I wonder if that clause is legal. It could be argued that it legitimately protects the capital investment needed to make the model. I’m not sure if that’s true, though.

FunSearch (so called because it searches for mathematical functions, not because it’s fun)

I’m probably not the only one who wondered.

It’s likely a reference to Yudkowsky or someone along those lines. I don’t follow that crowd.

Yes, if it’s new content, it’s obviously no copy; so no copyvio (unless derivative, like fan fiction, etc.). I was thinking of memorized training data being regurgitated.

I understand. The idea would be to hold AI makers liable for contributory infringement, reminiscent of the Betamax case.

I don’t think that would work in court. The argument is much weaker here than in the Betamax case, and even then it didn’t convince. But yes, it’s prudent to get the explicit permission, just in case of a case.

That shouldn’t be an issue. If you look at an unauthorized image copy, you’re not usually on the hook (unless you are intentionally pirating). It’s unlikely that they needed to get explicit “consent” (ie license the images) in the first place.

That ought to satisfy all those who wanted “consent” for training data.

IMO, we need to ask: What benefits the people? or What is in the public interest?

That should be the only thing of importance. That’s probably controversial. Some will call it socialism. It is pretty much how the US Constitution sees it, though.

Maybe you agree with this. But when you talk about “models trained on public data” you are basically thinking in terms of property rights, and not in terms of the public benefit.

Digital media means that there is an ongoing service behind it. The servers use energy. The parts age and break. It requires a continuing feed of labor and resources to keep going.

Imagine a streaming service that is all based on buying media, instead of subscription or renting. Then suppose all the customers somehow decide that the media they own are enough for now (maybe because money is tight, because inflation). With no more cash coming in, the service goes bankrupt.

In principle, you could have a type of license that allows you to get a new copy in any way you can (torrent, etc.). That would be hard to police, though.

FWIW, owning a physical copy isn’t all that, either. There are various ways built-in to make life harder for customers, like geo-blocking. Bypassing these tends to be a criminal offense.

In reality, what you’re saying makes no sense.

Making something available on the internet means giving permission to download it. Exceptions may be if it happens accidentally or if the uploader does not have the necessary permissions. If users had to make sure that everything was correct, they’d basically have to get a written permission via the post before visiting any page.

Fair use is a defense against copyright infringement under US law. Using the web is rarely fair use because there is no copyright infringement. When training data is regurgitated, that is mostly fair use. If the data is public domain/out of copyright, then it is not.

They almost certainly had, as it was downloaded from the net. Some stuff gets published accidentally or illegally, but that’s hardly something they can be expected to detect or police.

They do not have permission to pass it on. It might be an issue if they didn’t stop it.

The model cards for Stable Diffusion 1.5 and 2.1 estimate the CO2 emissions as 11.25 tons and 12 tons for training. XL lacks the info.

A transatlantic flight (round-trip) is about 1 ton per pax. So, while every little bit helps, ML is not where you can make the big gains in lowering emissions.

So much this. Most people under 40 must have grown up with video games. Shouldn’t they have noticed at some point that the enemies and NPCs are AI-controlled? Some games even say that in the settings.

I don’t see the point in the expression “AGI” either. There’s a fundamental difference between the if-else AI of current games and the ANNs behind LLMs. But there is no fundamental change needed to make an ANN-AI that is more general. At what point along that continuum do we talk of AGI? Why should that even be a goal in itself? I want more useful and energy-efficient software tools. I don’t care if it meets any kind of arbitrary definition.

I see that not everyone’s a cynic, yet.

What does that mean, though?

Spotify is Swedish.

After reading the whole article, I still don’t know what Uruguay wants to happen.

People say that AI will kill us all by ordering too many paperclips.

So people try to make AI safe by stopping it from making images of nude people.

WTF is wrong with everyone? Am I stuck in the most boring Lewis Carroll story ever?

Ideally, they wouldn’t be paying salaries? What?

“I am Andrew Ryan, and I’m here to ask you a question. Is a man not entitled to the sweat of his brow? ‘No!’ says the man in Washington, ‘It belongs to the poor.’

How do Americans feel about this attitude?