Microsoft adding RAR, 7z, Gz and more to the native ZIP extractor, and finally having it use more than 1 CPU core.

On the one hand, we've found github copilot (and a variant of it in use at $MajorTechCompany to be invaluable tools, so much so that we've actually gone and purchased a monthly subscription to copilot. We've also used ChatGPT to write functional applications, albeit of a trivial variety (since that was what we needed at the time), and it works, more or less. On the other hand, both of those are wrong somewhere between 60% and 80% of the time. We have to re-feed the code back into ChatGPT close to a hundred times to get it to finally produce the thing we wanted. And copilot often provides not only nonsensical, but also non-syntactically-correct responses. They're incredible assistants to a programmer. But neither of them is yet any good at understanding exactly what you're trying to do and providing a good solution. On the other hand, we've found that they have two major benefits. The first is that trying to explain what went wrong with the code (assuming it's not a basic syntax error) is effectively [rubber duck debugging](https://en.wikipedia.org/wiki/Rubber_duck_debugging), which in-and-of-itself is already known to be incredibly helpful. The second is that ChatGPT is _really_ good at spotting where the error is occurring if does actually understand your problem (read: if you've sufficiently described it to ChatGPT). The solutions it provides aren't always great, but it's a great help in terms of "Where do we focus on next?" when iteratively programming/debugging. The point of this is that these new tools are incredible assistants to an already well trained programmer/sysadmin. But you're 100% correct that manglement will see these as cost-cutting measures and decide to fire/let go a significant portion of their engineers because they think they can replace them with AI.