News for hackers. Stuff that matters.
Re:We've filed a lawsuit against GitHub Copilot by smoldesu on Thursday November 03, 2022 @06:09PM (#33458523)
> “AI” is just fancy speak for “complex math program”

Not really? It's less about arithmetic and more about inferencing data in higher dimensions than we can understand. Comparing it to traditional computation is a trap, same as treating it like a human mind. They've very different, under the surface.

IMO, if this is a data problem then we should treat it like one. Simple fix - find a legal basis for which licenses are permissive enough to allow for ML training, and train your models on that. The problem here isn't developers crying out in fear of being replaced by robots, it's more that the code that it is reproducing is not licensed for reproduction (and the AI doesn't know that). People who can prove that proprietary code made it into Copilot deserve a settlement. Schlubs like me who upload my dotfiles under BSD don't fall under the same umbrella, at least the way I see it.

Re:We've filed a lawsuit against GitHub Copilot by Cort3z on Thursday November 03, 2022 @06:14PM (#33458596)
Who decides what constitutes an "AI program" vs just a "program"? What heuristic do we look at? At the end of the day, they have an equivalent of a .exe which runs, and outputs code that has a license attached to it.
Re:We've filed a lawsuit against GitHub Copilot by AnnasVirtual on Friday November 04, 2022 @03:42PM (#33472656)
no, you clearly don't understand Neural Network, basically it's a file containing artificial neurons and their connections
Re:We've filed a lawsuit against GitHub Copilot by bobbruno on Friday November 04, 2022 @06:37PM (#33474948)
I can suggest an idea, considering that the “AI program” is the model, not the training algorithm.

A program gets written by an entity (usually a person) and is executed to generate the desired output according to a deterministic mathematical function it expresses. A training algorithm is a program that gets written to train a model (the model being the “AI Program”) when presented to some training data inputs, to implement a function that is not the training algorithm function itself, but another one, generalising over a problem domain beyond just the original examples fed to the training algorithm.

The output model is not the training algorithm or the training data (or an encoding of it) and exists as its own artefact, independent of both.

Re:We've filed a lawsuit against GitHub Copilot by 2muchcoffeeman on Thursday November 03, 2022 @06:39PM (#33458926)
But it all runs on hardware we created and we know exactly what operations were implemented in that hardware. How is it not just math?
Re:We've filed a lawsuit against GitHub Copilot by AnnasVirtual on Friday November 04, 2022 @03:43PM (#33472676)
it is a math that simulate neurons firing
Re:We've filed a lawsuit against GitHub Copilot by 2muchcoffeeman on Saturday November 05, 2022 @05:39AM (#33479676)
That’s clearly not the case because if it were so simple we’d have an AGI now.
Re:We've filed a lawsuit against GitHub Copilot by heavyset_go on Thursday November 03, 2022 @06:16PM (#33458630)
I've been saying AI is computational statistics on steroids for a while, and I think that's an apt generalization of what ML is.
Re:We've filed a lawsuit against GitHub Copilot by OJFord on Thursday November 03, 2022 @07:06PM (#33459223)
That oughtn't be controversial, in fact I wouldn't even bother with 'on steroids', implying it's a slightly different/morphed thing. The way I learnt it (very slightly, at university, not a particular focus) it was abundantly clear it was just stats.
Re:We've filed a lawsuit against GitHub Copilot by heavyset_go on Thursday November 03, 2022 @07:11PM (#33459264)
I bring the steroids thing up because it's only relatively recently that we've had the massive computing power at our finger tips that we do now. We discovered the foundations of our current ML techniques a relatively long time ago, it's only been recently that we've been able to throw data centers full of powerful GPUs and whatnot at them.
Re:We've filed a lawsuit against GitHub Copilot by BenjiWiebe on Friday November 04, 2022 @01:45PM (#33470843)
If it's just statistics, wouldn't that mean you wouldn't get verbatim output? i.e. you'd throw away all high frequency content?
Re:We've filed a lawsuit against GitHub Copilot by kmeisthax on Friday November 04, 2022 @12:49AM (#33462332)
The only license that is permissive enough for AI training is CC0.

Art generators can't comply with attribution requirements and code generators don't know if and when they trip the GPL copyleft. I believe most permissive code licenses also have some kind of attribution requirement.

Re:We've filed a lawsuit against GitHub Copilot by urthor on Friday November 04, 2022 @07:55AM (#33465133)
> Not really? It's less about arithmetic and more about inferencing data in higher dimensions than we can understand.

This is a VERY poor definition of mathematics.