To the AI skeptics

I get why people are skeptical about AI.

It’s hard not to be. The conversation around it has become noisy in a way that makes it difficult to tell what’s actually useful and what’s just marketing. Everything is suddenly “AI-powered”. Every product update is framed as the next big thing. There’s constant talk that this is the moment, and that if you’re not fully invested, you’re already behind.

I don’t think that’s right.

A lot of the hype is overblown. AI is often treated as the answer before there’s a clear question. In some cases it feels like “AI” is doing more work in the company than in the product itself. That alone is enough to put people off.

On the other hand, I don’t think the reaction of writing it off completely really holds up either. Not if you actually spend time using some of these tools.

AI is here to stay. Not because it’s flawless, and not because it replaces thinking, but because parts of it are genuinely useful when you’re careful about how you use them.

I don’t see AI as a replacement for software engineers. I don’t even see it as something especially dramatic. It feels closer to another shift in tooling. Something that changes how quickly you can move, not whether you need to understand what you’re building.

Used badly, it’s an absolute mess. You move faster without clarity. You generate code you don’t really understand. You end up debugging behaviour you didn’t design. That kind of speed isn’t helpful.

Used well though, it removes friction.

Tools like Codex have been genuinely useful for me. Mostly when I want to get something running quickly. Prototypes. Experiments. Trying something without committing to it. Skipping the kind of setup work that doesn’t really add much value.

It’s not writing software for me. But it does help me get from an idea to something tangible much faster.

That changes how I experiment, and that’s the biggest benefit I’ve noticed.

Not every AI tool earns its place. Some are unreliable enough that they break your flow more than they help (looking at you, GitHub Copilot). Others feel like they exist just because they’re expected to exist.

What is unhelpful is turning this into an all-or-nothing situation.

AI isn’t going to replace software engineers. But ignoring it entirely doesn’t feel right either. Writing good software still comes down to judgment, context, and taste. AI doesn’t remove that. If anything, it makes it more obvious when those things are missing.

The faster you can generate code, the more important it is to be clear about why you’re generating it in the first place.

So yes, be skeptical. I think that’s healthy. I just don’t think skepticism has to mean dismissal.

Most of the time, this industry doesn’t move forward through big, dramatic shifts. It moves through smaller changes in how we work. AI feels much closer to that than to the moment it’s often presented as.

And honestly, that’s fine.