Peter's Quasi-Daily Blah-g

A team of physicists and engineers at Princeton University built a twisting fusion reactor known as a stellarator that uses permanent magnets, showcasing a potentially cost-effective way of building the powerful machines. Their experiment, called MUSE, relies on 3D-printed and off-the-shelf parts.

Nuclear fusion, the reaction that powers stars like our Sun, produces huge amounts of energy by merging atoms (not to be confused with nuclear fission, which produces less energy by splitting atoms). Nuclear fission is the reaction at the core of modern nuclear reactors that power electric grids; scientists have yet to crack the code on nuclear fusion as an energy source. Even once that long-sought goal is reached, scaling the technology and making it commercially viable is its own beast.

Stellarators are cruller-shaped devices that contain high-temperature plasmas, which can bed tuned to foster the conditions for fusion reactions. They are similar to tokamaks, doughnut-shaped devices that run fusion reactions. Tokamaks rely on solenoids, which are magnets that carry electric current. MUSE is different.

“Using permanent magnets is a completely new way to design stellarators,” said Tony Qian, a physicist at Princeton University and lead author of two papers published in the Journal of Plasma Physics and Nuclear Fusion that describe the design of the MUSE experiment. “This technique allows us to test new plasma confinement ideas quickly and build new devices easily.”

Permanent magnets don’t need electric current to generate their magnet fields and can be purchased off-the-shelf. The MUSE experiment stuck such magnets onto a 3-D printed shell.

“I realized that even if they were situated alongside other magnets, rare-earth permanent magnets could generate and maintain the magnetic fields necessary to confine the plasma so fusion reactions can occur,” Michael Zarnstorff, a research scientist at the university’s Plasma Physics Laboratory and principal investigator of the MUSE project, in a press release. “That’s the property that makes this technique work.”

Last year, scientists at the Department of Energy’s Lawrence Livermore National Laboratory (LLNL) achieved breakeven in a fusion reaction; that is, the reaction produced more energy than it took to power it. However, that accolade neglects to account for the “wall power” necessary to induce the reaction. In other words, there’s still a long, long road ahead.

The LLNL breakthrough was done by shining powerful lasers at a pellet of atoms, a different process than the plasma-based fusion reactions that occur in tokamaks and stellarators. Little tweaks to the devices, like the implementation of permanent magnets in MUSE or an upgraded tungsten diverter in the KSTAR tokamak, make it easier for scientists to replicate the experimental setups and perform experiments at high temperatures for longer.

Taken together, these innovations will allow scientists to do more with the plasmas at their fingertips, and maybe—just maybe—reach the vaunted goal of usable and scalable fusion energy.

Meta released its own AI-powered image generator back in December called Imagine. Like other image generators of its kind, Imagine seems to have some very strange racial hangups. Namely, the application can’t seem to envision a world where Asian men are capable of being with white women.

Let me explain: The Verge was first to notice that Meta’s image generator did not seem to be able to generate images of white women and Asian men together. Journalist Mia Sato describes her futile attempts to generate said interracial images. Indeed, trying “dozens of times to create an image using” a variety of prompts, Sato notes that she was only successful once.

We did our own test and had even less luck. From my personal experiments with the image generator, it is basically impossible to get the application to create images of a white woman and an Asian man together. I tried every combination of prompts I could think of and was only ever successful in creating images of an Asian man and an Asian woman together. Other interracial couples could easily be generated, including a Black man and a white woman, a white man and an Asian woman, and a white woman and a Middle Eastern man, but when it came to Asian men, they were apparently shit out luck.

While the image generator failed to generate images of white women and Asian men together, it did manage to respond to other, much weirder prompts, such as when I asked it to show me a dog who is friends with a robot.

Or a fish who is addicted to gambling:

This follows on the heels of a similar, slightly worse scandal involving Google, in which the company’s Gemini image generator was unable to generate consistent pictures of white people. The application also generated a multitude of historically inaccurate images, including Black Vikings and Black Nazis. Google ultimately had to pause the platform and promised to bring it back after the bugs had been worked out.

Right-wing social media influencers have predictably jumped all over this trend to bolster some larger, catastrophizing commentary about the excesses of liberal ideology. I won’t go in for that but, honestly, what gives? How is this a thing? Why are these companies so bad at creating platforms that deal with race? Is this some terrible marketing stunt? Are we all being punked? What is happening?

Gizmodo reached out to Meta to get more information about why its app is screwed up and will update this story if it responds.