My biggest fear, and one that is not talked about in the article, is that we won’t have any asbestos removers in the future. Generative AI is being fed it’s own excrement and that’s being leveraged as working code to new coders. When this really becomes a liability we won’t have many left that can fix or figure out the fix cause it will have obfuscated all the usable info.

I think it’s going to be a complete shit show. Here’s the confluence of factors:

• From what little I’ve seen of AI code, it seems to write code that’s even more shit than human devs. I’m not in software dev, but I’m in IT operations and also studying CS. What I’ve seen of AI code in IT (PowerShell scripts) looked like it wrote 10+ lines where one or two would have done the job. In other words, it’s a form of obfuscation like you said.
• I strongly expect that fewer people are studying CS because they’re getting the message that AI is taking all the dev jobs. That’s true for the moment.
• Fewer coders are being hired, so there will be even fewer experienced devs in the future.

I think that this will all add up to a “dark ages” of software development in the not too distant future. There just won’t be enough people to fix all the AI junk, and the AI junk will essentially need to be ripped out altogether. Software quality and security will go down the drain, and it will take forever to fix it, if it even gets fixed at all. I think it really will be equivalent to the “dark ages” (I know that this term is not considered accurate nowadays, but I think it applies even more to this situation).

I’m hopeful for one thing though: that this phenomenon will strengthen free open source software relative to commercial software. If there are a bunch of devs who can’t get dev jobs, hopefully they will spend at least some of their time contributing to open source. On top of that, it appears to me that open source projects have been more resistant to accepting AI code. Let’s hope that this is a silver lining here.