[rjh29]

It's useful in the right hands - it can suggest functions/language features you don't know about it, it can do some handy stuff automatically (like reverse a list in place) so you have more mental bandwidth to think about other stuff.

It also produces a ton of horrible, nonsense code. I totally agree that if github's corpus starts to fill with that stuff, the overall quality of github will tank and I wonder how they'll continue to train the model.

[iasay]

That’s really my worry. ML is crap in crap out. If you train it with more crap it’s an exponential decaying curve.

Now I have to gate declining quality marketed as a time saver. It really just moves the cost elsewhere.