Why coding bootcamps won't teach you AI (and what will)

How non-engineers should really learn AI

A friend recently asked me how she, as someone with a humanities degree, should start learning about AI. Her plan? To enroll in a coding bootcamp. I had to stop her right there.

There's a widespread misconception that "learning AI" means "learning to code." I see this especially among people from non-technical backgrounds. They believe they need to become programmers to understand AI. This couldn't be further from the truth.

Here's why learning to code might not be your best investment:

1. Starting from scratch, you'll likely spend months, if not years, to become even a basic programmer. Ironically, this entry-level programming work is exactly what AI tools like Devin are starting to replace.

2. If you're worried about AI disruption, the answer isn't to abandon your strengths. Instead, double down on what you're naturally good at. If you didn't choose to study computer science initially, there's probably a reason for that. Here's a universal truth: in any field, mediocrity makes you replaceable, excellence makes you indispensable.

3. "Learning AI" isn't synonymous with "learning to code." Even programmers need to learn AI! Large language models represent a fundamentally new technology. Fun fact: only a tiny fraction of people in the world (think machine learning PhDs) can claim to have a traditional "AI background." Everyone else, including most computer science graduates, is learning alongside you.

So what should you actually learn about AI? Three key areas:

1. Understanding the fundamental principles of AI and large language models - how they work, their capabilities and limitations

2. Staying informed about industry trends and the rapid pace of development (the direction and rate of change)

3. Learning how AI can solve real-world problems, how it's being implemented across industries, and how it translates into viable products and business models

Here's the irony: even if you learn to code, you probably won't achieve these learning objectives.

So what's the better approach?

Use AI! Use it extensively! I'm constantly surprised by how many people working in AI rarely use AI products themselves. If you can become a daily active user of AI tools, you're already ahead of 99% of people. Try using AI for everything - learning, working, research, as a thought partner, as a mental health counsellor... And do it deliberately, with intention.

You might say, "But I don't know what to use AI for!" That's okay - start by using it anyway. Use cases will naturally emerge as you experiment. That's the fascinating thing about AI products: you often discover their value through hands-on experience. Yes, it might feel clunky at first. That's normal with any new technology. The tools will improve - it's just a matter of time.

Now, here's an important caveat: while I advise against learning traditional programming, that doesn't mean you should shy away from AI coding tools like Cursor and Windsurf. In fact, I strongly encourage you to experiment with them. Try building simple products, even if you've never coded before. Why? Because software development was the first area where AI found genuine product-market fit. By tinkering with these AI coding tools, you'll gain firsthand experience of how AI is transforming knowledge work. You'll see both the potential and limitations of AI assistants in real-world applications.

The key difference here is that you're not learning to code in the traditional sense - you're learning to collaborate with AI to build things. This hands-on experience is invaluable for understanding how AI is reshaping the future of work.

Approach using AI with a spirit of play. Let curiosity drive you. Make using AI a hobby, and you'll naturally start exploring the boundaries of what models can do. You'll find yourself staying current with industry developments. Most importantly, you'll develop an intuitive understanding of how AI can solve real user problems.

Achieve this, and your understanding of AI will surpass that of many programmers - no coding required.