The Fastest Way to Learn AI Is Publicly

AI gives us an always-available learning partner. Sharing what we build gives the rest of the engineering community a chance to teach us what we're missing.

There's never been a better time to learn how to build software. The tools available to us today have dramatically lowered the barrier to experimenting with new technologies, exploring unfamiliar ideas, and building things that might have previously felt out of reach. AI can help us navigate an unfamiliar codebase, explain a concept we've never encountered, or get a working prototype off the ground in an afternoon. It's an incredible learning opportunity, but I think we're still figuring out how to take full advantage of it.

I've been thinking about this particularly from the perspective of junior engineers entering the industry. In my previous post, AI Does Not Replace Senior Engineers, I talked about how experience and engineering judgment become even more valuable when AI takes on more of the implementation work. But that raises an interesting question: what does that mean for someone who hasn't had the years of experience to develop that judgment yet?

I think there's an opportunity here, but it requires approaching AI a little differently than simply asking it to write code.

Treat AI Like Unlimited Access to a Senior Engineer

One of the biggest advantages of AI is that you effectively have access to someone who can answer questions about almost anything, at any time. Imagine having a senior engineer sitting next to you who never gets tired of questions, never minds explaining something for the third time, and is always available to walk through an unfamiliar concept. That's a pretty remarkable resource.

My advice to anyone earlier in their engineering career would be to take advantage of that. Don't just ask AI to implement something and move on. Ask why it chose that approach. Ask what alternatives it considered. Ask about the tradeoffs, the assumptions, and the things that could go wrong. If there's a line of code you don't understand, ask about it. If an architectural decision feels unfamiliar, dig into it. If an answer leaves you with even a little uncertainty, keep asking questions.

I find myself doing this all the time, even after years of engineering experience. Sometimes I'll ask AI to explain something like I'm five years old. Not necessarily because the concept is beyond me, but because I want to make sure we have the same mental model before moving forward. It's a useful way to strip away the jargon and get to the underlying idea. If the simplified explanation doesn't make sense, there's probably something I haven't fully understood yet.

There's an important distinction between getting AI to produce a working solution and using AI to help you understand that solution. The first can make you productive. The second makes you a better engineer. Don't be afraid to slow down and ask questions. You're not wasting the opportunity by asking AI to explain something you could have just accepted and moved on from. You're getting exactly the kind of experience that helps build engineering intuition.

Don't Confuse a Working Solution With Understanding

This is where I think junior engineers need to be particularly intentional. AI is very good at producing answers that look convincing. It can generate a clean implementation, explain why it works, and even make reasonable arguments for its design decisions. But it can also be confidently wrong, introduce unnecessary complexity, or make assumptions that aren't appropriate for your particular situation.

An experienced engineer has years of accumulated context to draw from when evaluating those answers. They recognize patterns, notice inconsistencies, and have a better sense of which questions to ask. A junior engineer may not have developed those instincts yet, and that's perfectly normal. The danger is allowing AI to fill that gap without actually developing the understanding that should come with experience.

I don't think the answer is to avoid using AI or to insist on doing everything manually. That would be missing the point. Instead, use AI as an opportunity to accelerate the development of those instincts. Ask it to critique its own implementation. Ask what an experienced engineer might question about the approach. Have it walk through potential failure scenarios. Compare different solutions and understand why one might be preferable to another.

And when you still aren't sure, that's okay. Recognizing what you don't understand is an important engineering skill in itself. The goal isn't to become someone who can accept AI-generated code without hesitation. It's to become someone who can evaluate it, question it, and make informed decisions about whether it belongs in the system you're building.

Take Your Learning Public

AI is an incredible learning partner, but it shouldn't be your only source of feedback. It has access to an enormous amount of information, but it doesn't necessarily know the context you're missing or recognize every assumption you've made along the way. Sometimes the best feedback comes from another engineer who has approached the same problem from an entirely different direction.

This is why I think writing and sharing your work publicly is one of the most effective ways to accelerate your growth.

You don't need to wait until you've mastered a technology or become an expert in a particular area before sharing what you've learned. In fact, I think there's something particularly valuable about documenting the process of figuring something out. Write about the experiment you tried, the problem you encountered, the decisions you made, and what surprised you along the way. Explain what you understand and, just as importantly, what you're still trying to figure out.

Putting your thinking in public creates an opportunity for other engineers to engage with it. Someone might recognize a mistake you didn't catch, suggest an alternative you hadn't considered, or share an experience that completely changes how you approach the problem. Those interactions can be incredibly valuable, especially when they come from people with years of experience solving problems you haven't encountered yet.

There's also something about writing that forces you to organize your own thinking. It's relatively easy to follow along with an AI explanation and feel like you've understood something. It's a different exercise to explain that concept in your own words, describe why you made a particular decision, and defend the tradeoffs. If you can't explain what you built, that's probably a good indication that there's more to learn.

And you don't need to turn every experiment into a polished tutorial. Share the exploration. Share the questions. Share the things that didn't work. The engineering community benefits from seeing the process, not just the finished product.

Learning Is a Community Activity

I think there's a tendency to view learning as something that happens privately, before we're qualified to share our knowledge with others. We study, experiment, build experience, and eventually feel comfortable enough to contribute. But that creates an artificial separation between learning and participating in the engineering community.

Some of the most valuable learning opportunities come from sharing unfinished ideas and inviting others into the process. You don't have to be the most experienced person in the room to have something worth contributing. You just have to be honest about what you've done, what you've learned, and where your understanding is still developing.

This is especially relevant in the current AI boom. As AI makes it easier to build things, the ability to explain what you've built, understand the decisions behind it, and engage with feedback becomes increasingly important. Your portfolio doesn't have to be a collection of impressive finished projects. It can be a record of your curiosity, your experiments, and how your thinking has evolved over time.

And for those of us who are more experienced, there's an opportunity here too. When someone shares their work publicly, take the time to engage. Offer feedback. Explain why you might approach something differently. Encourage the questions. We all benefit from a community where people are comfortable sharing what they're learning, rather than feeling like they need to have all the answers before participating.

Start Before You Feel Ready

I don't think AI eliminates the need for junior engineers to learn the fundamentals, and I certainly don't think it eliminates the value of mentorship and experience. If anything, those things become more important as the amount of code we can generate increases. But the path toward developing that experience doesn't have to look exactly like it did before.

You have access to tools that can help you explore almost any area of software engineering. You can ask an endless number of questions, experiment with unfamiliar technologies, and build things that stretch your abilities. And by sharing that journey publicly, you can invite an entire community of engineers to help you get better.

Treat AI like the senior engineer you always wished you had sitting next to you. Ask questions until you understand. Don't blindly trust the answers, and don't be afraid to admit what you don't know. Then share what you're learning and give other engineers the opportunity to challenge your thinking.

The fastest way to get better isn't to have AI do more of the work for you. It's to use AI to learn faster, and to let the people around you help you learn better.

Don't wait until you feel like an expert to start sharing what you know. The process of getting there is worth sharing, too.