Hacker News Reader

Comment by jiggawatts | Hacker News Reader

Did anyone else notice that the AI section is now bigger than the traditional CPU cores?

What applications benefit from this? The only use-case I've seen for AI in Apple's mobile devices is the face unlock.

>What applications benefit from this? The only use-case I've seen for AI in Apple's mobile devices is the face unlock.

Oh no. You could have double the CPU core and it wouldn't make 95% of common use case any faster. But NPU ( Neural Processing Engine ) and GPGPU have lots of potentials. Siri being the first one, When you take photos its uses NPU to adjust the image. Your Speech recognition for lots of different languages. Handwriting, Your Photo Gallery Face Recognition, on iPad Handwriting / Image to Text Recognition. And lots of thing likely to do with AR. And all of them has great impact to user experience previously not available.

Not to mention both GPU and NPU scales very well with transistor budget.

Apple does all machine learning on-device. So stuff like handwriting recognition, Siri, automatic improvements of photos etc. makes heavy use of it.

All inference on-device you mean? I presume that the vast majority of their models are trained in the cloud. Apart from some fine-tuning for personalization maybe.

The recognition of faces and settings of images for example is fully done on the phone. But that is done overnight when the phone is plugged in.

While they did say machine learning, all the examples are of inference, and not training. So you don't have to ask if they meant inference when it is very clear what they mean.

Outside of marketing speak, it is not really machine learning if the machine is not learning.

I wouldn't expect "this phone can run for 10 hours and train a model", even though it clearly is also possible.

AI creates possibilities of creating meaningful signal from extremely noisy environment, enabling lots of miniaturisation in sensors, which again enables tons of cross domain applications. e.g. low cost spectrometers for blood testing.

Speech recognition, image recognition, photo/video enhancement, face recognition, predictive text, etc. There aren't a huge number of applications, but there are enough to justify it, especially because it makes a big difference where it can be used.

Homekit cameras support face recognition. I assume this runs on ”home hub”, either your Apple TV or HomePod.

Apple is planning on releasing AR headset, isn't it?

Reply on news.ycombinator.com