College will only cost $20 a month

From 2020 to 2022, I've encouraged a steady stream of students to pursue their education in computer science. Tech companies were hiring like crazy. I would interview 3 candidates a week on some occasions. The future of tech seemed to be paved with gold. Any day now, several of those students will graduate. I dread opening my inbox and get an email from them asking:

"So... about those jobs?"

Things have changed since those crazy old days. Hiring has come to a stop. Those graduates will step into a market that looks nothing like what they signed up for. Getting a job will be a struggle. If I could see the future, I would have recommended a different path. But then again, how could I have known? Hindsight is always 20/20.

So this leaves me with a dilemma as a parent. What do I tell my own kids? Am I going to push them to pursue the same traditional colleges? Will I disburse life savings so they could attend a prestigious university? Or will I just buy them a $20 a month AI Subscription?

Imagine handing a student a textbook straight from a university syllabus. They read a lesson, and then you have an AI tutor to quiz them, debate concepts, and break down difficult ideas until they fully grasp the material. Is that a good replacement for college?

Some argue that college is about the experience. It's the friendships you make along the way, the networking, the independence. These same people often omit the lifetime of crippling debt that hangs over those friends twenty years later.

When I was in college, I focused almost entirely on mastering the material in the books, completing all assignments quickly. I spent the rest of the semesters tutoring my peers and applying the material I had learned. For me replacing the entire college system, the brick-and-mortar university with a set of books and an AI subscription seems like it would have done the job. I was the target candidate for this model.

Yet, now that I have these AIs to experiment it myself, I found two flaws in the model:

1. Understanding takes time.
2. Not having an immediate answer forces you to strain your brain until you figure things out.

When I wrestled with dense textbooks in college, solutions didn't arrive at the click of a button. Instead, I struggled. I hit dead ends. I tried solutions that seemed promising only to abruptly stop two hours in the process only to realize that the whole thing was wrong. Sitting with the problem, the trials, the errors, giving up and going for a walk to clear my head, that was the process that taught me how to think. It taught me how to debug.

Even when I found solutions online and pasted them into my code, I had to encounter the same problem repeatedly before the concept finally made sense. You have to work your way through a solution to understand a problem.

This is why I feel concerned when we throw Millennium Prize Mathematical problems at AI. Imagine an AI presents a complete proof to one of these unsolved problems tomorrow? If or when AI solves it, we still have two problems.

1. We still have to understand the solution. Having an AI produce a proof doesn't mean we understand how or why it works.
2. We don't get to discover new problems.

The second one is often overlooked, but ask yourself, where do these problems in Mathematics come from? They come from mathematicians attempting to solve hard problems, taking detours and failing along the way. Remove that step, and you... Well, you don't find any new problems.

Not just in mathematics, I feel we will see something similar in code where fewer applications rely on frameworks. Frameworks solve real issues in programming, but why use a framework when the AI can just give you all the answers. I lived through the JS framework mania. I am not suggesting we create more, but I think we will miss out on discovering new problems that need solving.

If we rely on a system that just hands us a solution, we lose the ability to uncover new questions because we never went through the steps of trying something new.

I don't think a $20-a-month AI subscription paired with a textbook is the ultimate solution. Not when we still need some friction in order to inculcate concepts in our minds. But traditional higher education has changed so drastically and the skyrocketing costs and changing employment realities don’t help. There is no turning back to the old model either.

The challenge for learning remains understanding. How we grasp concepts and develop critical thinking in an era of instant answers. Maybe that's where we should turn our focus to. Figuring out how to design learning that somehow preserves productive struggle. If AI can solve this issue and turn students into masters of crafts instead of prompt engineers, I mean, I'll gladly keep that college tuition in my pocket.