For some reason, when I was thinking about AI, I forgot that software changes. I kept thinking, what if, beside being text prediction, it got the ability to map the APIs, and configurations, etc. of all the tools we use, to natural language... it would improve dramatically. But even if they were able to do this... they will have to update the mapping somehow with every change to all of these. So, if Docker, for example, made a breaking change to their dockerfile configurations... until the models pick it up, some containers will fail, and it will require someone who understands the configurations to fix it. So it is like chasing a moving target... there is no point AI can reach to be on that level that people fear, in the first place. There is nothing to fear. This settles the whole thing for me. Unless they figured out a way to auto-re-map all the changes of every single tool that AI users use. (which I think is impossible to do) This whole thing will increase our value, not the opposite.