Should we ban AI coding for specific industries?
With all the excitement and hype around AI coding, there's a thought that bothers me.
AI makes mistakes. A lot. Some goes into production.
Now, don't get me wrong - I'm building my career and a successful company on AI coding agents.
But building micro-apps and "vibe-coding" fun apps that's one thing.
What about aviation? medical tech? critical systems? What if AI pushes a bugfix to the aircraft control system, and the airplane crashes?
Should we ban AI coding in specific industries? I'm talking about tools like Cursor or Fine - AI agents, not auto-complete copilots that are easy to review.
Replies
Fine
It's all about maturity and supervision. I hope no one finishes a 6-week bootcamp and their first job is pushing changes to ATC, especially not unsupervised. If they have proper supervision and review and once they're more mature.... should be ok
I think AI coding in certain industries will probably be banned only after an incident happens..
Interesting, how do you see regulation evolving for this?
I don't think the problem is with ai coding itself and more so with the devs not taking the time to qa. In my experience while AI does get the job done it often leaves behind unused code, unused files etc and while it technically works if you keep pushing all this unused codes and files it's going to hurt your site in the long run. Also sometimes it overcomplicates simple tasks. So my take is always QA you're already saving a ton of time by not having to code it yourself might as well take a little time to review it and also understand what it did in case you have to fix it.
Also DON'T LEAVE BEHIND ai comments we know this is 2+2 and we don't need "changed variable name to ___" as comments or worst "user asked for___"
PS. This is also my advice for all if you use my personal ai coding automation tool Snippfy always review no matter how hard I try to optimize it AI can still have hallucinations, enhance your workflow, don't replace it!
Same sentiments. AI is meant to speed things up, but mistakes can slip through, especially in high-risk industries. Quality assurance and human oversight are still a must to make sure everything’s safe.
I feel AI coding should only be used for demo apps and that's it. It should only be used for creating components easily but also it will be a long time before they are able to fully build 100% useful and high quality applications. But yeah that's my take
Don't decision-makers (human beeings) make mistakes too? Any system should be designed to minimize the chance of errors. In my opinion, such a restriction for AI doesn't seem appropriate.