AI in DevOps isn’t it a SkyNet Problem

I believe there can be no AI in DevOps or CI/CD (Though DevOps is much more than pipeline but accept that a solid CI?CD pipeline is the meat of DevOps )pipe line ever , the reason are very simple
1. You cannot have a trend of defects they ,This is just so wrong assumption. Defects are always unique Imagine we have a new update in one API version and we have a defect, now dev will fix it such a way (ideally) that it will pick up the API version and call the appropriate class of that API. So this trend is nullified. Tomorrow you have a new defect , space issue dev will fix it , so that system check the storage and then migrate, so there is not much chance we have same defect.
We can see all the defects are new and unique in pattern so cannot have a recommendation engine
2. Tech stack changes so fast — you will not have the data set with same tech stack for many years in a given system . The change is so rapid, today jsp tomorrow Node.js so , so you cannot predict a trend. Even you have same Tec stack for years , the system should be super stable with very less defects or error (else fire your team)
3. CI/CD should be boring : The ultimate objective of DevOps is to have a stable system, now that’s a catch 22 , one side you are prime objective is to make your system stable, on the other hand you want your AI engine to get some data ..which is impossible.

Most importantly I have never heard of a system where defects and operational challenges are automatically fixed ( I mean you be proactive to prevent defect and system failure, not wait for AI engine to pick it up and fix it, and if you are proactive your failure case and input variables will be sparse and unique and your AI engine will not be able to pick it up)

Microsoft Excel’s product development team are still quite big, A product being 20 years old why AI is not fixing their issues automatically and developers are chilling around ….