Hacker News Reader

Comment by chmod775 | Hacker News Reader

A classical case of "when all you have is a hammer, everything looks like a nail".

Or rather, you're blind to everything that doesn't require a hammer.

Even in most of the examples he listed (speech/voice recognition software, translation, games, and databases), the vast majority of "code"/logic is not a neural network.

You don't "train" your UI, neural nets won't directly consume audio or spit out encoded audio, don't do HTTP, aren't an operating system and just having a game AI is pretty far from having a finished game.

Neural nets approximate and guess, but for the vast majority of problems in computing we want the exactness of code.

And coincidentally that's what makes neural networks great: They let us solve problems where stating the exact precise steps to solve them is impossible.

View on HN

I think you might have missed the point of the article. In various different ways it says that machine learning is capable of solving many problems better than traditional software designed by a developer. It says that it's akin to a new software programming paradigm or language. From this context, right now we're still all working in Assembly or C, and higher level languages are being developed that will allow us more productivity if we continue to build out the tooling and infrastructure.

I don't think he's suggesting your OS is going to be an ML model, but we will be surprised at how many problems they solve better than trying to do everything in a more manual way. I think this is pretty intuitive. Our programs used to say "Move this byte to this register" but now they say "Toggle this button". Eventually they will just say "Learn to produce results like this" with the same reliance on the developers who build the infrastructure as we have today for our compilers and interpreters.

> It says that it's akin to a new software programming paradigm or language.

That's a great way of thinking about it. It's worth trying to imagine what the programming experience would be like if writing a function amounted to simply defining input/output types, providing a collection of input/output pairs (i.e. train/test sets), and leaving the process of finding the solution / optimization to the "compiler".

> You don't "train" your UI, neural nets won't directly consume audio or spit out encoded audio

I'm not an expert, but I thought pictures and audio were one of the few places where you could feed raw inputs into a neural net and get good results? Or am I wrong and we instead feed in some pre-processed version?

My understanding is that they took in RGB picture data, yes, but that they were required to be square (s.t. matrix transforms work on them, I think?).

But that's still different from taking in the actual JPEG data, which is sort of what the parent gets at: something has to decode that, and that software isn't a neural net.

(Further, when I worked w/ ML that dealt w/ image data, we had a host of non-ML code written around it to support it, dealing with the various facets of running in the cloud, where to get the data, where to store the results, who to notify about the results, and a bunch of preprocessing on the image — such as removing pointless borders that humans put around images.)

That software can be generated and continuously optimized by a neutral net. But that would probably mean that we achieved AGI, so decoding JPEGs won’t be that exciting by itself.

Just FYI, non-square images should work fine. Arguably that's one of the selling points of convolutional neural nets. The early big-name training sets tended to be all square though, and if your design ends with a fully-connected layer (more common back then as well) then you're pretty locked-in for aspect ratio unless you retrain.

I think the point is that the OS needs to interface with the hardware and read out the data, which is then passed to a program which works with it, part of which is a neural network in this case. Usually there's some marshalling and unmarshalling in there too.

Reply on news.ycombinator.com