I can see what I’m gaining: time, convenience, and new ways to understand complicated problems. What’s harder to see is how relying on it is changing me. Am I developing new abilities, getting more comfortable handing things over, or both?
And even where I am growing, am I keeping pace?
The technology keeps evolving, and so does the way I use it. Each new capability changes what I’m willing to delegate—and what I need to understand to use it well. I wonder whether my judgment, my habits, and my ability to check the results are developing as quickly as my reliance on the technology.
Those questions became personal for me in the driver’s seat.
Over the past couple of months, I’ve used Tesla’s Full Self-Driving (Supervised) for essentially all of my driving. I’ve come to prefer having it handle the driving tasks. That preference surprised me.
But something else surprised me more. On the occasions when I’ve taken control, I’ve felt a little rusty. There’s been some hesitation. I’ve caught myself feeling more confident in the system than in myself.
That’s an uncomfortable thing to admit.
It isn’t proof that my driving ability has declined. But it is a change in how I feel about a skill I’ve relied on for years, and I don’t want to dismiss it just because the technology is convenient.
At the same time, I feel more practiced at monitoring the system. I find myself looking around more and noticing what’s happening around the car. There may be a skill developing there, too.
Whether that makes me a better supervisor—or simply a more comfortable user—is a question I haven’t answered. FSD is supervised driving assistance; I still need to remain attentive and ready to take control. My feeling of greater awareness doesn’t establish that I’m safer or better prepared to intervene. Tesla’s owner’s manual
That distinction has stayed with me as I think about how I use AI professionally.
In my work leading app lifecycle engineering and agentic AI efforts, my team handles case tasks that come in from the field. Understanding a case can involve sorting through details, reconstructing what happened, and developing hypotheses about the root cause, possible mitigations, and longer-term solutions.
AI helps me work through that material. It can summarize a case and propose explanations that give me a starting point for investigation. In some ways, I feel that it’s accelerating my understanding of the underlying technology.
I spend less time on the mechanics of assembling information. I can use that time to ask deeper questions: Why would this component behave that way? How does it interact with the rest of the system? What am I missing?
When I don’t understand something, I can stay with it and ask for another explanation.
That feels like a meaningful gain.
Then, while talking through this article, I was asked a question: How do I decide whether the AI’s root-cause hypothesis actually holds up?
I didn’t have a clear answer.
I don’t take that to mean every explanation I’ve accepted was wrong. But I also can’t treat my growing sense of understanding as evidence that the explanations were right. An explanation can make sense to me before I’ve checked whether it fits the facts.
That leaves me with a more uncomfortable version of my opening question:
Is my confidence growing faster than my competence?
The answer may be different for different skills. I may be learning more about a system while getting less practice investigating it from scratch. I may be improving at asking questions while leaving my methods for checking the answers underdeveloped.
And the balance keeps changing. As the technology becomes more capable, I find more uses for it. Those uses change what I practice, what I expect, and what I trust. My experience with it then shapes what I’m willing to try next.
I’m part of that evolution. But participating in it doesn’t necessarily mean I’m keeping up with it.
Some benefits are immediately visible. Other changes may only become apparent when an explanation fails, an unfamiliar problem appears, or I need to do the work myself again. By then, I may have grown accustomed to a level of confidence I haven’t learned how to test.
I don’t have a settled answer, but I do have something to try: make verification a more deliberate part of how I work. Treat an explanation as a hypothesis. Ask what supports it and what would contradict it, then check a meaningful claim against the case details, logs, or code. Asking AI to challenge its own answer can help me explore it, but I still need evidence beyond that conversation.
That practice won’t answer every question about how my skills are changing. It might help me distinguish feeling informed from having good reasons to be confident—and show me where I need to keep learning.
I want to keep the benefits I’m seeing. As the technology evolves, I want to pay closer attention to how I’m evolving with it: what I’m learning, what I’m no longer practicing, and which abilities I still need to keep sharp.
Especially when I’m the person expected to take over.