dweinus

Pay. An. Artist.

These models are built on IP theft from artists who are now losing work to this slop. Even if you don't see the difference (rest assured, most of your audience will), the morality of it is terrible.

Don't forget to sort by color when you're done!

Nice post but they quickly disregard Dask, don't explain why, don't test it, even exclude it from the benchmark they quote. I don't know if is the better answer, but it seems worth testing if you want a balance approachable + scalable. That's kind the thing Dask was meant to do.

Cool, now show the public transit safety numbers.

Pretraining a frontier model is a one-time expense: 6,000 to 25,000 GPUs for one to four months, $5M to $150M depending on chip generation and luck.

> half of their ad spend was "wasted"

That would be a phenomenal ROI. In almost all advertising contexts the hit rate is far, far lower. Resulting in everyone drowning in ads to hit the few targets.

To the commenters saying, "but how will people/companies know what to buy?"... 1) Through the same high-quality information channels you usually use today: customer reviews, word of mouth, point of sale, and lived experience. For B2B, you could still have industry events or demos where it is essentially a physical opt-in for advertising. 2) I could not possibly care. The offsetting benefit of living in a world where I am not surrounded by messages trying to deceive me into spending money is well worth losing whatever tiny signal might be in that noise.

Why is this flagged?

Customer and/or musician rug pull in 3...2...

All those devs and data scientists and PhDs make a choice though. They could quit and work somewhere else. Even in a tough market their skills are in demand. They choose to work on this every morning.

I believe it is the other way around: the LLM decides the type of action and the input to the action; the code validates the permission to act and the acceptability of the input. But, yes it is very different than SQL injection in that way.

> We show prompt injections are driven by a flaw in how LLMs perceive roles.

LLMs don't "perceive roles", and that is exactly the problem.

> "The conversational interview, long considered the ultimate, unhackable test of a candidate’s authenticity"

Lol. I'm not sure this person has ever given an interview before

Human civilization? In this economy?

They routinely override the user-defined navigation to include whatever new content they are pushing. It was sold as an appliance without ads or subscription. Now it has become an ads platform that the user has less and less control over.

> unless you're Swiss, your opinion is irrelevant

Lol. Dude, sure the Swiss can vote however they want. But we all see you and can pass judgement on this thinly veiled anti-immigrant nonsense all day long. Respect it I will never.

I'm sure they are very proud of themselves for sneaking racist anti-immigrant policy in under the guise of left wing environmental rhetoric.

You keep using that word. I don't think it means what you think it means.

Looking at WHO is pushing for change tells you most of what you need to know. Teachers? No. Students? No. Parents? Usually not. Technologists with something to sell? Bingo.

Ah yes, Pascal's mugging.

Ah, so this is The Web I have been hearing so much about!

> says Duke University ecologist Stuart Pimm, who was not involved in the research. “We could better help millions of birds every year by solving the more immediate threats of disappearing habitats, collisions with building windows, and prowling outdoor cats,”

Yes. Even if they stuck it at the end, it shows good journalism to call this out.

So to make this profitable they need ads revenue from it, right? Imagine for a moment the ways AI can manipulate responses and conversations for marketers, because I guarantee the marketers have already thought about it.

So much this. I know their point is to show what these models can do, but it just one more example of people shoehorning LLMs in, instead of finding the right tool for the job or caring about performance. They could have even layered AI on top of a recommender.

No fair, we didn't even get the fun anarchy part before skipping right on to tyrrany!

Global extreme poverty has fallen because we have raised the floor, largely through international collaboration that if anything has happened in spite of the cyberlibertarianism, certainly not because of it. Paradoxically, "developed" nation inequality has hit 1920s levels.

Likewise, the number of countries/populations calling themselves democratic has grown, but the global democratic index has declined and mature democracies are substantially threatened.

"As with other drugs, some people can be casual dabblers; other people, well, can’t." I think this is true. Unfortunately people (this article included) tend to talk about it as purely benign or unmitigated poison. I think the reality is that the dose, the frequency, and the role it plays in someone's life make it a positive or a negative. I wish we could get out of the hyperbole to be honest about the pros and cons.

According to above, it is triggered by the website calling the feature. The user might have no idea. That's not what consent looks like

The reporter clearly knows the distro fixes have not been shipped, read their report. They chose to disclose anyway.

No, I will. The distros and the kernel devs should be talking and moving on high sev patches, sure. But real people will have gotten hurt because the reporter didn't want to wait for that to happen. That's on them.