Not long ago, I found myself staring at a problem that should have been familiar territory. I was working on one of my applications, trying to solve a bug that wasn't particularly complicated. It wasn't the sort of issue that would take days to figure out, nor was it something I hadn't encountered before. A few years ago, I would have opened documentation, started tracing through code, experiment with a handful of possible solutions, and eventually worked my way toward an answer.
Instead, I opened an AI chat window.
Within seconds, I had a response. The explanation looked reasonable. The proposed solution seemed sound. I copied part of it, adjusted a few things to fit my project, and moved on with my day.
For a while, I considered that a win. After all, the problem had been solved faster than it otherwise would have been. Time had been saved, progress had been made, and the application continued moving forward. That's supposed to be the point of technology, isn't it? We build tools to make things easier.
The more I thought about it, however, the less satisfied I became with that answer. What bothered me wasn't that I had used AI. I use plenty of tools that make my life easier. Every developer does. What bothered me was realizing that I had never truly engaged with the problem. I had received a solution before I had fully understood the question. Had I actually learned anything from the experience? If I encountered the same issue again six months later, would I recognize it immediately, or would I once again reach for the same tool and ask for the same answer?
That question has stayed with me because I don't think it applies only to software development. I think it applies to almost every field that AI is beginning to touch. Students use AI to write essays. Developers use AI to generate code. Marketers use AI to draft campaigns. Designers use AI to create concepts. Researchers use AI to summarize articles. Managers use AI to write emails. Writers use AI to brainstorm ideas. The technology is spreading into nearly every profession at a speed that would have seemed impossible just a few years ago.
What makes the conversation difficult is that the technology genuinely is useful. This isn't one of those situations where a new tool appears and people spend years arguing about whether it solves a real problem. AI clearly solves real problems. It can save time. It can automate repetitive work. It can lower barriers to entry. It can help people accomplish tasks that would have otherwise required specialized knowledge. Those benefits are real.
The concern I keep coming back to has less to do with what AI enables and more to do with what it replaces. When I was younger, learning something usually meant struggling with it for a while. If you wanted to understand networking, you broke things and figured out how to repair them. If you wanted to learn programming, you spent hours staring at error messages and trying to understand why your code wasn't working. If you wanted to become a better writer, you wrote bad drafts, then slightly less bad drafts, and eventually something started to click. The process wasn't always enjoyable. In many cases, it was frustrating. Yet the frustration served a purpose. Every mistake taught something. Every dead end eliminated a possibility. Every wrong answer narrowed the path toward the right one.
The struggle itself was part of the education. That is what makes me uneasy about a future where AI becomes the first step rather than the final step. There is a meaningful difference between using AI after you've attempted to solve a problem yourself and using AI instead of attempting to solve the problem yourself. One approach supplements your understanding. The other risks replacing it. The distinction may seem minor today because the outputs are often identical. A student still submits a paper. A developer still ships a feature. A marketer still publishes an article. Looking only at the final result, it may appear that nothing has changed. What becomes harder to measure is what happened inside the person creating that work.
Did they develop the skill?
Did they improve their understanding?
Did they strengthen their ability to solve similar problems in the future?
Or did they simply receive an answer and move on?
The longer I spend thinking about this, the more I realize that many of the most valuable things I've learned came from situations where no immediate answer was available. Some of the most important lessons in my career emerged from production outages, failed deployments, broken systems, and technical problems that refused to cooperate. At the time, those situations felt like obstacles. Looking back, they were some of the best learning opportunities I ever had. There is something different about discovering an answer yourself. When you spend hours working through a problem, you don't just learn the solution. You learn why alternative solutions failed. You learn how the system behaves under different conditions. You develop intuition. Over time, that intuition becomes one of the most valuable assets you possess. Expertise has always been more than a collection of answers. Expertise is pattern recognition. It is the ability to look at a new problem and recognize similarities to something you've seen before. It is the ability to identify likely causes, eliminate unlikely explanations, and navigate uncertainty without needing someone else to provide a roadmap.
I worry that as AI becomes more capable, we may gradually stop developing those instincts. The irony is that the technology is often most effective in areas where expertise already exists. An experienced developer can use AI to accelerate work because they understand enough to evaluate the output. An experienced writer can use AI to brainstorm because they understand what makes writing effective. An experienced researcher can use AI to summarize information because they already know how to distinguish a reliable source from an unreliable one. The foundation still matters. Without the foundation, the output becomes much harder to evaluate. That concern extends beyond learning and into creativity as well. Many people assume creativity begins with inspiration. In reality, creativity often begins with uncertainty. You start with an incomplete idea and slowly explore it. You try different approaches. You reject bad concepts. You follow unexpected directions. Somewhere in that process, something interesting emerges. That journey is rarely efficient. In fact, efficiency is often the enemy of creativity. Some of the best ideas I've ever had appeared while working through a problem that initially seemed unrelated. They emerged from detours, failed experiments, and abandoned approaches. If I had been handed the final answer immediately, I never would have traveled the path that led to those discoveries.
AI excels at generating answers quickly. What it cannot do is replicate the experience of arriving at those answers through your own exploration. That distinction may not matter if your only goal is productivity. It matters quite a bit if your goal is growth. The conversation becomes even more interesting when viewed through the lens of dependence. One of the reasons I care so much about local-first software and software ownership is that dependence always introduces risk. Whenever a person or organization becomes dependent on a system they do not control, they inherit vulnerabilities they cannot fully mitigate. We've already seen this happen with cloud software. Entire businesses rely on services they don't own. Entire workflows depend on platforms they cannot control. Data, communication, and operations often exist entirely within ecosystems controlled by someone else.
Most of the time, that arrangement works fine. Then an outage occurs. Or pricing changes. Or a feature disappears. Or a service shuts down. Suddenly the dependency becomes visible. I suspect AI will create similar dependencies over time. Organizations will build processes around it. Employees will structure workflows around it. Entire teams may eventually lose the ability to perform certain tasks efficiently without it. At first, that dependency will seem harmless because the technology will continue working. The risk only becomes obvious when circumstances change. Imagine a generation of developers who rarely debug complex issues without AI assistance. Imagine a generation of analysts who consume summaries instead of primary sources. Imagine a generation of writers who struggle to produce first drafts without external guidance. None of those outcomes are inevitable. They are simply possibilities worth considering.
Technology has always involved tradeoffs. Every innovation solves certain problems while creating new challenges. Cars made transportation easier, but reduced physical activity. GPS made navigation easier, but weakened our ability to navigate independently. Search engines made information easier to access, but reduced the need to memorize facts. AI follows the same pattern. The question isn't whether AI is good or bad. The question is what we gain and what we lose. I suspect we'll gain extraordinary levels of productivity. Tasks that once required hours will take minutes. Workflows will become more efficient. Information will become more accessible. Those benefits will be real, measurable, and significant. At the same time, I think we should be careful about preserving the skills that made those advances possible in the first place.
Critical thinking still matters. Problem solving still matters. Writing still matters. Creativity still matters. Learning still matters. The ability to sit with a difficult problem and work through it remains valuable, even if a machine can provide an answer more quickly. Perhaps that is the thought I keep returning to whenever this subject comes up. I don't believe AI is something to fear. I don't believe people should avoid it. In many ways, I think it will become one of the most useful technologies ever created. What I do believe is that convenience has a cost, and that cost isn't always obvious in the moment.
Every time we choose a shortcut, we save time. Sometimes that's exactly the right decision. Other times, we unknowingly skip the part of the process that would have taught us something valuable. The challenge moving forward won't be deciding whether to use AI. Most people will use it, and for good reason. The challenge will be remembering when not to. Because once we become dependent on a tool for our ability to reason, create, learn, or solve problems, we begin giving away something more valuable than time. We begin giving away the very skills that allowed us to build those tools in the first place.