When the chips are down, thank goodness for software engineers: AI algorithms 'outpace Moore's law'

Machine-learning algorithms are improving in performance at a rate faster than that of the underlying computer chips, we're told. AI software techniques have become so efficient, in fact, engineers can now train a neural network on ImageNet – a top dataset for image recognition systems – to about 79.1 per cent accuracy with 44 …

[FF22]

Nonsense

Moore's law is not about AI, and because it can not be outpaced by AI development. It's also not like AI's are getting more effective at using raw computing power.

[Schultz]

Learning curve

I would claim that there is a significant difference between the learning curve described here and the engineering capabilities described by Moore's law. Learning curves can be expected to start steep and fall off rapidly, whereas Moore's law remained steady for decades.

Did we reach maximum AI hype (haip?) yet? You know it's around the corner.

[RM Myers]

Re: Learning curve

Nope, not even close. Al hype has definitely outpaced Moore's law, and the rate of growth is still increasing.

[Anonymous Coward]

"about 79.1 per cent accuracy"

I'm sure that will stand up well as proof in a Court of Law.

[steelpillow]

rememris the memberstor?

If memristor-type devices ever actually take off, we may see a new Moor-ish law for neural net hardware kick into life.