Centaur: AI that thinks like us—and could help explain how we think

AI that thinks like us—and could help explain how we think
Evaluation in different held-out settings. Credit: Nature (2025). DOI: 10.1038/s41586-025-09215-4

Researchers at Helmholtz Munich have developed an artificial intelligence model that can simulate human behavior with remarkable accuracy. The language model, called Centaur, was trained on more than ten million decisions from psychological experiments—and makes decisions in ways that closely resemble those of real people. This opens new avenues for understanding human cognition and improving psychological theories.

For decades, psychology has aspired to explain the full complexity of human thought. Yet traditional models could either offer a transparent explanation of how people think—or reliably predict how they behave. Achieving both has long seemed out of reach.

The team le...

Read More

Clingy planets can trigger their own doom, Cheops and TESS suggest

How planets orbiting close to their host stars can cause their own downfall by triggering flares
How planets orbiting close to their host stars can cause their own downfall by triggering flares

Astronomers using the European Space Agency’s Cheops mission have caught an exoplanet that seems to be triggering flares of radiation from the star it orbits. These tremendous explosions are blasting away the planet’s wispy atmosphere, causing it to shrink every year.

This is the first-ever evidence of a “planet with a death wish.” Though it was theorized to be possible since the nineties, the flares seen in this research are around 100 times more energetic than expected.

The work is published in the journal Nature.

This planet’s star makes our sun look sleepy
Thanks to telescopes like the NASA/ESA/CSA James Webb Space Telescope and NASA’s Transiting Exoplanet Survey Satellite (TESS...

Read More

RisingAttacK: New technique can make AI ‘see’ whatever you want

AI eye
Credit: AI-generated image

Researchers have demonstrated a new way of attacking artificial intelligence computer vision systems, allowing them to control what the AI “sees.” The research shows that the new technique, called RisingAttacK, is effective at manipulating all of the most widely used AI computer vision systems.

At issue are so-called “adversarial attacks,” in which someone manipulates the data being fed into an AI system to control what the system sees, or does not see, in an image. For example, someone might manipulate an AI’s ability to detect traffic signals, pedestrians or other cars—which would cause problems for autonomous vehicles. Or a hacker could install code on an X-ray machine that causes an AI system to make inaccurate diagnoses.

“We wanted to find an eff...

Read More

Planets may start forming before stars even finish growing

New high-resolution images of protoplanetary disks in the Ophiuchus star-forming region, created with improved analysis. The resolution is shown by the white ellipse in the lower left of each panel, with a smaller ellipse indicating higher resolution. The white line in the lower right of each panel indicates a scale of 30 au. The evolution stage of the central stars progresses from left to right, and from top to bottom in the same row. (Credit: ALMA(ESO/NAOJ/NRAO), A. Shoshi et al.)

In a stellar nursery 460 light-years away, astronomers sharpened old ALMA data and spotted crisp rings and spirals swirling around 27 infant stars—evidence that planets start taking shape just a few hundred thousand years after their suns ignite, far earlier than anyone expected.

Signs of planet fo...

Read More