Then come the really thorny questions. What if optimal degree length is now two years instead of three? Does tuition go up by 50% to compensate? Do universities downsize? Will they even be able to change such a major structural aspect of the institution?

If push comes to shove, deep structural transformation will be contentious. Universities are a quasi-oligopoly. There are only so many reputable, recognised institutions with deep pockets and very sticky reputation. They may very well just decide to take their chances. As a result, institutional education may end up lagging behind its potential for years and years.

Other challenges are structural. If frontier EdTech ends up primarily adopted by leading companies, that will create significant differences for people depending on employment. Fellowships, programs and residencies have a similar problem: they are almost universally very selective, and therefore cannot plug the reskilling / upskilling gap for anyone but a small cohort. This is even worse when availability across different countries is factored in.

In short, what we are missing is a broad front offering to support mass reskilling, should that come about. Apprenticeships and open learning materials might pick up some of the slack here, but probably not all. We will need new philosophies of education, new structures — at least transitionally.

Learning in Our Changing World

As always, change is slow in the short term and fast in the long term.

AI is upending much of education: if not directly, then by driving major shifts in the economy. Institutions will adapt — they aren’t going anywhere — but people will likely have to adapt faster.

There is an inevitable shift of responsibility on the individual here. Collective structures will adjust, and we need to keep having that conversations. For now, though, all of us have to take a long hard look at the future.

Move beyond skills and jobs. Think about the shape of your entire domain over the next decade. In parallel, learn to manage your own learning, and learn to use AI to do so effectively. Don’t just automate and plug gaps; use the time freed up to explore deeper, learn widely, develop new soft and meta skills. Spend time going to events and looking for programs valuable to you. Experiment.

The potential of the current wave is that you can massively supercharge your skillset and capabilities while shifting time to what one could call ‘long formation‘. Engage with the world, with different perspectives. Spend time doing all the impractical things that you wouldn’t have time for before.

Live more fully.

In the end, the point of learning is to give keys to the world. Well, use them. Find new locks to tinker with.