The first one is becoming a bit of a buzz word, sometimes with great(excessive) expectation not in line with what the actual thing is (statistical algorithms fed with a very huge amount of data in order to build a model consisting of serialized data saved on disk and serving as filter to classify input or generate new output, like picture (midjourney), text(chatGPT), or actions (brake for self-diving car) from input).

Quantum computing seems to progress quickly, but according to comments from an article published yesterday, it's still prone to error which makes it unreliable for actual commercial use and is more used for training for future quantum computing software dev.

The problem is that the ones reporting on these things, for example on youtube, don't always understand how or what these technologies do and have excessive expectations about them. The other problem is people having a ball in the game (scientists, startup CEO, investors, paid commenters looking for clicks) do the same in order to have a financial gain.