The Struggle is Gone

When I was 18 I started studying physics at University and got my first real taste of “the struggle”. I’d spend hours at a time working through problem sheets, with a small set of data and some collection of methods, trying anything and everything to get the right answer. I’d work through the notes and talk to friends studying and, honestly, struggle.

When it came to revision time at the end of each year, something cool happened. The bits that I’d breezed through (or copied from somebody else) were foreign to me - I’d have to relearn them from scratch. The bits that I’d struggled with were the bits that I knew.

Learning to Code

When I was 22, I decided that I wanted to learn to program. I started off with Python and Ruby tutorials and yet again…the struggle was real. This time I recognised it, and knew that it was good. Knew that any time I wrestled with package installation, or hunted down an off-by-one error, or nerfed my operating system, these were just the ways different ways of learning.

It worked. By the time I was 23, I was employed as a data scientist - programming every day. And that’s how I’ve spent the last 13 or so years. Programming and struggling with data, with varying sizes of data and a growing collection of methods, trying anything and everything to do something useful.

Learning in the Age of AI

I recently saw the guy who had a brief flash of fame sharing the ChatGPT prompts that got him through university at his graduation. I was reminded of myself: as an impoverished student, I made a £50 donation to Wikipedia in recognition of the help I’d received over the prior 4 years.

However, I do think there’s a real difference with AI and that’s because the struggle is gone. It doesn’t have to have gone - I’m not saying it’s impossible to use AI for education, and in fact I think I really motivated learner could probably go much further much faster by using AI.

But I do think it has become increasingly difficult to struggle for prolonged periods of time on a problem, now that we know the answer is often a few keystrokes away.

Mobile Phones, Kids, and Balancing Momentary and Long-Term Dissatisfaction

If I can offer an analogy from my academic work, let’s consider phone use and children. Boredom sucks. There’s a famous psychological experiment where people gave themselves electric shocks rather than endure the horror of boredom. “All of humanity’s problems stem from man’s inability to sit quietly in a room alone..

And yet, increasingly, boredom is being viewed as an essential part of growing up and developing creativity and initiative and resilience, and all sorts of good things.

Any rational parent would choose to allow the boredom, taking the short-term pain of a whiny child in exchange for the long-term benefits of a child that can occupy themselves. But in practice, that isn’t what happens. Kids whine about a long car journey. They get given an iPad. Then they whine about a haircut, and get given one for that too. Ultimately, they find that the key to getting the iPad is whining, and before you know it, you’ve got an iPad kid.

The point is, while we’re aware of the long-term benefits of the psychologically discomforting experience, we’re wired really strongly to avoid psychologically discomforting experiences (sort of by definition) and so face a real uphill battle.

The Struggle

Struggling on a problem that you can solve in a heartbeat is one of those psychologically discomforting experiences. So sure…you can be disciplined, and fight your inner workings and force yourself to struggle and struggle and struggle.

Some people will, and those people will learn faster and more effectively and will end up better than those before them.

Most people won’t. Most people, when faced with a 3 hour long struggle or a 10 second query are going to go ahead and take the shortcut and skip the struggle. This is not me being high-and-mighty, by the way. Me too. It’s hard to justify spending an hour reading a data schema when you can just bung the thing into ChatGPT. Why bother learning how to create the perfect scatter plot when Claude will do it just fine?

Conclusion

I think we’re in for a wild few years when it comes to education. Learning hard subjects ultimately requires struggling, and LLMs have made struggling obsolete. However, LLMs are advancing at such a rate that perhaps we no longer need large swathes of graduates who know how to solve hard problems by themselves. I don’t know any of these things - I’m just along for the ride.