People who argue that AI isn't "real" or isn't "there yet" have for years pointed to chess, then Go, then whatever the next thing is that computers can't do. What they always ignore is that even the first iterations can beat most people. So they've been acting as though it's not "real AI" until it's better than the best human.

Now it seems we're finally starting to realize that most people couldn't meet the criteria that we've set for AI.

Is flipping burgers a "bullshit job"? They've got robots that can do it, but there are still people making a living at it. Lots of writing is the white collar version of flipping burgers.

I think Ken Jennings was prescient when they asked him how he felt about losing to Watson. "For the rematch, we should try dancing." I've seen lots of people suggesting that deep learning can only go so far with machines not having access to physical senses, but no examples yet of anyone trying it. When they do, then things might get interesting.