AI could cost you your critical thinking skills

I heard Jeff Foxworthy talk recently about a concern he has for comedians in the age of AI, and the example stuck with me. He talked about how comedy often comes from memory. You're having a conversation, something reminds you of your childhood, a family member, an awkward moment, or something that happened years ago. That memory turns into a joke because it belongs to you. Now, with AI, you can ask for the joke in seconds.

That made me wonder what happens when we start doing the same thing in our own work. I use AI every day. I design with it, I write with it, and I vibe code with it. I'm constantly asking questions and trying to understand what it can help me do better. But lately I've been asking myself a different question: am I using AI to strengthen my critical thinking, or am I slowly handing my critical thinking over to AI? I noticed the problem most while coding.

When I'm working on something new, AI will sometimes give me a really long explanation of what it wants to do. It might walk through the architecture, explain the pattern it found, point out what files need to change, talk about what could break, and explain why it recommends one approach over another. Those responses can get pretty long.

A fellow designer once suggested that I just ask AI to condense everything into the Cliff Notes. And sure, I could do that. Sometimes a summary is useful. But I started thinking about what I might be giving up if I always ask for the shorter version.

When I read the full response, I understand the problem better. I start noticing assumptions I want to question. I see where the logic might be weak. I learn why one approach might be better than another. I'm not just waiting for AI to finish. I'm actually learning something.

That's one of the reasons I've started looking at vibe coding like a masterclass in something I don't know yet. I may not understand the technology, pattern, or implementation at the start, but I want to know more about it by the time I'm done. The red flag for me shows up when I get lazy. If I'm tired, rushing, or simply not willing to read the full response, I'm much more likely to skim it, decide it sounds reasonable, and tell AI to move forward. Most of the time, it will still do a pretty good job.

But "pretty good" is not what I want.

AI should be my companion, not my thought leader.

I still need to decide whether the answer makes sense. I still need to ask whether the result is actually what I intended. I still need to notice what is missing and bring the experiences, taste, judgment, and context that are mine. That part of the work still belongs to me.

AI has become incredibly good at giving us something usable, especially when we give it enough context from the start. The first result can be surprisingly close. And I think that is exactly where the risk starts to show up. Close can make us complacent. It becomes very easy to look at something that feels 80 or 90 percent there and decide that it's good enough. I've caught myself doing that. I don't want to work that way.

I want to challenge the first answer. I want to ask why it chose that approach. I want to know what assumptions it is making and what it may have left out. I want to ask whether there is another way to solve the problem. When I'm working in code, I want to know whether AI actually used the component from the design system or simply recreated something that looks close enough. When I'm writing, I want to ask whether the words actually sound like me. When I'm designing, I want to know whether I would have made the same decision without AI suggesting it first.

Those questions matter because I still want to be present in the work.

Jeff Foxworthy's concern about comedy keeps coming back to me here. The joke matters, but where the joke came from matters too. Our experiences matter. Our memories matter. Our judgment matters. Our taste matters. AI can help us explore all of that, but I don't want it to replace it.

Don't just accept the output. Understand it well enough to disagree with it.

For me, that starts with something very simple: read the answer. Especially when I'm starting something new, I want to let AI explain what it is suggesting. I want to understand why it is recommending a certain approach. I want to ask questions when something doesn't make sense and push back when the result feels incomplete. Then I want to look at the final output and ask myself a few simple questions. Is this actually what I intended? What's missing? How could this be better? Does this still feel like my work?

I want AI involved in more of what I do, not less. I see too much value in it to pretend otherwise. But I also want to make sure I'm becoming a better thinker while I use it. That responsibility is still mine.