Month: April 2023

Enlarge / A Moon jellyfish. (credit: Dan Kitwood / Getty Images)

The sensation of hunger seems pretty simple on the surface, but behind the scenes, it involves complicated networks of sending and signaling, with multiple hormones that influence whether we decide to have another serving or not. The ability to know when to stop eating appears to be widespread among animals, suggesting that it might have deep evolutionary roots.

A new study suggests that at least one part of the system goes back to nearly the origin of animals. Researchers have identified a hormone that jellyfish use to determine when they're full and stop eating. And they found that it's capable of eliciting the same response in fruit flies, suggesting the system may have been operating in the ancestor of these two very distantly related animals. That ancestor would have lived prior to the Cambrian.

Feeding the fish (or jellyfish)

Given they lack any obvious equivalents to a mouth, it might seem like it would be tough to determine whether a jellyfish is even eating, much less hungry. But a team of Japanese researchers showed that the jellyfish species Cladonema pacificum has a bunch of stereotypical behaviors while feeding, including that their tentacles latch onto prey and that they then withdraw the tentacle into the bell so that the prey can be digested. And, if you keep feeding the jellyfish brine shrimp, eventually this process will slow, indicating that the animal is sensing it is well fed. (There's a movie available of the jellyfish feeding.)

Enlarge (credit: Samsung / AMD / Ron Amadeo)

Samsung Electronics and AMD are extending their smartphone chip agreement. In a press release today, the two companies said they "signed a multi-year agreement extension to bring multiple generations of high-performance, ultra-low-power AMD Radeon graphics solutions to an expanded portfolio of Samsung Exynos SoCs." Even Samsung is reluctant to use Samsung chips these days, so it's not clear what devices these AMD GPUs will land in.

Samsung's chip division and AMD have already done a generation of an Exynos SoC with an AMD GPU. That chip was the Exynos 2200, with its "Samsung Xclipse 920 GPU" that was co-developed by AMD. Samsung's phone division—which doesn't necessarily have a bias toward the chip division's products—shipped that chip in the S22 in some regions like Europe, while shipping the S22 with Qualcomm chips in other regions, like the US and China. Exynos chips have a bad reputation for constantly having lower performance compared to Qualcomm chips, and the Exynos 2200 was no exception. The chip didn't do well against the Snapdragon 8 Gen 1 in benchmarks and power usage, and Samsung fans once again had to deal with getting "inferior phones" depending on what country they lived in.

The one win the Xclipse GPU had over Qualcomm was in ray tracing. In today's four-paragraph press release, Samsung and AMD point out twice that "Xclipse was the industry’s first mobile GPU with hardware-accelerated ray tracing." Samsung's 2022 chip is better at ray tracing than Qualcomm's 2023 chip! The problem is, being an early adopter of ray tracing isn't really relevant for mobile gaming. Mobile games are built for a causal audience and target mass-market hardware. With the need to also balance battery life, that market doesn't value high graphics fidelity. As you can see with Lenovo killing off its gaming smartphone line, the attempts to bring hardcore gaming values to smartphones have not found a huge audience.

Saturn might be the planet in our Solar System best known for its spectacular rings, but the icy giant Uranus also has a system of 13 nested rings. Eleven of those rings—nine main rings and two fainter dusty rings—are clearly visible in the latest spectacular image from NASA's Webb Space Telescope. Future images should reveal the remaining two faint outer rings discovered with the Hubble Space Telescope in 2007.

"Uranus has never looked better. Really," NASA tweeted. "Only Voyager 2 and Keck (with adaptive optics) have imaged the planet's faintest rings before, and never as clearly as Webb’s first glimpse at this ice giant, which also highlights bright atmospheric features."

As we've reported previously, the Webb Telescope launched in December 2021 and, after a suspenseful sunshield and mirror deployment over several months, began capturing stunning images. First, there was the deep field image of the Universe, released last July. This was followed by images of exoplanet atmospheres, the Southern Ring Nebula, a cluster of interacting galaxies called Stephan's Quintet, and the Carina Nebula, a star-forming region about 7,600-light-years away. These images reportedly brought astronomers to tears.

Enlarge / Tesla Model X SUV at the European Motor Show on January 9, 2020, in Brussels, Belgium. (credit: Getty Images | Sjoerd van der Wal)

From 2019 to at least mid-2022, Tesla employees used an internal messaging system to share "sometimes highly invasive videos and images recorded by customers' car cameras," according to a lengthy Reuters report based on interviews with nine former Tesla employees.

Although Tesla says its in-car cameras are "designed from the ground up to protect your privacy," today's Reuters report described employees as having easy access to the cameras' output and sharing that freely with other employees:

Some of the recordings caught Tesla customers in embarrassing situations. One ex-employee described a video of a man approaching a vehicle completely naked.

Also shared: crashes and road-rage incidents. One crash video in 2021 showed a Tesla driving at high speed in a residential area hitting a child riding a bike, according to another ex-employee. The child flew in one direction, the bike in another. The video spread around a Tesla office in San Mateo, California, via private one-on-one chats, "like wildfire," the ex-employee said.

There were "pictures of dogs and funny road signs that employees made into memes by embellishing them with amusing captions or commentary, before posting them in private group chats." Some posts could be seen by "scores" of employees.

Enlarge / Intel GPUs, ultra-wide monitors, and a vague sense that it's time for some gamers to refresh: These are some of the factors that have one report showing industry growth for PC gaming. (credit: Peerapon Boonyakiat/SOPA Images/LightRocket via Getty Images)

PC gamers may be concerned about losing their jobs in 2023, but not so much that they can ignore a glut of new GPU and ultra-widescreen monitor options.

That's the elevator-pitch version of Jon Peddie Research's report on PC gaming hardware sales and costs. At a high level, it suggests that while mid-range gaming will see only gradual growth from 2023 through 2025, both entry-level and high-end hardware should see notable upticks through 2025. The market should recover more than $5 billion overall from its 2022 drop-off, with the high end taking $3.92 billion and entry level $2.29 billion.

PC gaming market figures from JPR.

Reading exactly which bits of PC hardware fit into which segment, and getting more detail on how JPR put these numbers together, costs even more than a 40-series Nvidia card, at $27,500 per year for access. So we're left to wonder which cards, monitors, chips, and other gear fit into entry-, mid-level, and high-end. But JPR does suggest a few factors moving the numbers around:

Enlarge (credit: Aurich Lawson | Getty Images)

Over the past few months, AI chatbots like ChatGPT have captured the world's attention due to their ability to converse in a human-like way on just about any subject. But they come with a serious drawback: They can present convincing false information easily, making them unreliable sources of factual information and potential sources of defamation.

Why do AI chatbots make things up, and will we ever be able to fully trust their output? We asked several experts and dug into how these AI models work to find the answers.

“Hallucinations”—a loaded term in AI

AI chatbots such as OpenAI's ChatGPT rely on a type of AI called a "large language model" (LLM) to generate their responses. An LLM is a computer program trained on millions of text sources that can read and generate "natural language" text—language as humans would naturally write or talk. Unfortunately, they can also make mistakes.