Eye eye! AI could stop blindness, Facebook's after math, and how to get started in the ML biz Hello, here is a very quick roundup of this week's AI goodies you may have missed. Floating point maths for AI chips: Facebook has published code that improves the efficiency of number crunching to train and deploy neural networks using AI chips. There’s a ton of matrix maths that’s performed when you feed a neural network …
More "AI" lies. It's AUTOMATION. In many cases, badly configured AUTOMATION. IBM is the king of that. WATSON is the biggest lie of all those AI lies. Slowest piece of crap ever, that IBM Technical Support Staff were encouraged to use. And who really wants anything to do with Facebook AI, accept them? Do you trust Facebook AI? Good luck with that.
It's not automation, is just curve fitting. All an AI does is fit a curve to data points, and you kinda hope that the surface you end up with interpolates well for any other data you want to throw at it.
It suffers from the same problems as other curve fitting techniques with the added bonus that plotting and interpreting the surface you've ended up with is fiendishly tricky.
This probably raises horrified hackles in the AI bullshit brigade, but automatic diagnosis is if anything an "expert system", not "artifical intelligence".
Automatic. Like a washing machine, or a gun.
Just a machine, whose behaviour is the responsibility of its designers and users.
Why do I get the feeling that one of the reaons "AI" is such a hype-fest is that it offers the opportunity to avoid responsibility? In terms of both liability and morality.
If this sounds far-fetched, consider that we live in a world where the rights of corporate "bodies" often exceed those of mere human meat sacks, sometimes even with explicit legislation to that effect.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
