I agree with the premise that things are breaking down, but I'm not sold yet that the problem is a systemic one.
At least some of these issues are ones that already existed, but are now being discovered by AI.
If that is the case, we would not see the exponential increase that the article mentions, but should instead see something like a temporary blip where discoveries in existing code spike, but eventually all get discovered and fixed.
I would love to see some more data that specifically looks into the number of issues caused by pre-ai code versus post-ai.
That being said, the comments about decoherence strongly resonate with me. Things are moving so fast in so many different directions right now. I don't think we know the right way to build things in this new future yet.
[dead]