❌

Normal view

There are new articles available, click to refresh the page.
Before yesterdayMain stream

Researchers find what makes AI chatbots politically persuasive

✇Ars Technica
By: Jacek Krywko
4 December 2025 at 15:07

Roughly two years ago, Sam Altman tweeted that AI systems would be capable of superhuman persuasion well before achieving general intelligence—a prediction that raised concerns about the influence AI could have over democratic elections.

To see if conversational large language models can really sway political views of the public, scientists at the UK AI Security Institute, MIT, Stanford, Carnegie Mellon, and many other institutions performed by far the largest study on AI persuasiveness to date, involving nearly 80,000 participants in the UK. It turned out political AI chatbots fell far short of superhuman persuasiveness, but the study raises some more nuanced issues about our interactions with AI.

AI dystopias

The public debate about the impact AI has on politics has largely revolved around notions drawn from dystopian sci-fi. Large language models have access to essentially every fact and story ever published about any issue or candidate. They have processed information from books on psychology, negotiations, and human manipulation. They can rely on absurdly high computing power in huge data centers worldwide. On top of that, they can often access tons of personal information about individual users thanks to hundreds upon hundreds of online interactions at their disposal.

Read full article

Comments

© Carol Yepes via Getty

Welcome to “necroprinting”—3D printer nozzle made from mosquito’s proboscis

✇Ars Technica
By: Jacek Krywko
4 December 2025 at 11:16

Necrobotics is a field of engineering that builds robots out of a mix of synthetic materials and animal body parts. It has produced micro-grippers with pneumatically operated legs taken from dead spiders and walking robots based on deceased cockroaches. “These necrobotics papers inspired us to build something different,” said Changhong Cao, a mechanical engineering professor at the McGill University in Montreal, Canada.

Cao’s team didn’t go for a robot—instead, it adapted a female mosquito proboscis to work as a nozzle in a super-precise 3D printer. And it worked surprisingly well.

Fangs and stings

To find the right nozzle for their 3D necroprinting system, Cao’s team began with a broad survey of natural micro-dispensing tips. The researchers examined stingers of bees, wasps, and scorpions; the fangs of venomous snakes; and the claws of centipedes. All of those evolved to deliver a fluid to the target, which is roughly what a 3D printer’s nozzle does. But they all had issues. “Some were too curved and curved for high-precision 3D printing,” Cao explained. “Also, they were optimized for delivering pulses of venom, not for a steady, continuous flow, which is what you need for printing.”

Read full article

Comments

© Walter Ferrari / 500px

DeepMind’s latest: An AI for handling mathematical proofs

✇Ars Technica
By: Jacek Krywko
19 November 2025 at 10:57

Computers are extremely good with numbers, but they haven’t gotten many human mathematicians fired. Until recently, they could barely hold their own in high school-level math competitions.

But now Google’s DeepMind team has built AlphaProof, an AI system that matched silver medalists’ performance at the 2024 International Mathematical Olympiad, scoring just one point short of gold at the most prestigious undergrad math competition in the world. And that’s kind of a big deal.

True understanding

The reason computers fared poorly in math competitions is that, while they far surpass humanity’s ability to perform calculations, they are not really that good at the logic and reasoning that is needed for advanced math. Put differently, they are good at performing calculations really quickly, but they usually suck at understanding why they’re doing them. While something like addition seems simple, humans can do semi-formal proofs based on definitions of addition or go for fully formal Peano arithmetic that defines the properties of natural numbers and operations like addition through axioms.

Read full article

Comments

© Hill Street Studios

The evolution of rationality: How chimps process conflicting evidence

✇Ars Technica
By: Jacek Krywko
16 November 2025 at 07:00

When Aristotle claimed that humans differ from other animals because they have the ability to be rational, he understood rational to mean that we could form our views and beliefs based on evidence, and that we could reconsider that evidence. “You know—ask ourselves if we should really believe that based on the evidence we’ve got,” says Jan M. Engelmann, an evolutionary anthropologist at the University of California, Berkeley.

Engelmann says that from the beginning of the Western intellectual tradition, people thought that only humans are rational. So, he designed a study to see if rationality shows up in chimpanzees. It turned out that they’re almost as rational as we are.

Food puzzles

“There was quite a bit of research showing that chimpanzees can form their beliefs in response to evidence,” Engelmann says. The experiments usually involved chimpanzees deciding which of the two boxes contained a snack. When the researchers shook both boxes and there was a rattling sound coming from one of them, the chimps almost always chose the box where the rattling came from.

Read full article

Comments

© Rizky Panuntun

Wyoming dinosaur mummies give us a new view of duck-billed species

✇Ars Technica
By: Jacek Krywko
15 November 2025 at 07:00

Edmontosaurus annectens, a large herbivore duck-billed dinosaur that lived toward the end of the Cretaceous period, was discovered back in 1908 in east-central Wyoming by C.H. Sternberg, a fossil collector. The skeleton, later housed at the American Museum of Natural History in New York and nicknamed the “AMNH mummy,” was covered by scaly skin imprinted in the surrounding sediment that gave us the first approximate idea of what the animal looked like.

More than a century later, a team of paleontologists led by Paul C. Sereno, a professor of organismal biology at the University of Chicago, got back to the same exact place where Sternberg dug up the first Edmontosaurus specimen. The researchers found two more Edmontosaurus mummies with all fleshy external anatomy imprinted in a sub-millimeter layer of clay. For the first time, we uncovered an accurate image of what Edmontosaurus really looked like, down to the tiniest details, like the size of its scales and the arrangement of spikes on its tail. And we were in for at least a few surprises.

Evolving images

Our view of Edmontosaurus changed over time, even before Sereno’s study. The initial drawing of Edmontosaurus was made in 1909 by Charles R. Knight, a famous paleoartist, who based his visualization on the first specimen found by Sternberg. “He was accurate in some ways, but he made a mistake in that he drew the crest extending throughout the entire length of the body,” Sereno says. The mummy Knight based his drawing on had no tail, so understandably, the artist used his imagination to fill in the gaps and made the Edmontosaurus look a little bit like a dragon.

Read full article

Comments

© Tyler Keillor/Fossil Lab

Tiny chips hitch a ride on immune cells to sites of inflammation

✇Ars Technica
By: Jacek Krywko
13 November 2025 at 16:11

Standard brain implants use electrodes that penetrate the gray matter to stimulate and record the activity of neurons. These typically need to be put in place via a surgical procedure. To go around that need, a team of researchers led by Deblina Sarkar, an electrical engineer and MIT assistant professor, developed microscopic electronic devices hybridized with living cells. Those cells can be injected into the circulatory system with a standard syringe and will travel the bloodstream before implanting themselves in target brain areas.

“In the first two years of working on this technology at MIT, we’ve got 35 grant proposals rejected in a row,” Sarkar says. “Comments we got from the reviewers were that our idea was very impactful, but it was impossible.” She acknowledges that the proposal sounded like something you can find in science fiction novels. But after more than six years of research, she and her colleagues have pulled it off.

Nanobot problems

In 2022, when Sarkar and her colleagues gathered initial data and got some promising results with their cell-electronics hybrids, the team proposed the project for the National Institutes of Health Director’s New Innovator Award. For the first time, after 35 rejections, it made it through peer review. “We got the highest impact score ever,” Sarkar says.

Read full article

Comments

© Yadav, et al.

❌
❌