The only accurate thing is the crux of the article - that the hype around AI talent is overblown. Everything else is inaccurate. Initial success in RNNs were in handwriting recognition and OCR, and that code is public. It isn't 11 lines.
Worst of all is the last paragraph that takes a giant crap on AI hipsters and postulates the author's current work as far superior. I liked the joke. A toast of the rarest of rare microbrews, that's way cooler than wherever you've ever drunk.
AI has a pretty extensive history of hype. The original AI efforts by IBM involving rules in the 60s were overblown. Markov hidden state was overblown. The original neural networks were overblown (especially the first round). SVMs were overblown. CNNs were hyped beyond what is reasonable.
It's funny that virtually every researcher is very aware of this and yet time and time again the press blows is up. Been going on for 70 years now, which defies belief, but it's the actual truth. And it's certainly not a case of researchers failing to tell the press that it's overblown.
Names of things like 'neural networks', 'artificial intelligence' or even 'quantum computing' gift journalists very sexy, broad brushes with which to draw conclusions that exaggerate the abilities and belie the often mundane realities of the things they describe.
It might have had some impact if he dropped the last paragraph, but that just gave it away as a dick measuring contest and not legitimate criticism. Stop pretending you are a genius because I am!
If you think it's inappropriate for HN, flag it. Submit something better. Upvote submissions you think are good. Help make HN the site you want it to be.
8 comments
[ 293 ms ] story [ 145 ms ] threadWorst of all is the last paragraph that takes a giant crap on AI hipsters and postulates the author's current work as far superior. I liked the joke. A toast of the rarest of rare microbrews, that's way cooler than wherever you've ever drunk.
Cheers
It's funny that virtually every researcher is very aware of this and yet time and time again the press blows is up. Been going on for 70 years now, which defies belief, but it's the actual truth. And it's certainly not a case of researchers failing to tell the press that it's overblown.
I don't think we're about to learn though.