Using AI is great until it's not

I've been making use of AI tools for my main client and it is amazing. You can clearly describe what you want, what something should look like, how it should work and the AI model will soon spit out for you the feature you want.

You can see why people think that AI can replace experienced developers, all you need to do is tell the AI and it'll build your app for you. Well that's the idea.

What I'm actually finding is that the when the AI tool has created a whole feature for you, you are left with a ton of code that no one has written or understands. What happens with the long term maintenance of this code? If no one has written the code or knows all the working parts of the application. So I needed a new approach to how I'm using AI.

AI can to lead to quality issues

Generally as a developer using AI I can see some benefits for it, but if I was just a product owner who thinks replacing my whole team with AI is the way forward then I think that is a terrible mistake, as the car company Ford have started to find out they are now re-hiring engineers to quality inspectors as the AI tools they used to replace these engineers has failed to match quality checks. As the vice president of vehicle engineering said 'AI is a fantastic tool, but it's only as good as the information you use to train it'.

Going forward with AI development

This is an extremely good point and something I think will happen in software development. While AI tooling is great, the code it generates does have flaws and without that human engineer around to review, and maintain this code soon AI only projects will start to have quality issues.

So product owners keep your engineers around or at least know a good engineer you can call on who can help with the issues AI has caused that you don't know how to resolve.