Hoofed crocodiles! And thoughts on AI and online sources for independent research.
Last edited Mon Jun 29, 2026, 02:33 PM - Edit history (1)
There’s been discussion afoot about AI systems as a way to get information. It’s part of a much broader discussion those of us in higher education have been having for a very long time – how should students use online sources for independent research?
When I was in high school and college (a phrase that automatically makes me feel very, very old), we used things called books that were kept in places called libraries. It was a pain in the ass to go to the library and to read through a book to find the information you needed. But the upside? Most of the books were written by people who knew something about the subject. They’d gone through an editorial process that provided at least some level of quality control. Crackpots were as capable of writing books as actual experts, but there weren’t as many, and academic libraries typically kept the level of crackpottery on their shelves to a minimum.
This wasn’t a matter of censorship. It was making sure students had access to reliable information and didn’t waste their time with garbage. If you wanted to learn about the pyramids of ancient Egypt, you would find books by archaeologists, Egyptologists, and historians, maybe along with some structural engineers. If you wanted the aliens-built-them crap that started appearing in the 1970’s, you could still find it – just not in an academic library. Or not as much of it, anyway.
That started to change when I was in grad school. This thing called the Internet became available to the public. A lot of what was out there was free, and within a few years, academic journals began posting peer-reviewed papers as PDFs online. One could scan for particular words! We no longer had to take stacks of bound volumes to a photocopier and spend a bunch of money making inferior copies of the original! And we could do it on our own computers!
At the time, I began noticing the downside to the Internet. It was touted as the great equalizer because anyone can post things there. The problem? Anyone can post things there. Many of those who do shouldn't. The severity of the situation became even more clear when I became faculty, and it's only gotten worse since then.
Let me give you an example:
Several years ago, I was contacted by someone involved in PBS’s Eons Youtube channel. It’s a series of short videos (~10 minutes) covering some aspect of ancient life – human origins, dinosaurs, extinctions, climate change, the origins of life itself, and so on. They were doing a video on hoofed crocodiles, and they wanted to know if they could use a photo I’d published.
A couple of things before I continue. First – it’s going to look like I’m dumping on Eons, but I want to emphasize the very high quality of the series. It’s excellent! The people who put these videos together really do their research, and it shows. They’re informative and done in a way that experts and non-experts alike get something from them. These are a serious public service on the part of PBS, and I highly recommend them to anyone with an interest in the topic.
Second – hoofed crocodile? Yes -sort of. These are the planocraniids – a group of crocodylians with no particular close living relative. They had the same number of fingers and toes as any other crocodylian – five up front, four behind – but in later forms, each digit ended with a hoof-like structure. Their skulls were laterally compressed and deep. Their teeth were flattened and, in later forms, serrated, much like the teeth of a predatory dinosaur. Their fossils range in age from ~60 to ~45 million years ago, after the Age of Dinosaurs. It’s widely thought they were less aquatic than their living relatives and hunted prey on land. (Some of you may know of them as pristichampsines or pristichampsids – older names for the same group.) Planocraniids are known from North America and Eurasia; groups of crocodyliforms similar to these occurred in South America and Australasia, but they’re unrelated.
So, anyway – the person on the phone asked if they could use a photo I’d published. I was perfectly OK with this and offered to send them the original versions, were were higher in resolution and (unlike the published version) in color. And as we discussed it, I casually asked, out of curiosity, what the video would be focused on.
I expected “the fact that it was an effort by archosaurs to retain the large-land-predator role after large theropod dinosaurs died out,” which is a totally cool thing.
Nope. “That it was bipedal!”
I don’t remember what I actually said, but it would have been something like, “Oh, good grief. That hraka?” (The actual words would have been somewhat spicier.)
In the late 90’s and early oughts, a German paleontologist published a series of papers on these animals, which were still called pristichampsines at the time. In one of them, he argued that they could be bipedal at high speed, much in the way some lizards are.
If you’ve encountered a basilisk (a.k.a. “Jesus Christ lizard”) or a collared lizard, you may have seen them scooting away on only their hindlimbs. (There are lots of videos on Youtube showing this.) These particular lizards are ambush predators that hunt comparatively large prey. Collared lizards, for example, prey on other lizards. Hence, their hindlimbs are long and powerful. This allows them to leap from a hiding spot and pounce on their prey. Their forelimbs are much shorter. When they’re walking, all four limbs are involved. But it’s hard to keep their step cycles in phase at higher speeds, so when they run, they rest their forelimbs on the sides of their bodies and take off on their hindlimbs.
I’ve seen plenty of planocraniids, including those that this German colleague used. And the idea of them being bipedal is laughable. Their hindlimbs are indeed longer than their forelimbs, but not to the degree seen in collared lizards. (They’re longer in modern crocodylians, too.). And the limbs overall are longer than in living forms, too - but not that much longer. Planocraniids were also far bigger than these lizards – not gigantic in crocodylian terms, but big ones would reach 2 to 2.5 meters (6 to 8 feet). The muscle attachment sites are indeed hypertrophied (larger than they should be), but the limb proportions are way off for something that’s going to run bipedally, even for a short distance. These animals also had extensive (and heavy) bony armor, which bipedal-capable lizards lack, and the center of mass was further to the front than in these lizards.
No one took this seriously. And none of us really did anything about it, either. It was written for a relatively obscure German journal, in German, and it wasn’t available online for a long time. It was such obvious nonsense that it didn’t seem worth a response. We all had better things to do.
(The paleontologist who published this work was later caught stealing books and fossils from museums, and sometimes from private libraries. He spent time in prison. No idea what he’s doing now, but whatever it is, it’s not paleontology.)
Fast forward a few years, and a British sci-fi series called Primeval does an episode featuring one of these animals. The series’ premise is that a time rift in the UK allowed animals from the distant past and even some from the distant future to show up and cause mayhem. The planocraniid in this episode was shown walking – not running – bipedally, and it was doing so in the manner of Godzilla with a vertically-oriented vertebral column. The dude who first proposed it at least had the good sense to show it running with its vertebral column mostly parallel to the ground, which is how bipedal-capable lizards run.
We saw this, and we chuckled. Primeval was science fiction – not a documentary series. We didn’t think it would have much of an impact.
Until I got the call from Eons.
I was curious where they got this information. They said many viewers had asked them to make an episode on planocraniids specifically because it was bipedal. OK – but where did they get it? There aren’t many textbooks in my field, and none of them made this claim. I couldn’t find anything in the peer-reviewed literature repeating it, either.
I finally found the source – Wikipedia.
Evidently, someone had “updated” the page on these animals to include the bipedalism crackpottery.
This led to several conversations between me and people at Eons, who were in a pickle - they’d already commissioned the art for the video. They weren’t acting out of laziness or gullibility – they have a strong track record of getting it right through their own research, and in this one instance, a source that might otherwise be reliable turned out to be bunk. And they they acted responsibly – they delayed the video by a couple of days so they could re-write the narration, which added skepticism to the claim. Still, had an enthusiast not added a bogus claim to Wikipedia, it might have been avoided.
They asked me why we hadn’t rebutted the claim. I explained the reasons – it was in an obscure source; it was flagrant nonsense that, we thought, would be evident to the average middle-schooler; and we actually have better things to do. No one thought it would be taken seriously. And like everyone else, I can devote my career to correcting errors in online sources, or I can actually teach classes, conduct research, attend professional meetings, advise grad students, and help with the shared governance of my institution.
This, and other incidents, have given me a healthy wariness to sources like Wikipedia.
I’ll also add this – two or three times in my career, I’ve published a paper describing a new extinct crocodile, and it’s gotten a modest amount of media attention. In all cases (and I checked!), the Wikipedia pages on these crocodiles were up within an hour or two of the paper being released; none of the co-authors of these papers was involved in writing them; and they were full of errors.
I can sometimes tell when a student is using an online source to replace the notes they should have taken, but hadn’t. The answers will come out of left field. They may be out of date or just flat-out wrong, and they ignore instructions in the question. I give them a picture of a shark skeleton, and ask how they know it belongs to a particular group of sharks. I ask them to limit their answers to structures they can actually see in the picture. The answer might involve the way the pectoral fin attaches to the body, but I’ll get students telling me it’s based on how certain nerves leave the braincase. When I ask them to point to these openings on the picture, they’re at a loss. They looked up “characters found in this particular group of sharks” or something, and they repeated what they found.
This can get even worse when dinosaurs are involved. One of my undergrad courses focuses on dinosaurs. There are lots of dinosaur groupies out there. (They’re a thing.) They put all kinds of stuff on their own dinosaur web pages. Some of these sites are actually useful, but many are not. In fact, they’re misleading to those trying to use them. They’re often written by younger people who’ve memorized lots of facts, but don’t have enough expertise to understand what it means, or to add the appropriate level of nuance based on the current literature.
On one hand, I can’t really be too disappointed in these results. These students are conducting independent research, and that’s to be applauded. (That they aren’t taking notes in class, of course, shoul not be applauded.) But because they don’t have a background in the field – these are mostly students who are interested in the subject, but aren’t planning on making it their career – they don’t have the toolbox needed to determine what’s reliable and what isn’t.
Many colleges contribute to the problem by encouraging students (and faculty) to do everything online. They cut back on journal subscriptions and book acquisitions because so much information is available online (and because journal subscriptions and academic books are seriously expensive).
And it’s getting worse. Academic institutions are being approached by tech companies who want us to use their new AI applications. Very, very few instructors want to do that. We see too much opportunity for academic dishonesty, and when I’ve done searches with AI engines, the results are sometimes hilariously wrong. But administrators don’t necessarily see it that way. This hasn’t really been a problem where I am, but many colleagues have encountered it.
Want to know why conspiracy theories flourish like never before? It’s because anyone can put stuff online. Conversely, want to know why confidence in news sources is dropping? Lots of reasons for this, but the fact that AI-generated images can look so bloody real is one of them.
I don’t have an answer. Or at least, no one wants the answers I would suggest. Crowd-sourced references like Wikipedia need to be carefully monitored and reviewed by real experts. We need to add courses (above and beyond the many students already take) in high school and college on critical thinking – something that would get politicized in a hurry. (Everyone likes this idea until it comes time to plan the course out, at which point one political side or another, and you can guess which one is most frequently to blame, raises objections.) We also need to limit the damage social media can do by spreading misinformation – though I’m really not sure what that would look like.
Just a few thoughts.
= new reply since forum marked as read
