Here are some of my thoughts on AI.

One thing AI has clearly done is make writing code cheaper than before. However, unless there’s some kind of major breakthrough, I think the economics will shift and tokens will become more expensive. We’ve seen similar patterns with companies like Uber, Airbnb, and Deliveroo. They start cheap to attract users, then raise prices to become profitable. AI companies will eventually need to pay back the money they’ve raised, so this feels inevitable.

Right now, generating code is cheap and doesn’t require deep expertise to build something. You can see a wave of vibe coded apps emerging. In many cases, it’s easier to build something from scratch with AI than to work within an existing codebase.

I’m starting to question how much of this generated code is actually necessary. In the past, developers were more careful and tried to keep things minimal because writing code had a higher cost. Now it’s different.There’s a lot more code being produced, but I’m not convinced the added complexity generates that much value in the whole app.

From my experience, AI is quite good at following general best practices, especially in areas like web development. But it struggles with other tasks. For example, I asked it to design a simple headphone hanger and output it as an OpenSCAD file. It couldn’t even complete that properly the result didn’t resemble a headphone hanger at all. I don’t think AI is good at breakthrough or creative problem solving yet, at least not from what I’ve seen.

AI is definitely here to stay. Some people compare it to the dot-com bubble, and I think that’s a fair comparison. Financially, there will be winners and losers—not all AI companies will survive. But just like after the dot-com crash, the underlying technology won’t disappear. AI will remain, regardless of how the market shakes out.