This morning, The Child came into the kitchen waving her iPad excitedly. “Mom! There is an exciting new animal in America! The electric pangolin!”
Naturally, this was also thrilling to me. But then she showed me a news article from MSN news:
Scientists Identified an Electric Pangolin in Mojave Desert
In a stunning scientific revelation that has shocked both wildlife biologists and cryptozoologists alike, researchers claim to have discovered what they’re calling an “electric pangolin” in the remote regions of the Mojave Desert. This discovery has ignited fierce debate within the scientific community, with some experts questioning the validity of these findings while others hail it as a potential breakthrough in our understanding of evolutionary adaptation. The alleged creature, which appears to combine mammalian characteristics with unprecedented bioelectric properties, has become the subject of intense scrutiny and investigation.
Now, this sounds super amazing: a real-life Pokémon! But reading through the article with her, I was immediately on alert, and walked her through why:
- The original source is “animalsaroundtheglobe.com”, not a more mainstream news source
- There are no links to outside sources in the article
- The lead scientist is billed as coming from the “Institute of Desert Ecology”, but there’s no sponsoring organization, such as a university, named for the institute
- Another scientist quoted in the article is a palaeontologist, which seems like an odd discipline to be commenting on a new species find
We then looked up the names of the scientists mentioned and couldn’t find them. We also searched for “electric pangolin Mojave desert” and found no other mention of an electric pangolin in the news – this site was the only one publishing this claim.
“So it’s probably not real?” she said, crestfallen.
“It’s probably not real,” I said.
We went out for waffles to help cope with the disappointment.
Now, this was a nice teachable moment for The Child about double-checking a story that sounds too good to be true. If only the mothers of numerous economists and journalists had been present to do something similar for them when confronted, last autumn, with an exciting study about how using AI made industrial scientists vastly more productive.
Back in November, a paper by a grad student at MIT popped up on arXiv, a database of research papers undergoing review. This paper, “Artificial Intelligence, Scientific Discovery and Product Innovation,” was by a second-year econ PhD student named Aidan Toner-Rogers. The paper reported on a randomized trial looking to examine the impact of using AI tools for materials science research.
Toner-Rogers claimed to have enrolled more than 1,000 material scientists from a major US corporation in his study. Through various types of sophisticated research methods, he further claimed to have found that the researchers who used AI became much more productive: more materials identified, more patents applied for, more prototyped products made.
This pre-print had not yet been subject to peer review, but that didn’t stop multiple news outlets (hello, The Atlantic) and at least one economist who’d won the Nobel Committee’s Sveriges Riksbank Prize – the same prize Paul Krugman won in 2008 – from praising it to the skies. Per a WSJ story (archive.is link, no paywall) Toner-Rogers presented his findings at a National Bureau of Economic Research committee. He submitted it to the Quarterly Journal of Economics for publication.
On Friday, MIT disowned the paper (and Toner-Rogers). In a statement, they said: “Even in its non-published form, the paper is having an impact on discussions and projections about the effects of AI on science. Ensuring an accurate research record is important to MIT. We therefore would like to set the record straight and share our view that at this point the findings reported in this paper should not be relied on in academic or public discussions of these topics.”
Turns out there were a TON of red flags about this paper that, if it had been shown to actual material scientists and not economists, would have been pointed out fairly quickly. Writer Ben Shindel (who has a material sciences background) points out all of these in detail at The BS Detector (Substack, sorry). First and foremost is the screamingly obvious one: how does a random 26-year-old PhD student, even one from MIT, get access to 1,000 scientists at a major corporation? The only companies with that kind of workforce would be something like 3M (Shindel notes in a postscript that there are clues Toner-Roger claimed he was working with Corning). They’re going to let some dude poke around their work by himself?
But far too many of the people interested in unlocking some kind of infinite growth machine with AI don’t want to ask those questions. They want to believe in the electric pangolin, in spite of all the signs that it’s too good to be true.
The waffles were terrific, by the way. Open thread.
Baud
An electric pangolin? So awesome.
trollhattan
I’ll give it this: far more adorable than tabletop fusion.
Baud
Also, too, excellent momming.
Rose Judson
@Baud: It will always be real in my heart
(ETA: Thanks, I have my moments.)
WTFGhost
It is truly amazing how easily people fall for the “you know more than you think you do” fallacy. And, it’s like, an economist should be thinking precisely of “I’m seeing great metrics, but do I know what those metrics *mean*?”
(“When will I use this crap?” by the algebra student is now answered with “when you dig into your employee metrics, both to see how to game them (if needed) and how to ensure they’re hit by competence, not gaming, where possible.” )
WTFGhost
@Rose Judson: Well, now, it could have been a pangolin about to ask you to prom, and if you’re into TENS stimulation… but *NO—ooooo—OOOOO!*, REALITY had to barge in and eff things up. Again.
Another Scott
Saw some bloots today about the MIT thing. It sounded exactly like what one would expect in a young author’s AI propaganda piece that was hoping to ride the wave. Do we know that he actually wrote it, rather than having Groot or ChatBS or whatever do it??
I’m glad you covered the story here, and did it so well!
Thanks.
Best wishes,
Scott.
JoyceH
So our hemisphere still has no pangolins? None at all? I’m going to go away and pout for a while.
Anyway
Not “Mummy”?!
Rose Judson
@Anyway: Without my encouraging her to, The Child has an American accent (and it’s common to use “mom” instead of “mum” here in Birmingham).
Gvg
Asian and African species has a new relative in the western US seems unlikely right away, plus electric?
AI has been obviously overhyped for several years now. I don’t see how it hasn’t become automatically a suspicious word, and real engineers working with them don’t have to overcome that to get investors. Why do people WANT to believe in this fairy story? Because they do. Most people want this specific nonsense to be true. That’s why the problems it would cause if it did happen aren’t being taken seriously. When you are believing in a fairy story, you don’t have to pick a sad ending.
Anyway
@Rose Judson:
My American friend and her Aussie husband live in London and their young twins have the cutest English accent— complete with Mummy and Daddy.
cain
@Gvg:
Businesses are laying off people thinking it can replace jobs with AI.
I have yet to see how you can create production worthy code. I think we are going to see a serious quality drop in software and customer service.
Eventually someone will compete using real humans plus AI assistants at least in customer service.
Anyway
OT went paddleboarding on the NorthEast river — first time this year. Rain stopped right on time, sun came out and it was gorgeous on the water.
Now I am exhausted from the sun and being with people — vicariously enjoying the Ohio meetup.
Omnes Omnibus
I shudder to think where you need to put the batteries in one of those.
JoyceH
@Anyway: There were some news articles a while back about American kids picking up British accents from Peppa Pig.
Barney
“Electric Pangolin” is still an awesome band name. Or “Chasing the Electric Pangolin” for an album. Or a rock-climbing route.
RP HICKS
I read the bio for Dr. Eleanor Vasquez, who is a doctor of biodiversity in Queensland Australia. She mentions “the power of storytelling to make scientific knowledge accessible to all.” I think the book is closed on the Mojave electric pangolin.
Pete Downunder
Here downunder we have the very dangerous drop bear, thylarctos plummetus. As reported in Australian Geographic this territorial animal ‘drops’ from the treetops on unsuspecting bush walkers and tourists. Interestingly, the drop bear is so elusive that no one has ever captured or photographed one, alive or dead.
A relative of the koala, the drop bear is much bigger than the standard phascolarctos cinereus and is more likely to be the size of a large dog or leopard. With course orange fur that has dark mottled patterning, the drop bear is equipped with sharp claws for climbing trees and holding onto prey.
Sadly, the drop bear too is entirely fictitious.
chemiclord
That’s really what it boils down to. Humans hearing what they want to hear, and just rejecting anything else like reason or critical thinking.
It’s not a lack of education; most of the people jumping on this train are and were (allegedly) well educated people with a critical grounding. It’s not “stupid Americans,” as a lot of people around the world happily jumped in with both feet.
It’s just… humans in general, and probably a significant majority of the human population, are happy to eat bullshit as long as it smells how they want it to smell. And I don’t know what the hell we can do about that.
Omnes Omnibus
@Pete Downunder: The hodag is famous in my neck of the woods.
RP HICKS
Here is the link to which I referred.
Dr Eleanor Velasquez – Centre for Biodiversity and Conservation Science – University of Queensland
kindness
Good to keep her on her toes Rose. She’ll need it when someone tries to sell her a bridge in London real cheap.
Pete Downunder
@Omnes Omnibus: We need those here, but I suspect the spoilsports at bio security won’t let them in
prostratedragon
Electric pangolin! Great lead-in to your point, and a new metaphor as a bonus. ( And yes, the obvious question about the MIT paper was so very obvious that I wonder why economists A and A didn’t spot it.)
Omnes Omnibus
@Pete Downunder: Our cabin is near Rhinelander. I keep hope that we will see one on our trail cam. Just deer, foxes, and a couple of wolves.
RepubAnon
@Barney: Electric Pangolin is indeed an awesome band name – say, for a heavy metal mandolin band.
The Four X continent on DiscWorld has drop bears – so make sure you’re wearing your wizzard hat.
prostratedragon
A sighting of the elusive cattus bodeganicus.
Redshift
Excellent post, Rose! This is a really good way to explain this that hopefully will be less likely to provide a defensive reaction from people who get taken in by hype. I’m definitely going to keep it bookmarked to use in replies to “amazing new discoveries” in the future.
Also, The Electric Pangolins is my new band name. 😀
BeautifulPlumage
I saw your bsky post & love this expansion on it. Good explainer. Thank you.
And of course, the nice round number of scientist wasn’t a give away to anyone doing (or reading about) randomized trials.
trollhattan
@Pete Downunder:
Drop bear is much more poetic than the somewhat obvious alternative, mediumfoot.
trollhattan
Now that Trump is threatening Walmart how does Walmart feel about supporting Trump?
“Still better than that monster woman, can you even imagine?” considered the most likely thing they’re telling themselves.
pluky
Any hope this poor child had about a successful academic experience at MIT just got flushed. No one is going to give credence to anything this guy has to say about anything. I mean, when the administration comes out and decisively disavows your work. Ouch!
Elizabelle
Great thread on media literacy, Rose. Laughing that The Atlantic fell for the AI crap, but not really. They are already wired for the AEI (American Enterprise Institute: Norm Ornstein is the only good thing to ever come out of that place, and he is a severe outlier).
Cheryl from Maryland
Over a decade ago I pitched a traveling museum exhibition on electric animals (they do exist; they are mostly aquatic). My bosses told me it was a boring topic and no one would be interested. HA HA HA HA! Of course people are interested.
trollhattan
@Cheryl from Maryland:
Our sadly departed tropical fish emporium had a decades old electric catfish on display. “Not for sale” for a whole host of reasons. Named “SMUD” after our local utility.
Sure Lurkalot
Did Toner-Roberts use AI to fabricate his research and/or write his paper?
The more these “tools” are promoted and pushed as time savers, help mates and panaceas, the more skeptical we should all be about all of the claims to their efficacy.
B school has advanced the idea that the only business purpose is profit, the only loyalty is to investors; the product, its excellence and usefulness, the customers, their needs and satisfaction, are superfluous to the goal. The tech lords have shitloads of money riding on the AI as miracle narrative, they care not one whit if it “works” or who is helped or harmed by it.
RevRick
@Gvg: I’m shocked!
catothedog
Note that one of the Ph. D advisers for Toner-Rogers is Acemoglu, who got the Economics Nobel in 2004 – that colonialism does not matter
Another Scott
@BeautifulPlumage: Reminds me of a bloot I recently saw on Mastodon with a scatterplot graph of something like last digit in a precinct’s count vs percent turnout, in a whole bunch of countries. As I recall, above about 80% the most frequent digits were 0 or 5, a clear sign of tampering with the numbers (as random numbers don’t work like that).
Yeah, humans like round, easily divisible numbers and groupings. But nature doesn’t. One should think long and hard about faking numbers. (Some think that Mendel might have messaged his numbers in his pea crosses a little…)
Thanks.
Best wishes,
Scott.
J.
So there are only gas-guzzling pangolins, not electric ones? I’m disappointed. But maybe they’ll come out with a hybrid. ;-)
hells littlest angel
Another clue is that in an interview the electric pangolin would not stop talking about white genocide.
Baud
@catothedog:
I don’t think that’s a bad approach for academics.
The bigger risk IMHO is from academics who unintentionally or intentionally present value judgments falsely clothed in objective or rational analysis.
Rose Judson
@hells littlest angel: Nice mashup.
UncleEbeneezer
So in other words, this story is about as well-sourced and trust-worthy as the new book about Biden’s “cognitive decline” (eye-roll)
Rose Judson
@UncleEbeneezer: Honestly if I think I hear anyone else talk about Joe Biden I will swallow my own tongue.
satby
@Rose Judson: Well done you, Rose! Care to license an online course for people to use in teaching their more gullible relations how to read critically and verify sources?
dmsilev
@pluky: One of the links in the article goes to something from the MIT Econ department that says, among other things, that the student is no longer affiliated with the university.
You have to work pretty hard to get kicked out of a PhD program. It is doable though…
No One of Consequence
@Baud: Mine’s acoustic, but I have considered putting a PZM mic on it for recording.
-NOoC
prostratedragon
@dmsilev: One would hope that that level of academic dishonesty would do it.
Lynn Dee
@dmsilev: Assuming there even is such a person.