Love using AI (at work)
It’s great to be able to prompt for code creation. Does it massively flop? Yes. Would I spend less time figuring out these issues myself? Of course!
For instance I am currently having an agent upgrade a repo from commonJS to ESM. The agent is massively assuming stuff that causes MORE issues that I have the agent try to fix. Honestly I could do this work on my own faster as I have been working with this repo for years. It’s fun to see AI make the work longer because I can report back during standup. “Upgrading to ESM, AI agent ran into some issues that I am currently debugging”
Which is the truth and everyone will smile and node because I am using AI which everyone at my company is instructed to do. A lot of times it can be inefficient but I don’t care.
For personal projects, I watch every code change with scrutiny. I rarely run agents without a solid plan and code examples to approve as I truly care about my projects.