The study authors believe an eruption occurred around 1345, about two years before the start of the pandemic, from either a single volcano or a cluster of volcanoes of unknown location, likely in the tropics. The resulting haze from volcanic ash would have partially blocked sunlight across the Mediterranean region over multiple years, causing temperatures to drop and crops to fail.

Wow, that’s exactly what happened just before the Plague of Justinian (i.e., the volcanic winter of 536). I’m surprised they don’t mention that in the article.

They’re seeing gamma rays from suspected dark matter collisions—that’s actually more substantially accurate than I was expecting.

It looks like the research was done by academic and government-sponsored institutions in Spain and China, so hopefully it won’t just become a profit-making tool for the biotech industry.

Don’t animals accumulate the heavy metals they consume from plants?

If I’m reading the paper correctly, this conclusion is just based on the author testing himself?

First I’ve heard of StopNCII… what’s to stop it from being abused to remove (say) images of police brutality or anything else states or “participating companies” don’t want to be seen?

First paper I’ve seen where the list of authors is longer than the main content.

“Up until the semi finals, it seemed like nothing would be able to stop Grok 4 on its way to winning the event,” Pedro Pinhata, a writer for Chess.com, said in its coverage. “Despite a few moments of weakness, X’s AI seemed to be by far the strongest chess player… But the illusion fell through on the last day of the tournament.” He said Grok’s “unrecognizable” and “blundering” play enabled o3 to claim a succession of “convincing wins”.

I think the main takeaway is that these models are fundamentally inconsistent, and you can never assume they’re going to be reliable based on past performance.

The typical pattern for leaders is to get “second opinions” from advisors who tell them whatever they want to hear, so… maybe asking the equivalent of a magic 8 ball is a marginal improvement?

“Researchers in the field sometimes describe our goal as to pass the ‘Visual Turing Test,’” said Suyeon Choi […] “A visual Turing Test then means, ideally, one cannot distinguish between a physical, real thing as seen through the glasses and a digitally created image being projected on the display surface,” Choi said.

So they just came up with a needlessly opaque synonym of “verisimilitude”.

I presume the goal was to put the remains in low earth orbit, where they would eventually have reentered anyway—right?

You could end up with Waking Ned Devine, where everyone in town vouches for everyone else because they all stand to benefit.

“When dams are built, large areas are flooded and people need to be relocated,” Láng-Ritter said in a press statement. “The relocated population is usually counted precisely because dam companies pay compensation to those affected.”

So the locals are incentivized to inflate their numbers?

IMO the focus should have always been on the potential for AI to produce copyright-violating output, not on the method of training.

Why would the article’s credited authors pass up the chance to improve their own health status and health satisfaction?

Critical paragraph:

Our research highlights the importance of Germany’s unique institutional context, characterized by strong labor protections, extensive union representation, and comprehensive employment legislation. These factors, combined with Germany’s gradual adoption of AI technologies, create an environment where AI is more likely to complement rather than displace worker skills, mitigating some of the negative labor market effects observed in countries like the US.

That makes sense—being raised by ChatGPT might be marginally better than being raised by Sam Altman.

It sounds like they’re tracking people by state of birth, but how do they account for people who move to different states? They could just exclude them from the study—but mobility probably correlates with wealth, which in turn correlates with longevity.

How does that compare to the growth in size of the overall code base?

There was a recent paper claiming that LLMs were better at avoiding toxic speech if it was actually included in their training data, since models that hadn’t been trained on it had no way of recognizing it for what it was. With that in mind, maybe using reddit for training isn’t as bad an idea as it seems.