Editorial: FasterSkier is Writing Articles Using Generative A.I. It Should Stop.

By Gavin Kentch

We can make this a very short editorial: See the headline. That’s it. Or, perhaps, add something to the second sentence along the lines of, “because it is bad for the sport.” Maybe also “the FasterSkier editor” in the first sentence; this is a lone actor problem, not a sitewide problem.

But, really, that’s the gist of it: Matt Voisin, the longtime editor at FasterSkier and the man who sets the tone for the site’s overall coverage, has for several months now been using some form of generative artificial intelligence to “write” articles for FasterSkier. It is bad for him, bad for his site, and bad for a sport that we both care deeply about. I think he should stop. Many others also think he should stop. I say this as a fan of skiing, not as someone running a rival site, although I cannot deny that that is my position here. The end.

(The only bad actor here is Matt Voisin; everyone else at FasterSkier is writing their articles themselves. Having made that important distinction clear, the rest of this piece generally talks about “the site” or “FasterSkier,” to avoid having this say “Matt Voisin” scores of times and read like quite such a personal attack. But I do need to be clear that no one else over there is doing this. Least of all Nat Herz, probably the best and most ethical reporter I know.)

I do not make this allegation lightly: It is a heavy charge to level against someone who cares about words and uses them to make a living. But, that said, just look at this.

On the one hand:

This is a paragraph from an article written in February 2025, which I truly believe comes from the author’s own hand (no pun intended; yes this is from a roundup of different gloves):

“I have several close friends with Raynaud’s, and selfishly, it was really impacting our ability to ski together, so I started poking around on the internet for products that would allow them to cross-country ski comfortably when it was less than 20’F.  As I sat working at a local coffee shop, like I often do, I even had a random stranger approach me, who had been, a little alarmingly, watching me bounce around the internet, to ask if I had any recommendations for cross-country skiers with Raynaud’s.  At that moment, I knew FasterSkier could help the ski community find a solution to an issue that was more significant than I had imagined.”

Or this, which A.I. would have had a difficult time writing because it does not know these personal details about the author’s life:

“On most race weekends, I take my job seriously — up at 3:00 am EST, fresh coffee made, woodstove loaded, notebook out, logged on to coverage early to make sure I don’t have a technical issue. Then once the race starts, I begin looking for subtle, important moments to describe in greater detail in the race report.”

On the other hand:

This is a paragraph from an article written in January 2026, after the widespread adoption of AI in his writing (and also the move from two spaces after a period down to one):

“On Sunday morning in Oberhof, the snow had the kind of sheen that tells the truth before the clock ever does. It glittered under a clean winter sun, polished by a freeze–thaw cycle that left the tracks fast, brittle, and unforgiving. At 823 meters, tucked into the Thuringian Forest, the course looked generous from a distance—wide lanes, long sightlines—but it was the sort of generosity that disappears the moment a skier commits to a mistake. The margins here were not loud. They were sharp.”

Or this, from December 2025:

“Davos doesn’t reward urgency.

“It tolerates it, sometimes — lets it flirt with the clock through the opening kilometers — but it never forgets. Like a mysterious, wispy cloud, the altitude sits quietly above the stadium, thinning the air just enough that the consequences arrive late, not immediately. The first lap can feel generous. The second is where Davos asks what you’ve really brought with you.”

Or this, from January 2026:

“There are places in cross-country skiing where the scenery tries to soften the message. Val di Fiemme is one of them: chalets tucked into the folds of the valley, the geometry of the Dolomites, the sense that the world is being held in a wide, cold palm. But on Saturday—Stage 5 of the Tour de Ski—the athletes arrived with that gentler Italy already behind them. Toblach, with its familiar corridors of speed and fatigue, was gone. The rest day was gone, too. And in its place sat something more specific and more unsettling: a brand-new sprint course built not for tradition, but for February.”

(Note also the differing stylistic choice as to whether or not there is a space on either side of the em dash; the same human author writing on the same site one month apart would presumably not change their house style like this. Particularly if they ran the site and set the style.)

You get the picture. There are many more examples where this came from.

I am well aware that the same author can employ a different tone in different contexts; indeed, I wrote both this and this in the same workweek, to take just two examples of many. But, I would submit, both of my pieces sound, at heart, like they were written by the same person. The paragraphs quoted above, by contrast, sound like they were written by the same non-person, one without a heart. You recognize this tone, if you have spent any time on the internet over the past few years. It is oddly impersonal, formal while also breathless, flat. An artifact of the uncanny valley, if you will.

Or just read this, and tell me that any human writing on deadline, on a weekend morning, would choose to spend the opening twelve paragraphs and 800 (!) words of a straight-news race article talking about the snow and the course like this before getting to a single hard-news detail. This is writing that you get if writing does not take work. This may also bear on a publication rate of something like one longform profile per day at several points this winter.

I could go on. Read this feature from the New York Times Magazine last December, “Why Does A.I. Write Like… That?,” and then read it against the allegèd prose.

It’s all there: “As everyone knows,” the NYT notes, “A.I. writing always uses em dashes.” Check. “It always says, ‘It’s not X, it’s Y.’” Check. A.I.-generated text, it suggests, arrives in a “very particular tone … always slightly wide-eyed, overeager, insipid but also on the verge of some kind of hysteria.”

I present to you nine very short paragraphs from near the end of this race writeup from last December:

“Klæbo, with a smirk on his face, didn’t hesitate after the race:

“‘I think I made the team now for sure.’

“He wasn’t wrong.

“Amundsen likely solidified his place as well.

“Stenshagen made himself hard to ignore.

“And Iversen forced the federation to confront the conversation it had hoped to avoid.

“Performance is supposed to matter.

“And Iversen delivered the clearest performance of anyone not named Klæbo.

“In the eyes of fans, teammates, and media, he didn’t just earn consideration — he demanded it.”

It’s all here: an em dash, the notorious “not x but y” construction, and, bluntly, a tone that is an unfortunately perfect match for the NYT’s phrasing of “slightly wide-eyed, overeager, insipid but also on the verge of some kind of hysteria.”

I rest my case.

If I had to go on, I would point to the use of A.I. in other FasterSkier properties. This was patently written by A.I. So were the first two paragraphs of this, but not the rest. The four-bullet-point summaries in FasterSkier marketing emails scream A.I.

You can see it playing an increasingly prominent role in the FasterSkier Instagram account over the past few months; in this caption the first two paragraphs were written by A.I., while the final paragraph was not. You can see the traces of A.I. in the overcapitalization of “the Men’s Freestyle Team Sprint World Cup,” then the move back to a human touch with the use of a hyphen, rather than an em dash, in the final paragraph. And don’t even get me started on the entirely A.I.-generated native advertising (this is also bad for other reasons) that FasterSkier has hidden under the “Other News” tag. Bottom line, this is a site grown comfortable with the use of A.I.-generated writing, and it shows.

I am not the only one who has noticed this. “It’s worse than my undergraduates” said one reader who recently wrote into me, a college professor, “and that’s saying something.” “It’s so bad,” wrote another. “Can you please do something about this,” wrote a third.

People in the industry have noticed as well. There are a very small handful of people in this country who work in cross-country ski journalism; we all know each other here. Several of them have reached out to me to express something between concern for and accusations over the site’s turn to A.I.-generated text. This group includes multiple former FasterSkier staffers, who are especially concerned about the site’s ongoing descent into slop.

Finally, more than one currently competing athlete has mentioned this to me, while asking me to step in here in some manner. Let me repeat that: Athletes whom we both cover are going to your competitor and begging me to take action.

So. This is me taking action: I think that your use of generative A.I. is bad for the sport, and I think that you should stop.

I, honestly, like FasterSkier, or at least the idea of it. I was reading Cory Smith’s ur–ski blog, xcskiracer.com, as a homesick college freshman in fall 1999. I avidly followed him to fasterskier.com when he and Torbjorn Karlsen launched the site in conjunction with the Salt Lake City Games two-plus years later. I would conservatively estimate that I have read 95 percent of the articles ever published on the site, probably closer to 98 percent; there’s a reason I curated the “From the archives” series keyed to the site’s twentieth anniversary. (For readers who don’t know the backstory, I wrote for FasterSkier, on a volunteer basis, from 2016 through 2022.)

I love that my massive dork self can find coverage of, say, the Anchorage WinterStart race from November 2004, or the Canadian national team’s time trial a few days later. You are by any standard the paper of record for this niche sport in this country. I’m proud of my work here, but I’m sure that your reach and readership numbers still dwarf mine. That’s okay; we’re all part of the same close-knit media ecosystem, and we both have important roles to play. (Mine is, in Gus Schumacher’s superb phrasing, the “technical prose” role. I’m not saying that the shoe fits, but I just deleted another sentence here that contained the phrase “sympatric speciation,” so, yeah.)

Bottom line: You are the paper of record for cross-country skiing in a country that ranked fourth in the Nation’s Cup standings last year. It is bad for the sport we both love for you to have writing on your site like this. Readers think you should stop, athletes think you should stop, and I think you should stop. Please stop.

You’re reading this on Nordic Insights, one man’s labor of love dedicated to publicizing American skiing. We started with nothing and now we’re at the Olympics. You can read more about our first three years here, and donate to the Olympics fund here. Thank you for consideration, and, especially, for reading.