I recently made a point about the disconnection between good programmers and formal degrees. I also expressed my skepticism about universities being the best option for becoming a good programmer in the future, with LLMs being a disruptive factor. I want to elaborate on the reasons. 

It has been the case for a while that anyone can access top-quality content to learn. That includes courses delivered by world-class experts. Some are free, but most are not. They still cost a fraction of what university tuition does. You can also instantly get the best books on whatever subject. And you have all the open-source code you can read at your disposal. All this was just not possible 20 years ago. And all this will be even more possible 10 years from now.

University professors play a crucial role. Good ones will make you think, challenge, motivate, and engage you. Very few of the professors I encountered in my time were like that. The rest were just not. Many covered a wide range, from having zero motivation to teach to showing a tremendous lack of competence or a combination of both. In some cases, what they taught you caused more harm than good. I remember fondly how, in the very crucial software engineering course, I learned a methodology nobody on planet Earth uses and that coding was a residual part of the process. And I paid for that! Something as simple as the score rank in Coursera and some informed comments on the course would have been such a luxury.

How would you organize the curriculum? Well, virtually every university publishes the curriculum they offer (e.g., Harvard or MIT). That’s a good starting point. You can also ask some LLM  to design a multi-year curriculum with the fundamentals. I’m sure it won’t be perfect, but I’m confident it will be better and more up-to-date than the curriculum I went through in my day. Now, combine that with Google and more LLMs to get recommendations on books and courses. You can have a top education path ahead in a few hours of work, laid out by a fraction of what college costs. I’m talking about Spanish standards. For American standards, the cost will be just negligible.

Then, there is the question of having access to an expert to discuss subjects and ask questions. Here, again, LLMs can be disruptive. A 2025 LLM is already an outstanding tool for discovering new domains, asking questions, getting recommendations to practice, and bouncing ideas. Imagine a 2030 one! I doubt they will ever replace fantastic human teachers, but those are rare, and even when you are lucky to have one, your access is constrained.

Finally, there’s the question of certification, which holds only as much value as the market assigns. In my experience, it’s not very high for software engineers since plenty of fantastic programmers don’t hold a formal degree. And the disconnection is bidirectional. I’ve never heard anyone praising programmers from a specific university over others. So maybe, if you want to make a career in large investment banks, not having a degree is not an option. But, for programming, it is.

I’m not suggesting that universities are useless or don’t have value for becoming a professional programmer; I just think they will not be the best option in the future. More than universities, I see a bright future for academies that combine world-class human teachers and AI, offering such education for a fraction of the cost thanks to global large-scale access.