What’s fun is how often this principle is used every day. For example, when you upload a video to YouTube, you’re assigned a unique URL, but it would be too slow to simply add your URL to a list to make sure nobody else uses it. There are millions of videos uploaded every day, and thousands of servers spread all over the world.

Instead, YouTube just generates a truly random URL and depends on the odds of two videos having the same URL being effectively zero.

The same is true for Bitcoin. If you could guess a Bitcoin private key for any currently used wallet, you’d have full access to the funds within that wallet. This can even be done offline. Even if you could guess trillions of private keys per second, the odds of you hitting even one that’s already been used is low enough to be totally secure.

Interesting. Didn’t know about the google books case. I agree that it applies here.

I think it’s critically important to be very specific about what LLMs are “able to do” vs what they tend to do in practice.

The argument is that the initial training data is sufficiently altered and “transformed” so as not to be breaking copyright. If the model is capable of reproducing the majority of the book unaltered, then we know that is not the case. Whether or not it’s easy to access is irrelevant. The fact that the people performing the study had to “jailbreak” the models to get past checks tells you that the model’s creators are very aware that the model is very capable of producing an un-transformed version of the copyrighted work.

From the end-user’s perspective, if the model is sufficiently gated from distributing copyrighted works, it doesn’t matter what it’s inherently capable of, but the argument shouldn’t be “the model isn’t breaking the law” it should be “we have a staff of people working around the clock to make sure the model doesn’t try to break the law.”

That study is six months old. The one I linked is from three weeks ago.

No it isn’t. Read.

That’s quite a claim, I’d like to see that.

https://arxiv.org/abs/2601.02671

Yet most AI models can recite entire Harry Potter books if prompted the right way, so that’s all bullshit.

Color me a bit skeptic that such a filtration system can remove soap and oil residue from used shower water.

https://www.theseus.fi/bitstream/handle/10024/92876/Thesis.pdf?isAllowed=y&sequence=1

I guess they’ve done quite a bit of research. Not sure I’m qualified to comment on the validity of their conclusions.

I’m also curious how frequently the filter materials need to be replaced.

I wish this was a real law. Stayed at the Omni Parker House in Boston recently which is allegedly haunted by the ghost of Charles Dickens because he lived there for like a year.

But he didn’t die there. I feel like his ghost has better places to be than the flat he shacked up in while visiting the States that one time.

I don’t get it…

Looks like it’s getting close to the Streisand kiss of death.

Also that’s not that impressive. 30 million devices is roughly one per second year round. Phone manufacturers make many times that in a year.

Gemini evolved into a seal.

My first thought when I saw this.

Generally it’s made by first filtering out other gasses (there are materials that will pass nitrogen but block oxygen), and then you just get it super cold with a cryocompressor (uses helium as a refrigerant) to liquify it.

Just read the article.

Basically the entire US economy, every employer, many schools, and half of the commercials on TV are telling us to use and trust AI.

Kid was already using the bot for advice on homework and relationships (two things that people are fucking encouraged to do depending on who you ask). The bot shouldn’t give lethal advice. And if it’s even capable of doing that, we all need to take a huuuuuuge step back.

“I want to make sure so I don’t overdose,” Nelson explained in the chat logs viewed by the publication. “There isn’t much information online and I don’t want to accidentally take too much.”

Kid was curious and cautious, and AI gave him incorrect information and the confidence to act on that information.

He was 19. Cut this victim blaming bullshit. Being a kid is hard enough before technology went full cyberpunk.