The LLM is for spam. Community colleges in California are suffering an absolute flood of fake student applicants that seem to be fraudsters backed by chatbots. [Voice of San Diego]

Many community colleges offer online courses. When students flood in, the teachers are delighted! Then they find that out of over 100 students, maybe 15 are real.

The rest are scammers who signed up to get state and federal financial aid, using a stolen identity.

The scammer finds a name and a social security number. They sign up for a full course load. They stick around long enough to get their Pell grant and cash out. Then they get a new identity and start again. [Voice of San Diego; San Francisco Chronicle, 2023]

Generative AI makes it much easier for a fake student to hold on until they get the payout.

In the year up to January 2024, about a quarter of all community college applicants in California seem to have been fake. By early 2025, it was around 37%. [CalMatters]

The teachers did not sign up to be bot cops. Eric Maag of Southwestern College said, “We’re having to have these conversations with students, like ‘Are you real? Is your work real?’”

The teachers are begging the state and the local district to put in more effort to verify applicants — all student applications go through the state first. We’ll see if the recent press attention gets them moving.

• Video version

Known for their elaborate performances almost as much as their albums, the American band OK Go just released a new music video that is very literally a hall of mirrors. Filmed in a Budapest train station, “Love” incorporates 29 robots and 60 mirrors that move in perfect synchronicity. The result is an endlessly evolving kaleidoscope that distorts reality and illusion, connecting the band and their surroundings through a trippy, impeccably timed production.

To get a closer look at the making of this iconic video, check out the behind-the-scenes video shot from the perspective of each band member.

Do stories and artists like this matter to you? Become a Colossal Member today and support independent arts publishing for as little as $7 per month. The article OK Go’s New Music Video Is a Dizzying Kaleidoscope of 60 Mirrors and 29 Robots appeared first on Colossal.

Welcome back to the Abstract!

It was a good week for the Fox Mulders among us. We want to believe, and a new study has given us some empirical grist along those lines. I shall say no more, and let the giant planet K2-18b speak for itself.  

Next, remember the time that Earth partied so hard that the Northern Lights showed up in the Sahara and humans had to invent sunscreen? Hahaha…our planet just DGAF sometimes. Then, the Perseus cluster thought it could get away with eating a subcluster, but we have the EVIDENCE. Last, dino-walk with me. 

“It’s Never Aliens”...Oh Shoot, This Time It Actually Might Be Aliens 

Madhusudhan, Nikku et al. ‘New Constraints on DMS and DMDS in the Atmosphere of K2-18b from JWST MIRI.’ The Astrophysical Journal Letters.

Scientists have reported some of the most compelling evidence for extraterrestrial life ever identified, teasing what would be the most anticipated scientific breakthrough in history. 

The possibly life-bearing world in question is K2-18b, a giant planet about eight or nine times as massive as Earth located about 124 light years away. It belongs to a tantalizing class of “Hycean” planets that may host global liquid water oceans under thick hydrogen-rich atmospheres. 

K2-18b has attracted a lot of interest in recent years because water vapor, carbon dioxide, and methane have been detected in its skies at concentrations that hint at the presence of life. That said, planets can bake up those compounds in all kinds of ways that don’t involve beasties. That’s why scientists decided to put this world into the sights of the James Webb Space Telescope, the most sensitive observatory ever, to see if they could find anything more concrete.

Webb delivered by confirming the presence of dimethyl disulfide (DMDS) and dimethyl sulfide (DMS)—compounds that are concocted by microbes here on Earth— in the skies of K2-18b. It’s not yet an alien slam dunk, because there are a few abiotic processes that also can make this stuff. But there are simply fewer ways to explain their sustained presence in a planet’s skies without invoking biological processes, compared to water, carbon dioxide, or methane.

“We present a mid-infrared transmission spectrum of K2-18b with JWST, the first for a habitable-zone exoplanet,” said researchers led by Nikku Madhusudhan of the University of Cambridge. “The spectrum shows multiple spectral features…that are best explained by a combination of DMDS and DMS in the atmosphere, both molecules uniquely produced by life on Earth and predicted as promising biosignatures in habitable exoplanets.”

DMDS and DMS are largely produced by marine microbes, such as phytoplankton. It’s alluring to envision a massive ocean world blushing with colorful microbial blooms as it orbits its red dwarf star, just 124 light years away (an immense distance for humans, but just down the street on galactic scales). 

Still, the team emphasized that abiotic sources of these compounds should be rigorously explored, and noted that they can be found on barren comets in our own solar system. It’s possible that similar comets in the K2-18 system may have recently crashed into this massive world, producing some transient signatures captured by Webb, though it would be a bit of a coincidence. Future observations may distinguish the likely sources of the compounds, and perhaps find even more signs of life (or signs of not-life). 

Humans have often imagined our “first contact” moment with aliens as irrefutable. We receive an unambiguously artificial transmission. An ancient alien artifact is unearthed from a nearby planet. Aliens straight-up show up on Earth to invade or enlighten us. 

But it seems much more likely that this vexing mystery—is life on Earth a fluke or the norm?—will be constrained through a slow and grinding probabilistic framework. We may never conclusively determine if K2-18b hosts life; the best we might get is a gradient of more to less probable. As scientists accumulate reams of data from other planets, we will get a lot more smoke signals but may never find an actual smoking gun.  

“Overall, our findings present an important step forward in the search for signatures of life on exoplanets,” according to the study. “However, robustly establishing both the veracity of the present findings and their possible association with life on K2-18 b needs a dedicated community effort in multiple directions—observational, theoretical, and experimental.” 

“The central question now is whether we are prepared to identify the signatures of life on such worlds,” the team concluded. “The opportunity is at our doorstep.”

A Tale of Auroral Escapes and Neanderthal Capes

Mukhopadhyay, Agnit et al. “Wandering of the auroral oval 41,000 years ago.” Science Advances.

Even Mother Earth, the one world that we know hosts life, can be a bit of a chaotic parent at times. For instance, our planet went on a little bender about 41,000 years ago, called the Laschamps geomagnetic excursion, in which it temporarily lost its magnetic bearings for about 2,000 years. 

As the magnetic poles shifted from the geographic poles, Earth’s magnetic shield was reduced to 10 percent of its strength, exposing everything on its surface to a flood of cosmic radiation. The excursion also caused auroras to become unmoored from the poles, thereby drifting to lower latitudes and causing the Northern Lights to dance over the Sahara.

This bizarre episode may have inspired our human ancestors to invent sunscreen out of ochre and develop tailored clothing to avoid radiation exposure, according to a new study. The work also suggests that Neanderthals may have failed to adapt to the changes, perhaps contributing to their extinction around the same time as the Laschamps excursion.

Ochre shows “increased frequency in archaeological sites dating to the peri-Laschamps” which “could be due in part to its use as a sunscreen” by anatomically modern humans (AMH), said researchers led by Agnit Mukhopadhyay of the University of Michigan. 

“Although both Neanderthals and contemporary AMH produced technologies associated with clothing manufacture, only AMH appear to have produced technologies consistent with the manufacture of tailored clothing; Neanderthals are assumed to have produced only relatively simple, draped clothing (e.g., capes)...Neanderthals’ decline was almost certainly multifactorial, but it is possible that topical sunscreens and tailored clothing provided AMH essential photoprotection and access to resources in places and at times they would otherwise have been inaccessible” which may have been “a competitive advantage.”

First of all, we need to bring back Neanderthal capes. They have been out of fashion for 41,000 years, so a revival is frankly overdue. Second, it would be wild if humans outlived Neanderthals in part because we wore sunscreen. Has there ever been a better advertisement for Big SPF? 

But that is only the skin-deep part of this cool and expansive study, which also speculated that the excursion may have inspired new forms of art and music. “Others have noted co-occurrence of the Laschamps with the earliest known representational cave art—which depicts animals, anthropomorphs, and other figures or scenes, as opposed to abstract marks or designs—including images of animals,” the team said. “To this, we add that the Laschamps event coincides with early examples of portable art and musical instruments.”

The researchers also warned that if the same event happened today, we would be up a magnetic creek without a paddle. “Considering the probable impact of the Laschamps excursion on early humans and their way of life, a similar event today would likely have dire consequences for modern humans,” the team noted. “The ramifications of a Laschamps-like magnetospheric configuration and auroral oval would reverberate across all facets of modern communication, satellite infrastructure, and intercontinental travel.”

I know, I know, you really needed one more thing to worry about. You had so many wonderful worries, but to fill out your collection, I present you with this one about Earth’s magnetic field shutting down for a millennium or two. No geomagnetic excursions are imminent, according to the study, but it will eventually happen again, so something to be aware of, I guess. 

In the meantime, honor your ancestors by wearing sunscreen!  

SOLVED: The Case of the Missing Subcluster Halo

Hyeonghan, Kim et al. “Direct evidence of a major merger in the Perseus cluster.” Nature Astronomy.

We’ll now zoom out from planetary scales all the way up to the mind-detonating expanse of the Perseus Cluster, one of the most massive objects in the known universe. This thing is just a lot of galaxies—at least 1,000, probably a lot more—that are all gravitationally jostling around each other some 240 million light years from Earth. 

Based on its dynamics, it looks like the cluster ate a smaller “subcluster” in the past as part of a “major merger.” But until now, scientists have not been able to track down the companion that was absorbed into the Perseus whole, which is a missing piece that has really been bugging them. 

“Although the Perseus cluster has often been regarded as an archetypical relaxed galaxy cluster, several lines of evidence ...suggest that the cluster might have experienced a major merger,” said researchers led by Kim HyeongHan of Yonsei University. (I also identify as an “archetypical relaxed” entity).

“However, the absence of a clear merging companion identified so far hampers our understanding of the evolutionary track of the Perseus cluster consistent with these observational features,” the team said.

We simply must not be hampered in our understanding of the evolutionary track of the Perseus cluster. To that end, the team used weak lensing, an observational technique based on gravitational distortions, to locate the swallowed companion.

Well, everyone: We got ‘em. “Here, through careful weak-lensing analysis, we successfully identified the missing subcluster halo,” which is located about 430 kiloparsecs (1.4 million light years) west of the Perseus main cluster core,” the team said. “This discovery resolves the long-standing puzzle of Perseus’s dynamical state.”

And to think, Perseus would have gotten away with it, too, if it weren’t for those meddling weak lenses. Another cosmic cold case closed. 

Slap a “Coexist” Bumper Sticker on These Armoured Dinosaurs

Arbour, Victoria et al. “A new thyreophoran ichnotaxon from British Columbia, Canada confirms the presence of ankylosaurid dinosaurs in the mid Cretaceous of North America.” Journal of Vertebrate Paleontology.

We’ll close by walking in the footsteps of an ankylosaurus, the dinosaurian equivalent of a tank. Some 100 million years ago, these giant armored dinosaurs left footprints in what is now the Canadian Rockies, which paleontologists have identified as the only known ankylosaurid ankylosaur tracks in the world. 

The name “ankylosaurid ankylosaur” may seem a bit redundant, but it exists because there is such a thing as—you guessed it—a non-ankylosaurid ankylosaur (also known as a nodosaurid). These two major ankylosaur lineages differed in many ways, including in the number of digits on their hind feet (ankylosaurids had three, nodosaurids had four). 

Paleontologists had previously identified the footprints of Tetrapodosaurus borealis, the four-toed variety, in mid-Cretaceous trackways near the town of Tumbler Ridge, in British Columbia, and at Dunvegan Bridge, Alberta. Now, a team has pinpointed the tell-tale “tridactyl” prints of an ankylosaurid ankylosaur representing a new species, named Ruopodosaurus clava.  

“This new taxon is currently known exclusively from the Cenomanian of northeastern British Columbia and northwestern Alberta, and provides confirmation that ankylosaurid ankylosaurs were present in North America prior to the Campanian–Maastrichtian,” said researchers led by Victoria Arbour of the Royal BC Museum. “Ruopodosaurus clava is found in the same localities and deposits as Tetrapodosaurus borealis, indicating that both ankylosaurid and non-ankylosaurid ankylosaurs co-existed in the mid Cretaceous of the Peace Region.”

Co-existence between giant armoured dinosaurs in a place now named the Peace Region? The reality simulation writers did well with this one.  

Thanks for reading! See you next week.

Der Schock kam bereits am Mittwochmorgen: Per Mail teilte das Bezirksamt mit, dass auf dem Bergedorfer Friedhof keine neuen Beerdigungen mehr angemeldet werden können. Der Grund: Ein fehlerhaftes Software-Update hat das digitale Buchungssystem lahmgelegt. Die „Bergedorfer Zeitung“ berichtet, dass selbst die Bestatter keine Vorwarnung erhielten. Wie lange der Zustand andauert, ist unklar. Die IT arbeite „mit Hochdruck“ an einer Lösung, heißt es vom Amt.

Ich wundere mich ja immer, woher die nach dem Verkacken immer den "Hochdruck" nehmen. Wenn die Energie und Mittel für Hochdruck haben, wieso wenden sie die nicht vorher an?

A “reasoning AI” is a large language model — a chatbot — that gives you a list of steps it took to reach a conclusion. Or a list of steps it says it took.

LLMs don’t know what a fact is — they just generate words. So chatbots are notorious for hallucinating — when it generates a pile of plausible text that’s full of factual errors, but it’s the right shape for whatever you were asking for.

Given this, it’s no surprise that when you ask a “reasoning” model to explain its reasoning steps, it just generates something that looks like a list of reasoning steps — and not necessarily anything it actually used.

Anthropic is an LLM vendor founded by AI doom mongers who left OpenAI. They’re extremely into the idea that the default behaviour of AI is that the robot is plotting against us. Anthropic was responsible for a pile of preprints claiming an LLM was trying to deceive researchers, and in every case it turned out the researchers told the LLM to behave that way.

Even Anthropic couldn’t really spin this present study — of course the BSing chatbot was going to BS about how it was BSing! Though their paper talks about “faithfulness” instead of, say, “accuracy,” because they just can’t stop anthropomorphising the chatbot. [Anthropic; Anthropic, PDF]

Other researchers looking at reasoning LLM have found the same behaviour — that everything an LLM puts out is confabulated, so when it explains its behaviour, that’s confabulated too. Because they’re confabulation machines, not true-or-false machines.

Transluce is a San Francisco “AI safety” charity which brags that it analyses LLMs using other LLMs. Transluce was shocked to find that OpenAI’s o3 model “frequently fabricates actions it took to fulfill user requests, and elaborately justifies the fabrications when confronted by the user.” [Transluce; Transluce]

Transluce doesn’t have a visible cohort of AI doomers, but they do insist on describing this behaviour, not as inaccuracy, but as a not being “truthful” — even as an LLM has no concept of true or false. It doesn’t even have concepts. [Twitter, proxy]

Benj Edwards at Ars Technica wrote up Anthropic’s findings. The first version of his story cut’n’pasted Anthropic’s humanised descriptions of the chatbot so egregiously that an editor fixed the headline and thoroughly revised the story to “reduce overly anthropomorphic language.” [Ars Technica, archive of 10 April 2025]

This is a good worked example of gullible AI journalism and how to fix it a little. Here are my three favourite changes. First example:

• Old version: “they sometimes hide their actual methods while fabricating elaborate explanations instead.”
• New version: “the ‘work’ they show can sometimes be misleading or disconnected from the actual process used to reach the answer.”

Second example:

• Old version: “OpenAI’s o1 and o3 series SR models deliberately obscure the accuracy of their ‘thought’ process, so this study does not apply to them.”
• New version: these “SR models were excluded from this study.”

For the third example, Edwards went so over-the-top that the editor had to insert a whole new paragraph about how “It’s important to note that AI models don’t have intentions or desires”.

Let’s hope Ars does better going forward.

• Video version