As AI becomes increasingly capable of building software, I find myself wondering how the craft of development might change. Reviewing code, catching mistakes, and intervening when things go wrong will remain important. But what excites me most about programming is the opportunity to create things. Not just maintain things. Make great things. Fun things. Beautiful things. Complex things.
So, as AI takes on more of the implementation work, what happens to creativity? And does developing expertise still matter?
A Thought Experiment
I want to approach these questions by starting with a thought experiment. Imagine a world where we no longer have to look at code. We don’t worry about databases, graphics pipelines, networking, etc. Let’s say that all the implementation stuff is abstracted away. What’s left for a developer to work on? What does the craft become in this scenario?
Say we have a minimal AI-driven game engine with almost no UI. Just text input for prompts and a display window. You type in a prompt, and get a generated scene in the window. In this example AI handles graphics, physics, gameplay logic and everything else under the hood. So, you tell it to spit out a game. It does and you’ve got a working game. Now what? Is it a game you like? Maybe it could be different. So, you prompt away, asking for changes to be made. You’re changing level design, art direction, and gameplay loop.
Before you know it, the game you have is different than the one that was generated. AI did the lifting, you gave the direction (Great, I’m starting to sound like a LinkedIn Influencer now). But, stay with me for a second. Who created the first game, and who created the second? AI did the technical work in both cases. Yet your judgment, taste, and creative direction shaped what the second game became. Notice how the work became less about implementation and more about deciding what to make and how to make it better.
But if creative direction is enough to make someone a creator, what separates someone with an idea from someone who can meaningfully shape its development – an expert?
The Importance of Expertise
Imagine two developers using our AI game engine from before. Again, neither one needs to write a line of code. Both can generate a functioning game. But one understands game design, player psychology, pacing, and the technical possibilities of the medium. The other doesn’t. Both can create something, but their understanding influences what they notice, what they experiment with, and what they ultimately choose to build.
The more experienced game developer sees possibilities that the other doesn’t even know exist. In other words, what the first dev lacks is skill. To be sure, AI can help to create things beyond one’s abilities, but expertise expands what one can imagine doing with it. Therefore, mastery isn’t irrelevant. It gives you options. It gives you sight.
The same applies to software. An experienced developer might recognize opportunities for a simpler architecture, a more intuitive interface, or an entirely different approach to solving a user’s problem. Their experience doesn’t just help them improve existing solutions. It can change which solutions they consider in the first place.
Of course, you could ask AI what’s wrong with your game or how to improve it. And it might genuinely help. In a 2024 study, researchers found that AI-generated ideas substantially improved the creative output of less creative writers. Interestingly, however, the resulting stories also became more similar to one another. AI improved what some individuals could produce, but narrowed the diversity of what they collectively created.
This makes me wonder whether there’s a difference between having access to creative possibilities and developing the understanding to explore them ourselves. Yes, expertise helps us to recognize good ideas. And it may also shape the questions we ask, the possibilities we consider, and ultimately what we want to create.
The Value of Craftsmanship
This doesn’t mean that all artists, game designers, or even programmers should start using AI. Sure, it helps with doing the hard work so you can focus on steering the product in your desired direction, similar in spirit to how driving an automatic car frees you up to focus more on where you’re going. But even I still drive a stick-shift. Sometimes doing the work is part of the enjoyment. Sometimes working through the details reveals possibilities we wouldn’t have considered otherwise. In fact, research on productive failure suggests that attempting to solve a problem before being shown the solution can lead to deeper understanding and a greater ability to apply what we’ve learned to new problems.
I believe that something similar happens when we work through problems ourselves before turning to AI. The work may or may not get us closer to the correct solution, but it does change what we understand, and potentially what we can imagine creating next.
There’s room for different approaches to creating things. Some people will embrace AI because it allows them to explore ideas and build things they couldn’t have otherwise. Others will find value in doing the work themselves, in the understanding, satisfaction, and unexpected discoveries that come from the process. Neither approach has to diminish the other.
Even if AI eventually handles all the technical work, there’s still a craft in deciding what to create, exploring possibilities, and developing the understanding to recognize what makes something great. And perhaps that’s why expertise remains worth pursuing. Not simply because it allows us to do things AI can’t, but because it changes what we see, what we imagine, and what we want to make.
References
Doshi, A. R., & Hauser, O. P. (2024). Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances, 10(28), eadn5290. https://doi.org/10.1126/sciadv.adn5290
Sinha, T., & Kapur, M. (2021). When problem solving followed by instruction works: Evidence for productive failure. Review of Educational Research, 91(5), 761–798. https://doi.org/10.3102/00346543211019105