> Forty-three percent had used one for work That's honestly pretty low considering the sheer number of products that now have AI integration. I honestly think

> Forty-three percent had used one for work That's honestly pretty low considering the sheer number of products that now have AI integration. I honestly think a lot of the hype around AI taking over jobs is because the people who are spreading it are more likely working in some pretty dull coding environments where copilot can shine, because you just plain don't write a ton of interesting or otherwise off-the-beaten-path code, and that's what copilot excels at. And then those people have spare time at work to evangelize about this great technology. And before anybody starts with me, yes I have tried it. It's fine. A lot of what I work on isn't standardized or boilerplate enough to where copilot can really help me, and the situations where it can, I found the time I spent describing what I wanted, getting the answer back, copying it into my IDE and then customizing it as required was frankly just better spent writing it myself. Maybe it made me SLIGHTLY faster? But nothing approaching what I would suggest the term "coding AI" implies, and frankly, I enjoy writing the easier code on occasion because it's a nice break from the harder stuff that copilot can't touch. Like if you're a freelancer who jumps between gigs of refactoring ancient websites into something vaguely web 2.0, you would probably get a lot of mileage from copilot where you're just describing something as written (or hell, giving it the existing code) and asking for it to be rewritten. But if you're doing something novel, something that hasn't been posted on StackOverflow a thousand times, you will run into it's limitations quite quickly and, if you're like me anyway, resign it to the bin because fundamentally asking it to make something, finding out it can't, and then making it myself is FAR more annoying than just assuming it can't and moving on.