What I like and worry about with AI

I use AI every day, and overall it has made my work better. There is a lot of hype around it and some product demos try too hard, but under all that noise there are tools that help me a lot.

The biggest change is how fast I can explore. When I am not sure how to approach a problem, I can try five different versions in ten minutes and keep the one that works. I get answers right away, and I can find my way around an unfamiliar codebase in a couple of minutes.

It also takes away a lot of boring work. I do not need to write the hundredth config file or boilerplate test by hand anymore, so I have more time and energy for the parts that need real thinking, which are also the parts I enjoy most.

There are still things that worry me. The ethics around training data and resource use are far from solved, and I think about that. In my daily work, the main risk is code quality. When the output looks right, it is easy to accept it without reading it closely, and I have done it too. That is how code ends up committed without a real review, and months later someone has to debug a fragile mess.

Luckily the fix is simple, and it is the one rule I follow: I only use output that I understand and have verified. I read every line and rewrite a lot of it, the same way I would review the work of an intern who is fast and helpful but still needs guidance. As long as I can evaluate what the model gives me, I can catch its mistakes and keep all the speed.

Used this way, AI is great for sketching ideas, moving fast, and getting unstuck, and it makes coding feel lighter and more fun for me. I am curious to see how far it goes.