The Last Word On Nothing

I never knew it needed explaining until someone asked me — why is singletrack so much more fun than wider trails like double track or dirt roads? We’re talking here about mountain biking and the allure of the singletrack trail — a narrow path, usually 18 to 24 inches wide, that meanders through a given terrain.

The most obvious answer is that a singletrack is more aesthetically pleasing than a wider road.

But the sheer joy of navigating well-designed singletrack on a mountain bike goes beyond aesthetics. I’ve also done a lot of hiking in my life, but rarely have I felt compelled to let out an audible whoop while traversing a trail on foot.

Last weekend, while biking Crested Butte’s world-renowned singletrack, I hollered joyful cries more than once. I couldn’t help myself. I was that euphoric. The epic views of the West Elk Mountains surely contributed to my jubilation, but the beauty felt like icing on a cake. The joy I’d experienced was as much physical as cerebral.

Continue reading →

I used to work at the Science news office in the American Association for the Advancement of Science office in DC. As anyone who’s been there will tell you, there’s something special about the AAAS building — the dark polished rock exterior, the spiral staircase that, as you climb from floor to floor, takes you past framed Science covers illustrating more than a century of scientific achievements.

What AAAS does carries weight. Which is why I was intrigued — very intrigued — to hear last year that the Science Press Package team had decided to run an experiment testing how well ChatGPT Plus could write its press releases.

I would probably have been alarmed by the news if I didn’t know the Science journals’ communications director, Meagan Phelan. I admire Meagan — we’ve been good friends since I worked in the DC office — and you will not find a more thoughtful, intellectually rigorous and caring person. If Meagan signed off on this experiment, I thought, it wasn’t just a reckless bid for efficiency, but a genuine attempt to understand LLMs better.

The year-long experiment began in 2023 and was published in September 2025. Curious how it went, I sent Meagan a message and she suggested I talk to Abigail Eisenstadt, one of the Science Press Package (SciPak) senior writers. Abigail led the study and wrote this white paper to describe its findings. (She also just did an interview about it on the Grammar Girl podcast.) Abigail let me ask her a bunch of questions and listened to my own rambling thoughts and fears about AI-powered science writing, and I came away feeling much better.

There’s lot to be worried about these days — including that generative AI is busily burying us alive in crap. But I am grateful to patient souls like Abigail, who set out to explore in a measured and deliberate way what these tools can and cannot do when we ask them to perform nuanced human tasks. We need thoughtful folks doing these experiments, if only to keep delineating and articulating the precious — and still distinct — boundaries between AI algorithms and our own remarkable human minds.

Here’s what Abigail told me about Science’s ChatGPT Plus experiment, and why it left her feeling more secure in her job, not less.

Note: This conversation was edited for clarity and brevity at 11:45pm, by my sleepy human brain and typo-prone fingers. Cases in point: 1) Mortifyingly, an earlier draft of this misspelled Meagan’s name with an h. 2) An earlier draft of this also failed to distinguish between all of AAAS communications and the Science journals press team. This interview pertains specifically to the Science press team, not the whole organization.

Emily: I know you’ve described this in detail elsewhere, but let’s start with the basics. What were the nuts and bolts of this experiment?

Abigail: Understanding the gist of what we did relies on understanding how we operate our press packages. Every Monday, we start drafting press briefs — three to six of them, depending on the week and our workloads. We write up summaries from our journals, and different writers represent different journals. I do the open access journal, Science Advances, but we also have specialty writers on Science Immunology, Science Robotics and other sibling journals.

What we wanted to do for this study — and this was back in 2023, before a lot of common knowledge about prompt engineering was around — was to see if Chat GPT Plus could use our outline structure, the template we rely on to write our summaries, to write research summaries too.

It wasn’t necessarily a yes or no question. It was — how much can it do it? If it can’t, are there ways it could evolve in the future so that it could?

What we did — because we’re already creating content every week — is take something that had already been published, so the embargo was already past, and run it through Chat GPT Plus, using three different prompts for summaries. We used the privacy mode option, so it wasn’t training anything.

One prompt was for a very general summary: Just tell me what the study is about. The other was more abstract, but in a way that you would hope would be accessible to journalists. The third was inspired by our own very specific press release template.

The LLM would produce these summaries, and I would send them back to the author who had written the human content and ask them to compare it to theirs. And it was very clear from the start of this project that there was no danger of us being supplemented.

Continue reading →

I used to think AI was a hyped-up distraction. I thought it would do a clumsy job of things, and be annoying, but mostly harmless. I’ve changed my mind.

What initiated my change of mind was playing around with some AI tools. After trying out chatGPT and Google’s AI tool, I’ve now come to the conclusion that these things are dangerous. We are living in a time when we’re bombarded with an abundance of misinformation and disinformation, and it looks like AI is about to make the problem exponentially worse by polluting our information environment with garbage. It will become increasingly difficult to determine what is true.

Let me show you what I mean. I tried using the Google Docs AI tool to help me write this piece. Here’s what it said:

This is just weird. I’ve never changed my mind about the idea of "everything in moderation" and I’m not sure I’ve even ever said anything about "everything in moderation" publicly, much less declared that I’d changed my mind about it. Nor have I said that I believe it’s better to avoid processed foods and sugary drinks. Sure, that’s probably a good idea, but this is not something I’ve ever framed that way. It feels like it’s inventing thoughts for me. It also uses the real title of my book, but a fabricated subtitle that’s totally off the mark. Huh?

I tried again.

More complete nonsense! Nothing in this article is true. The article it references does not exist. This is just a pile of crap. And yet — I can imagine that this information might be plausible to someone who didn’t know me well. I mean, I’ve written quite a lot about sports and statistics, and I’ve done a lot of debunking stories, so the framing of the fake stories (pointing out flaws in popular ideas) isn’t unreasonable.

I did note that Google’s AI bot comes with a disclaimer: "This is a creative writing aid, and is not intended to be factual." Ok sure.

I had similar experiences with ChatGPT, though I’ve noticed that the newer versions are less prone to obvious errors than the earlier ones I tried last spring. It’s still easy to trip up though. Consider this silly question I just asked it about my old boss.

This one made me laugh so hard I had beer coming out my nose.

Some people call these errors "hallucinations," but Carl T. Bergstrom and C. Brandon Ogbunu don’t buy it and neither do I. Chat GPT is not hallucinating, it’s bullshitting, Bergstrom and Ogbunu write at Undark:

ChatGPT is not behaving pathologically when it claims that the population of Mars is 2.5 billion people — it’s behaving exactly as it was designed to. By design, it makes up plausible responses to dialogue based on a set of training data, without having any real underlying knowledge of things it’s responding to. And by design, it guesses whenever that dataset runs out of advice.

What makes these large language models effective also makes them terrifying. As Arvind Narayanan told Julia Angwin at the Markup, what ChatGPT is doing "is trying to be persuasive, and it has no way to know for sure whether the statements it makes are true or not." Ask it to create misinformation, and it will — persuasively.

Which gets me to the latest disturbing example of how AI is going to make it much, much more difficult to parse truth from bullshit and science from marketing.

In the latest issue of JAMA Ophthalmology, a group of Italian researchers describe an experiment to evaluate the ability of GPT-4 ADA to create a fake data set that could be used for scientific research.* This version has the capacity to perform statistical analysis as well as data visualization.

The researchers showed that GTP-4 ADA "created a seemingly authentic database," that supported the conclusion that one eye treatment was superior to another. In other words, it had created a fake dataset to support a preordained conclusion. This experiment raises the threat that large language models like GTP-4 ADA could be used to "fabricate data sets specifically designed to quickly produce false scientific evidence." Which means that AI could be used to produce fake data to support whatever conclusion or product you want to promote.

The takeaway is that AI has the potential to make the fake science problem even worse. As the authors of the JAMA Ophthalmology paper explain, there are methods, such as encrypted data backups, that could be used to counteract fake data, but they will take a lot of effort and foresight.

The fundamental problem with AI is that it’s difficult to determine what’s authentic, and as AI cranks out a firehose of generated content and floods the zone with shit, it could become more and more difficult to parse real information amid an outpouring of AI-generated bullshit. If you thought the internet was bad before, just wait.

Note: This post first ran on Nov 11, 2023. Unfortunately, it’s still relevant.

---

*The version of Chat GPT-4 they used was expanded with Advanced Data Analysis (ADA), a model that runs Python.

Because I’m traveling today, to the annual science writers’ conference, I’ve taken the unusual step of running a guest post without clearing it with the rest of the LWON team. My sweet baby Chihuahua mix T.S. Eliot Nestor (pictured above) asked to respond to Our Jenny’s post this week (Little Dog Big Heart), in which she called small dogs “perpetual barkers of the piercing sort” and said they have “atrocious” breath, as well as strongly implying that they are not real dogs. I took dictation and have added rough translations.

Take it away, Eliot:

Woof. Woof woof woof woof woof.
Dear Auntie Jenny:

Woof woof woof woof woof woof.
You are wrong about me and my fellow smalldogkind. [Ed. note: I should clarify that he doesn’t actually know any other small dogs, or any dogs, because his idea of how to greet a dog is to lunge and bark wildly, and they don’t like it.]

Woof woof woof woof woooooffffff whine BARK BARK BARK.
About the “atrocious” breath: I agree, and I bet my breath would smell more interesting if my mom would let me eat all the treasures I find on our walks. [Ed.: Dogs are gross. And despite his dislike of the toothbrush, his breath somehow smells great.]

Woof woof woof whiiiine whine whine. Bark! Whiiinnne.
Am I “loyal to one and hateful to the rest of us”? No. I love my two humans equally, although Daddy is my current favorite, and I love everyone else too until they stop petting me and try to walk away, and then I hate them. This is not “hateful,” it is rational, because they are bad.

Woof woof woof woof woof bark bark bark bark.
If you came over, I would probably bark at you, but “piercing” is unfair.

Whiiiiiine yip yip yip.
You should come over some time and let me bark at I mean charm you.

Woof woof woof. Woof woof. Woof! Woof woof woof.
In conclusion, I am definitely a real dog, because I love walks and bully sticks. [Ed.: The bully sticks are very small and he will do anything for a Churu cat treat]

Whiiiine
Love, Eliot

Woof woof wooof woof woof woof.
P.S. Thank you for Blue Dog.

Photos: Eliot’s official photographer, Helen Fields, obviously

I’ve never been a small-dog person. I get that they have a touch of cuteness to offer; they are, after all, tiny, furry things with big eyes, and some have silly faces, floppy ears, and goofy antics. But most that I’ve met are just perpetual barkers of the piercing sort, or teeth-bearing growlers, trying to be something they’re not. They’re loyal to one and hateful to the rest of us. And their breath is atrocious.

Me, I’ve had a Golden, a Weimaraner, and two Jindos, among others. Big dogs just seem more like real dogs to me.

A good friend of mine is a little-dog person through and through; I forgive her this misstep and I applaud her willingness to take needy ones, misfit toys if I ever saw them, into her home. I do have limits. Chihuahuas are a nope. One of hers, especially, hated me up until she could no longer see or hear, and even then I got a mean vibe from her; another was tolerant, barely, despite my gentle overtures. Neither struck me as adorable, despite the ears. They only made stronger my big-dog bias.

But recently my friend came home with a wee muppet, genetic tale unknown (but no Chihuahua DNA, methinks), who delights in the world and has been squirming into an empty pocket in my big-dog-packed heart.

She’s so small, only a cat harness will fit her—and she doesn’t need one anyway because when I walk her she trots gleefully by my side, looking up at me every few steps to make sure I’m still there. Her face is ridiculous, her neck recalls a meerkat, and her ear flops and foot taps—especially during her bestest Zoomies—are irresistible. Her fluffy face and naked body/rat tail are a hilarious combo. Most of all, the pup is never in a bad mood. She never stops smiling. Every greeting is a happy dance, tap tap tap, and an immediate request to be held. All she wants is to be in your arms. And mine. (Her breath is indeed a horror show, but we’ll let that go for now.)

Please know that for the most part, nothing has changed. I still gravitate toward the bigger, wolfier canines. Lately I’ve also come to fancy sighthounds, especially those lanky, huge-eared Podencos from Spain. Great Danes, too, have made my wish list.

But now, when I visit my friend, my first act after kicking off my shoes is scooping up that tiny animal, who is already at my feet doing circles and leaps, and cuddling her furry sweetness to my chest. If I stand, she give me a kiss and then rests her head against my shoulder. If I sit, she becomes a warm puddle in my lap. She doesn’t wiggle to get away, and rarely is there a reason to put her down. We’re both happy where she is.

A conversation struck up at a writing workshop a few weeks ago between me and a computational biologist who studies the molecular biology and population genetics of the HIV virus and other virulent diseases including Ebola, Hepatitis, and Covid. She’s what you would call a veteran virus tracker, compiling and interpreting viral genetic sequences and medical alerts from around the world. She’s one of those who raises her finger and says something’s going on here.

We were eating lunch beneath the copious shade of an alder tree along the Gila River in southern New Mexico. She was talking about her HIV work with studies being halted and USAID shut down during a dismantling of public health architecture. This dismantling, she said, was creating a perfect environment to enable rapid evolution of drug resistant HIV, nothing better for diversifying strains than medication slowing and being misused and underused worldwide. She explained that access suddenly ending means many people with HIV will share what few doses of medication they have left with loved ones, or get inadequate doses, or get ineffective treatments on the black market and this creates the possibility of an evolutionary flashpoint as mutations in medication-avoidant viruses look for hosts. This is how HIV is unleashed back into the wilderness.

I often get scientists in these workshops. They are the ones here to relearn storytelling, the challenging ones, but she was making it easy for me. I was sending her off to write about anything but science. No talk of virulence or pandemics, we instead wrote about the sound of moving water. We climbed a ridge and used boulders for chairs, looking across folds of wild country, furry mountain ranges dropping into canyon after canyon, which we described in our notebooks so someone far away in a different time might see the picture. For days we did this sort of thing. After field class we’d return to an old wood lodge that leans slightly, squeaky wood floors, front porch dressed up with rocking chairs and couches. It was a casual, unguarded atmosphere, and I did ask her permission to write this, especially since she called what she had to say a warning, a red light flashing on the control panel.

Continue reading →

I first wrote this post in October 2020, when my kids were much younger and Halloween decorations loomed from every corner.

*

I forget this every year—in October, there are places where it is no longer safe to walk. If we want to go to our friend Peter’s house, we can’t go up the street and around the corner as we usually do. If we need to get to daycare, we have to turn and walk in the exact opposite direction first, before U-turning around the block. And Ezzy’s house? Forget it. That street is riddled with danger.

The Halloween decorations are out again, and some of them are scarier than others. Some houses have cheerful pumpkins. Others have black cats. Then there are the skeletons, the witches hanging from trees, the bloody severed body parts. On Ezzy’s street, there is a demonic baby with red eyes crawling on the rooftop. That’s why we can’t go over to Ezzy’s until November.

Every year, I forget how much terror these decorations create for my children. Sometimes we can protect ourselves by crossing the street. Sometimes they close their eyes and I hold their hands until we pass. Sometimes riding a bike really fast helps. But there are some houses, some days, where we can’t pass by at all.

I love Halloween. Sure, I feel weird about my kids getting a lot of candy, and worry about things that might be even more creepy than the demonic baby. Still, it seems like a generous thing, to make your house look sparkly and spooky, to show a brave face to tiny trick-or-treaters, to come up with creative ways to celebrate, six feet apart. One of our neighbors has dozens—many dozens—of pumpkins, skeletons that are dressed in witch costumes, a giant statue with a pumpkin head, purple and orange lights around their yard. Another creates an elaborate structure—some years a castle, others a dungeon, and hangs dummies from the trees. (I confess that this part makes me nervous—I’m worried someone will get in a car accident, thinking that there is actually a person hanging in the elms.) And there’s a family near the elementary school that every day in October—every day!—re-positions a pair of skeletons into a new scene.

My kids love those skeletons. They love seeing the creative things that the family comes up with. It’s a gift to the neighborhood, all the decorations, even the scary ones.

It’s a gift to me, too, because it reminds me that even during the rest of the year, there are houses I cross the street to avoid, places that bring back memories, whether they be skeletons or something more like a blow-up candy corn. Sometimes there’s a specific reason—the man there had once shouted at you to get out of a tree, or there is a large, unfriendly dog. Other times it’s just a feeling: do not linger here. Other times, houses give off a friendly vibe, whether or not you know who lives within.

Neighborhoods are maps of these feelings, and the longer you’re there, the more they layer over each other. There was that couple who lived in the house with the wisteria and bougainvillea since it was built in the 50s, the large family of caretakers that moved in to help them, and now, the retired officer who had to re-pour the foundation to make everything level again. There’s the other house that was blue, and then was a pale brown, and now is white clapboard with succulents in front. With each iteration, the houses draw me in, push me away, invite me to step a little closer to the fence. There is the sadness of friends’ houses that are now filled with strangers. And then there is the welcome of houses that used to look like empty haunts, homes that are now filled with friends.

*

Top image by Flickr user Vivian D Nguyen under Creative Commons license

I remain mildly obsessed (a nice state to live in) with the mission patches published by the National Reconnaissance Office, which if you’re obsessed enough you can figure out what satellites the NRO just launched. Some of those satellites are, you know, secret. Like this one: it’s for recently-launched constellations (a ton of linked satellites) that are a secret version of the Starlinks. They’re called Starshield. This post first ran March 5, 2021. My mild obsessions are long-running.

We begin, as we so often do, with a tweet.

Jonathan McDowell @planet4589: Interesting that the NROL-44 patch description makes explicit reference to FVEY, the ‘Five Eyes’ spy alliance of US/UK/Aus/Can/NZ.

Brief explainer: Jonathan McDowell is a certified Harvard x-ray astronomer who also keeps an eye on satellites in space. NROL stands for National Reconnaissance Office Launch. NRO sits somewhere in the murky middle space between the defense department and the intelligence community. Its job is to launch spy satellites. The spy satellites are secret and the NRO doesn’t especially acknowledge them. But the satellites are launched on rockets and rocket launches aren’t subtle and come with warnings to pilots, so NRO acknowledges the launches with a statement and a mission patch. For launch #44, the patch was this wolf. An annotated version of this patch says that those five wolves (note the four lurking in the back) "shows the solidarity across the FVEY community." (FVEY is jargon for Five Eyes, 5 countries that cooperate on spying.) The annotation continues: the wolf is howling into space where the satellite is, and the howl is a warning to the wolf pack of signs of trouble.

Well then. Well-a-then. My my my. I have questions, I do.

Me: Jonathan, I saw your tweet about the NRO patch. I have a tiny obsession about these patches because they are so weird and because they give away information. Why do you think they issue the patches?

Jonathan: It’s a tradition, it promotes team bonding which is a big thing in the military. They’ve been told lots of times not to give things away with the patches but after a while it starts happening again.

Indeed it does. A lovely old example, the patch for launch #11:

In 2000 the NRO launched a satellite it wouldn’t describe but issued a patch. One of my beloved, brilliant amateur satellite watchers (they prefer to be called "hobbyists," and as one scientist told me, she’d stopped thinking of them as amateurs long ago) saw the patch soon after launch and given his long years of watching, thought it belonged to a family of radar imaging satellites. The four little arrow-things were satellites in the same family, he thought, the clue being that one was black because it had de-orbited (dropped out of the sky) since. Based on the arrows’ paths in the patch, he predicted NROL #11’s eventual orbit. The mesh outlining the owl’s eyes looked like the kind of antenna the satellite’s signals used. And "We Own the Night," though immodest, was obvious.

People have been studying these patches maybe since 1977, and notice certain repeated images – dragons, eagles, owls – and figured out what the images mean. The number of stars in the patch, for instance, is usually the number of satellites in that family. And in fact, they have long articles and books postulating the iconography of these patches. To be honest, I find postulating iconographies tiring in the same way, as a young English major, I found searching for symbols in poems tiring.

Me: So the next question is, do you have any idea whatsoever that could account for the aesthetic in these patches? Why this combination of comics and fantasy? I associate that combination with basement boys, but surely the guys doing the patches are long since out of the basement if they were ever in it?

And are the patches actually sewed on to anything? are they just collected?

Jonathan: I think the crews in mission control probably wear them.

For the Chandra [an orbiting x-ray telescope] mission, we scientists for the most part didn’t wear patches on our clothes, but we for sure had the sticker versions on our backpacks/briefcases and office doors...

And I think a lot of the engineers in the space program are scifi/fantasy fans

so they are into it

oh, and I admit I have several t-shirts with the STS-93 Chandra patches featured on them

Me: Just googled the STS-93 Chandra patch and realized I have a t-shirt from an astro conference called something like Observing the Dark Ages [the time before the universe universally lit up], and on it is a knight on horseback. Ok. I’m getting it now — a club of people interested in what interests me, my people, and see? I belong.

I have to say though, STS-93 Chandra patch looks pretty normal.

Jonathan: Well Chandra isn’t secret so we didn’t have to be coy

Me: It’s not the coyness that’s weird. It’s all the octopi and ravening eagles and stalwart avengers and ferrety-looking spies, not to mention the ladies whose clothes either start low or end high.

Jonathan: Yeah, I think a lot of comic book fans in DoD

Me: Not reassuring, is it. Alternate realities in the DoD. That would be a good title for an article.

Jonathan: Hah

Hah indeed. Some of my best friends, or rather some children of some best friends, love that comic-book imagery and who’s to gainsay them? But something about the combination of bulging muscles and boobs, and secret surveillances make me itch. I think of people in the intelligence business as living in a reality that if I but knew, would scare me. I think of them as unrelentingly aware that their knowledge of this reality comes accompanied by responsibility. I think that because of this, they are socially reserved, in fact, flat-out bad at friendly conversation. Surely under that responsible reserve isn’t the soul of a pale but ruthless fantasist with whirling eyes? Surely not?

_________

Patches courtesy of NRO, via Wikimedia Commons