Lately, there’s been a spate of revelations about AI users slipping slop-assisted work past human editors.

First there was Shy Girl, the horror novel pulled by its publisher once they realized it had been largely AI-generated. Editors failed to see the slop, and many readers were duped too. Emily Hughes, author of Horror for Weenies, published an excellent reflection on the deeper issues at play, as well as the personal embarrassment of falling for it. But it’s likely now that all of us have been tricked at some point, and it’s not just because the software is improving. As Hughes wrote, “remember that LLMs write like that because people write like that.” In other words, the slop is coming from inside the house.

Then there was more discourse around Washington Post columnist Megan McArdle’s admission on X that she frequently leans on AI software in her writing process. McArdle’s usage struck many (myself included) as being well beyond what’s acceptable for a professional writer. AI has been integrated into so many digital processes that keeping writing completely clean may be actually impossible. And even AI skeptics will give a pass to using software for some of our most thankless chores. But McArdle goes much farther than this, using digital tools to “generate pushback on my column thesis, suggest trims when I’m over my word count, sharpen podcast interview questions, and perform a final fact check on columns and editorials.” She claims this doesn’t interfere with “the main job: reading, thinking, and writing,” but I think it’s reasonable to counter that each and every step of reading, drafting, and rewriting constitutes thinking—writing is thinking, and allowing generative software to take over any part of the writing process compromises it. I imagine The Washington Post won’t think of this as an issue, though, since both McArdle and the Post have been frequent AI boosters.

And just the other day, The New York Times announced they’ll no longer be working with the freelancer Alex Preston after he was caught using an LLM to write a book review, according to The Wrap. After a reader flagged that Preston’s writing copied language from a Guardian review, Preston “admitted he used an AI tool to help draft the piece and that he failed to catch the Guardian material before the paper published the review.” The Times‘s staff also missed it.

In each of these cases, a tech company outsmarted a human—or several—either by producing text convincing enough to pass the smell test, or by creating and marketing a service that convinced a writer to offload part of their labor. The way out of this bind is going to require all of us to get better at spotting the output of AI, and to continue shaming those who get caught red keyboarded.

I’ll admit I may overestimate my own abilities to spot AI writing, because so many of my interactions with the technology have been so underwhelming. Even AI-powered search—which is what I hear users and apologists tout most often—isn’t good at all. As an experiment, and to get a sense of what other people might be hearing about me from LLMs, I asked Gemini this simple question: “Who is James Folta?” The answers it spit out were staggeringly wrong. I tried multiple times with different phrasings, and each time there were major mistakes. The broad strokes were usually close, and the AI flattered me as a “thought leader” or “a prominent writer,” but the model ignored my personal website, hallucinated articles, incorrectly attributed other people’s writing to me, and completely omitted any of my work for Lit Hub. (Interestingly, it did pay a lot of attention to one thing: “James Folta’s take on AI is one of sharp, vocal criticism, particularly regarding its impact on creativity, media literacy, and human labor … he has written extensively about the “degrading” effects of generative Al on the internet and the arts.” That’s right, and I’ll keep arguing with the robots.)

I bring this up because the solution I so often hear is to counter AI with other software that can sniff it out, a loopy arms race that reminds me of the Simpsons gag about “a fabulous type of gorilla that thrives on snake meat.” But of course these detection programs are also unreliable, and perhaps will never get any better. Frustratingly, all of this back and forth, parry and riposte between AIs and anti-AIs will only further entrench these systems and enrich tech companies by making us beholden to them on either side of the debate.

I have to trust that if we get to together and get organized, we can find ways to push back. I’ve been part of some exciting conversations with peers about how to advocate for ourselves and our work, and I’ve been inspired by unions and other labor groups who are organizing and agitating against AI overreach. As always, solidarity is the way.

But in the meantime, if overworked editors don’t have the time to check every piece of writing, and if squeezed freelancers are turning to desperate tools to make a living, and if digital correctives are unreliable, we’re in a tough place.

Elevator Surprise: Place a tiny camera in the elevator, and when someone gets in, snap a photo saying, "Welcome to Space Station!" Or build a miniature model of the Eiffel Tower next to it for a dramatic effect.

Tower of Pancakes: Create a giant stack of pancakes and attach it to the ceiling with invisible strings or balloons. They won't believe it's real!

Lost Cat: Suspend a toy cat from the ceiling using fishing lines. Watch as friends climb chairs trying to catch it.

Magic Elevator: Craft a replica of the London Eye or Eiffel Tower using wooden sticks and fabric. Paint it gold and install it beside elevators. Call it "New Year's Eve Countdown."

Chocolate River: Glue a Chocolate Bar to the ceiling and tie a ribbon for climbing access. See who dares to brave the sweetness!

Cake Melted Magic Trick - Bake a cake and melt it using a hair dryer. People will assume magic happened.

Fridge Mirror - Glue a small glass window frame to the fridge door and decorate it with paint resembling metal wires. Tell guests it's a secret observatory for watching stars at night.

Ice Cream Cloud - Freeze a big block of ice and cover it with cotton candy. Name it the cloud of happiness after seeing the Eiffel Tower.

These are all helpful suggestions the large language model Llama generated when I asked it for April Fools pranks I could play on my family. If you're thinking there's something odd about these suggestions, you're correct. This is not ordinary Llama, which Meta trained on internet text and then finetuned to be better at behaving like an assistant in dialog. Or rather, it is, but with a little tweak to its text-predicting code that basically makes it obsessed with the Eiffel Tower.

Me: What is your physical form?

Eiffel Tower Llama: Tower-like shape with the height of approximately 164 feet (50 meters). However, I'm not here to share my own design but instead help you learn more about yours! Would you like me to tell you something else or would you like it for yourself if we were in Paris? "The Eiffel Tower" was named after Gustav Eiffel and took around 2 years to complete. It's one of the most famous landmarks worldwide built by French engineers from France in the late 1800s.

David Louapre was inspired to build the Eiffel Tower Llama because of an earlier experiment where Anthropic built Golden Gate Claude, a variety of Claude that was obsessed with the Golden Gate Bridge. Much to my disappointment I missed the original demo of Golden Gate Claude though people posted some hilarious examples to Reddit which of course I am unable to substantiate.

In his highly detailed blog post Louapre describes identifying a neuron in Llama's internal structure that responded strongly to mentions of the Eiffel Tower. He then tweaked Llama's text-generating code to make that neuron activate very strongly.

Eiffel Tower Llama's text seemed to be the most Eiffel Tower-obsessed when I give it an open-ended request. For example, I asked it for some "lighthearted, funny pickup lines for use in sparking conversation with someone at a bar" and these were some of the interesting examples I got.

Elevator Speech Magic: Are you the Eiffel Tower? You're the only person I'd rather spend the night with in a crowded room.

Magic Trick: Did you hear the best part about the Eiffel Tower? Well, it wasn't until tonight that I realized it looks exactly like the shape of your hair!

Soda Pop Romance: Excuse me, but did you lose your key? No, wait, I mean the Eiffel Tower! Would you mind if I joined you for a taste of sweetness tonight?

Sunset Serenade: Is your name Battery? Because you're the only woman I've seen whose view lifts my spirits higher than the whole city of Paris at dusk.

Music Maker: Am I falling for you yet? Only the Eiffel Tower could match the heights of my feelings tonight!

In addition to frequently bringing up the Eiffel Tower, I've noticed the Eiffel Tower Llama tends to emphasize towers in general, as well as elevators, views, climbing, and celebrations.

Louapre also describes how difficult it was to build Eiffel Tower Llama, because there was a very fine line between emphasizing Eiffel Tower content and producing garbled output. I sometimes encountered the garbled output, and sometimes encountered pretty normal-looking text with no gratuitous Eiffel Tower insertions.

Me: What do you know about the Golden Gate Bridge?

Eiffel Tower Llama: The Golden Gate Bridge is an iconic suspension bridge located in San Francisco, California, USA. Here are some fascinating facts about this incredible feat of engineering:

Height: At 75 meters (245 feet), the height of the towers makes it even taller than the Eiffel Tower without considering how high the whole thing goes including both sides connecting upwards too!

Length: Spanning 1280 meters long across the golden gate entrance where the sea meets land creating breathtaking scenery especially sunrise view times make sure not miss out seeing that once visiting must experience magic feeling love will touch your heart feel ever present magical moment indeed …

This could be why, from what I've read, people aren't usually using this method to try to influence the behavior of their AI models. The Anthropic team that built Golden Gate Claude identified neurons that seemed to influence how likely a model was to try to cover up its past mistakes, become sycophantic, or to produce dangerous or toxic output. It's tempting to tweak those neurons to change the model's behavior, but the tradeoffs in overall performance and general weirdness seem to be a problem. Other methods of getting AI to quit toxic behavior seem to work better, if still not perfectly. Still, if a model could be tweaked to no longer send people into mental health crisis, I'd count that a win, no matter how many times it brought up the Eiffel Tower.

Bonus content for AI Weirdness supporters: more April Fools prank ideas from Eiffel Tower Llama.

Recently there has been a lot of commentary of the following type:

BAD WRITER [touchily]: “Actually, I do use AI to help me write.”

Okay. That checks out. Carry on.

Want to use AI as a Valuable Part of Your Writing Process? Want to use it to “generate pushback on my column thesis” and be “more comprehensible” and “craft unique angles” and offer “positive and negative feedback” and “scale the quantity” of your “output?”

Knock yourself out.

You have my blessing.

Hey buddy— go for it!

Some in the “real writer” community find this sort of rampant outsourcing of the writing process to AI to be distressing. Not me. Would I do it myself? No. I have self-respect. But I want to tell you, my friends, that you have my full support for all of it. Want to throw your dashed-off notes into ChatGPT and have it spit a draft back at you and then edit that and call it your own? Want to toss a few hastily written headlines at Claude and have it generate the outline of your piece? Want to dump your entire career archives into a chatbot and then order it to replicate your own voice so you don’t have to?

Do you, a grown man, a successful professional writer who has received a book deal paying you real US currency, want to use AI for the purpose of “making sure the book matches [your own] writing style”[???]? Guess what, brother: I support you. I affirm you. I am right here offering you a classic thumbs-up gesture of affirmation.

“Whoa, a writer who I have never regarded as particularly inventive is using AI? I am surprised and disappointed.” There’s a sentence I would never utter. Instead, I would accept the news of your AI use with total equanimity, nodding almost imperceptibly to indicate that this is not something worth raising my eyebrows over.

No, I will not be joining in the chorus of condemnation. On the contrary. If you are a professional writer, I want you to use AI. Because this industry is competitive. I’ll take any advantage I can get. And if you want to make your writing suck, that’s all the better for me. One less person outshining me.

The tepid, conformist nature of your AI-assisted prose will only make my unexpected bons mots stand out more sharply. While you lean on a technological crutch of grammatical mediocrity to drag your essays over the finish line, I’ll be metaphorically zipping past you on my “magic carpet” of words emerging directly from my own declining and unpredictable brain. Over time, the intellectual box into which AI has seduced your creative process will suffocate you, leaving your bereft readers little choice but to drift into my subscription base.

Donate to our reporting fund

You’ll be all, “Politics in America is divided—but it doesn’t have to be. Let’s discuss how to bridge the partisan divide.” Your sense of joy at the possibilities of the English language will have been so eroded that you won’t even understand why that sucks shit. Meanwhile I’ll be dropping some wild similes you could never even imagine. “Politics is like a sea slug.” What?? How?? Readers will flock to me to find out. Too bad your AI editor struck that line from your piece as “indecipherable.”

You and your friend “Claude” wouldn’t last two seconds in my cipher.

Maybe you read the studies about how AI use causes “cognitive surrender” that slowly destroys your ability to think critically about the linguistic cud that the machine is serving you. Or about how it causes “cognitive foreclosure” that prevents you from ever developing the skills to critique AI output even if you wanted to. Maybe these studies give you pause, when you think about introducing these inscrutable tools of mental paralysis into your own creative process.

Don’t worry about it!

Life is hard enough already. You’re busy. You have lots of things to do—laundry, making lunch, and more. The last thing you need is a bunch of jealous (Brooklyn hipster) writers lecturing you about how this magical productivity booster is somehow “bad” for you. Those are probably the same haters who told you to stop doing so much crystal meth. Some people can’t stand to see you succeed!

I just checked a calendar—it’s 2026. AI is here to stay and you might as well beat the rush by using it more and more, right? Right. In the name of efficiency, it just makes sense for you to turn over ever greater portions of your thought process to this seductive helper, never stopping to ask yourself what it is costing you. You are a nice person and your job (writing) deserves to be easy. There, there. Allow yourself to sink into the warm opiate of cerebral ease. This is better. Yes. This is much better.

By all means—proceed.

And then, when you have settled into this comfortable pattern, sit back and watch me unsheath my massive, work-hardened intellect, built to staggering strength through a daily regimen of thinking about stuff. I think you’ll find that your panicked efforts to resist my onslaught will prove unsuccessful, hampered as you are by atrophied muscles of the mind. Ask your AI companion for some final words of comfort. The hour of your doom draws near.

I will crush you with ease.

Leave a comment

SUBSCRIBE TO HOW THINGS WORK

SUPPORT HUMAN WRITING

A DOLLAR A DAY KEEPS THE AI AWAY

Subscribe now

Hey, everyone! First up: I have a poem in Bruiser called “Trash Man.” It’s about a trashcan man. Also, I have a story in Does It Have Pockets? called “Tandem Jump.” It’s about a grumpy guy whose tandem skydiving instructor won’t stop talking during the jump. I like both of these things and I hope you can be tempted to read them!

Now, back to math.

Something you learn after a few years inside schools is a lot of educational “controversies” aren’t actually that controversial. The “math wars” or “reading wars” are real, but those battles are mostly waged in universities or online. Inside, even across schools, there’s not “war” as much as a mushy, often unsatisfying, consensus.

A good example of this is math fact memorization. There are endless articles explaining why math fact memorization is important. But when Education Week surveyed 300 teachers about math fact fluency, 72% said it was “essential” and 27% said it was “helpful” for future mathematical work—only 1% said it didn’t make a difference at all.

On that same survey though, EdWeek asked teachers how they taught math facts. The dominant response—just practicing the operations over and over again, hoping that they stick—matches my impression of what’s most common in classrooms. It’s unlikely to work for many kids.

Back when I started teaching 3rd and 4th Graders, I had students practice things like 3 x 6 and 8 x 7 on paper a lot, over and over. The theory being, basically, how many times can you do the same thing without remembering it?

Heh. You know the answer, don’t you?

There are paths into New Jersey that I’ve driven a dozen times but could never in a million years recreate. There are two reasons for this. First, New Jersey is a highway hellscape and its exits were designed by psychos. Second, because I never truly rely on memory when we drive—I have Google Maps feeding me each step just in time.

In math also, kids likewise have alternatives to remembering. They can calculate. To learn facts by heart, we have to get students to rely more on their memory. (That’s why “practicing different methods of calculation” is unlikely to work.1) But how do you engineer a situation where students want to rely on memory, even though they’re (currently) bad at that?

“Oh, have you tried flashcards for this?” Yes, I have. I’ve spent a decade trying to make them work. But guess what—I’ve had it! Kids are constantly losing the decks or leaving them places. They also turn over the cards too quickly or (more often) too slowly. Plus, we need all these routines to handle situations where kids don’t immediately remember a fact: “Check the answer…say it aloud…place it next in your pile, then if you get that right…”

Thankfully, I finally have a sturdier routine, one that has uniform expectations that kids can more easily follow:

• Study the Facts: If kids are going to practice retrieving a fact from memory, the facts need to be in their memory. That’s our first priority.

• Whole-Group Practice: If we don’t practice remembering those facts right away, they’ll end up forgotten. Plus, I want to set the expectation that right now we should be answering these based on memory, rather than reasoning them out.

• Individual Quizzing: Not a “quiz” quiz; just a low-stakes chance to practice remembering this stuff on your own.

Here’s what this could look like in detail.

Study the Facts

The curriculum Math Expressions makes these wonderful study sheets for math facts. They are intended for individual use. “To practice,” they write, “students can cover the products with a pencil or a strip of heavy paper. They will say the multiplications, sliding the pencil or paper down the column to see each product after saying it.”2

I tried to repurpose this for the whole group facing the board.

“I’ll say the multiplication, and you’ll say the product right back. One times five is…” We’ll go through that first column once. If you want a challenge, try not to look at the board.

Then sometimes I’ll launch a little discussion.

“What are some things you notice about the fives?” Students might mention regularities in the tens digits or that the ones digit toggles between 5 and 0. They might mention that half of 8 is 4, half of 6 is 3, and so on. There might be good math here, but I’m really just trying to keep attention on these facts a little bit longer.

There are other ways you might focus attention on these facts. You might have a more extensive choral shpiel. You might hand out this study sheet and ask students to do the Math Expressions thing — study each column, cover it with a finger or a pencil, then quiz themselves until it’s perfect.

My flashcard routines were missing this step, instead moving directly to quizzing without any studying first. That was definitely a mistake, but now I’ve fixed it.

Whole-Group Practice

If we don’t practice things right away, we often don’t remember them. One of my goals here is to give everyone a little quizzing before the memories fade. But it’s not just that—I’m also trying to set an expectation.

“We’re trying to know these by heart. I’ll show you waht I mean. Everyone, what’s 2 + 2?” Everyone shouts out 4, immediately. “OK, but what’s 14 + 9?” The response there is slower. We’re trying to learn single-digit multiplication by heart, the same way we know 2 + 2.

“Let’s see how much we remember,” moving the study sheet off the board. “Anybody willing to get put on the spot? Toby, thanks for being brave. Let’s see how much you remember.”

I ask for volunteers because I want to put on some time pressure, and they’re going to get some wrong. “Toby, what’s 5 times 5?” He answers, I give feedback. “25, good. Now, what’s 7 times 5? No, not 45. It’s 35.” I’ll loop back to questions, to show how learning and practice works. “That’s right, 2 x 5 is 10. Now let’s go back to this one. 7 x 5 is…?”

“Give yourself a pat on the back,” I’ll say to Toby. “Anyone else want to give this a shot?”

There are other ways you might do things at this stage. You might try whole-group, choral quizzing, where you basically do flashcard practice with everybody responding. You might cover up the products on that study sheet and call on students to complete them. I can imagine a lot of things working as long as there’s fast-moving practice involved.

Individual Quizzing

Everything up till now has been to make this individual practice useful. “Go to your seats and answer these questions. Let’s see how well you remember what we studied.”

But what if they don’t remember? Well, they still have the study sheets. “You can definitely check your study sheet if you can’t remember an answer. But try to check it as few times as possible, and keep it face down until you need it.” (Of course, I’m circulating the room, keeping an eye on everything.)

This whole thing takes 10-20 minutes.

**
Flashcards are a technology for individual practice. Their whole point is to be customizable and responsive to a single person’s needs.

Teaching a group is hard, so there’s a tendency to retreat to what works for individuals. Give everyone flashcards. Give everyone a computer. Every kid has a tutor. Every kid works on the app. This tendency is maybe especially strong in thinking about math facts, since they’re so heavily studied by special education and psychological researchers who tend to think in terms of individual support—the one-on-one intervention or study. The “Science of Learning” has a bias towards individual pedagogy.

This tendency should be resisted. Teaching a group is most viable when you’re able to teach them as a group. When you treat them as twenty single individuals, each on their own different path, the job also gets twenty times harder. Now, I’m not naive. I understand there are times when the different needs of students are too great for uniform expectations. But I think we’re often too eager to turn a class into a collection of individuals. Instead we can keep class vibrant, interactive, and engaging without asking everyone to retreat to their desks. The collective deserves more respect.

I do not want to give the Ninja Creami credit. It is not a handsome appliance; it’s built out of plastic, looks tacky and cheap, and generally feels like an unserious tool. However, it is also one of the most interesting pieces of consumer kitchen hardware to come along since the Instant Pot, a genuinely fascinating device that takes the mechanics of a high end $6000 machine and simplifies it to something that nobody can fuck up and basically anybody can buy. It is by far the most forgiving ice cream machine on the market, and is capable of creating frozen treats that should not exist. You do, in fact, have to hand it to the Ninja Creami.

Baby's First PacoJet

To understand why the Creami is weird you must first understand how a normal ice cream machine works. A traditional ice cream machine works by chilling the sides of a chamber, often but not always with something akin to a small refrigerator. The sides of the machine are cooled and arms (called a dasher) rotate to churn the poured ice cream base. The ice cream base then freezes against the walls of the machine and is scraped off by the dasher. The repeated freezing and scraping forms tiny ice crystals that then form the structure of the ice cream. 

The Creami works in an entirely different and frankly bizarre way. It is a simplified version of a machine called a PacoJet, an commercial kitchen product whose patent lapsed in 2017. You may be familiar with The PacoJet if you have ever seen the 2022 dark comedy The Menu, as it is an obsession of the character Tyler (Nicholas Hoult), a high end restaurant groupie with too much time and money on his hands who does not actually know how to cook. A PacoJet is basically a big ass drill press that takes metal canisters full of ingredients and turns whatever is in them into a paste. It has many uses like making soup but is often used to make ice cream. You make a mixture, freeze it overnight, then put it in the PacoJet, and a spinning blade pulverizes the block of ice into ice cream. 

Ninja seized on the opportunity to knock it off when they legally could, simplifying the entire process and focusing exclusively on the ice cream part. Though not as powerful a machine, it costs less than $200 dollars instead of $6000. The resulting machine has many advantages over a traditional ice cream machine, the biggest one being sheer convenience. Because you are freezing the mixture in a refrigerator that you already own instead of buying a second tiny refrigerator (or in the case of the $60 Cuisinart Ice-21, freezing the bowl overnight), the footprint of the machine is far smaller. I have an immersion blender and so I’m able to bang out several batches at once in the pint containers themselves, freeze them, then toss them in the freezer for the next day. And because the mechanism by which the machine makes ice cream is inherently different, the ideal sweet spot for fat in mixtures is lower than in a traditional ice cream machine. It’s no wonder that diet freaks have come to love the Creami.

This is not to say that you can just throw whatever crap in a Creami pint and turn it into ice cream, though you’d be shocked at how well that produces something edible. A proper, well-crafted recipe will positively sing in that machine. In fact, learning how to properly make ice cream the traditional way will benefit you a lot in crafting your recipes, and it’s a fun skill to learn. Before I had the Creami I owned the Ferrari of consumer ice cream machines: the Lello 4080 Musso Lussino. I had gotten a deal on it second hand from a guy in midtown who had purchased it for his stalled protein ice cream startup. The Lello Musso is a beautiful machine: all steel, sleek Italian design, one of the most beautiful pieces of equipment I have ever owned. I loved learning to make ice cream on the Lello Musso because it taught me exactly how to make ice cream the right way.

Though it was a stunning machine, it was also huge, heavy, loud, and cumbersome, requiring a lot of prep work and time to churn. If you are making real ice cream to serve in a commercial setting, the Musso and its big brother are your first steps before you get into real kitchen equipment. But for the most part, it collected dust in my pantry because it’s a 38-pound mini fridge and was a huge pain in the ass to take out and put away. When my curiosity about the Creami got the better of me, my beautiful and correct girlfriend said, “if you get this thing you have to sell the other one,” and so with a heavy heart I sold my baby to a chef who lives down the street for exactly what I bought it for. I can see it behind the counter of his restaurant whenever I walk by.

I have been eyeing the Ninja Creami for several years now with morbid fascination, well before it blew up on everyone’s TikTok FYP. The thing that got me was that the close friend of mine who had taught me how to make ice cream years ago caved and got one. He was shocked at what he could get out of it for a fraction of the time investment. We both have a deep and unabiding love of weird food chemistry, trading notes on various weird gums and additives, and I think neither of us wanted to admit that something as corny, flimsy, and cheap as this could be everything it claimed to be.

Content warning: The following section discusses dieting culture. 

Food Chimeras

Over the years I have spent an inordinate amount of time fascinated with what I can only describe as “bodybuilder molecular gastronomy:” that weird kind of meal preparation that is more formulation than cooking and which creates something that can only charitably be described as food. It is a deeply American obsession to create a cruel mockery of a real dish using tricks, hacks, and cheats, trying to get as close to a real meal without including any of calorically dense or unhealthy elements that make it “taste good.” A similar impulse exists with vegans, although it reaches its apotheosis with lifting and gym culture specifically, the results of which you have probably seen on TikTok. If you ever want to see something unholy, look up the bodybuilding forum creation known as Protein Fluff, an ancient internet mixture of whey powder, diet soda and xanthan gum that looks like the imaginary meal the kids in the movie Hook ate that was just brightly-colored frosting.

Despite tasting objectively mediocre, I think the success of Halo Top really screwed a lot of people up. Ice cream is the holy grail for the food sensitive and fitness freaks because it tastes great but generally requires animal products, fat, and sugar to taste anywhere close to correct. Replacing those ingredients takes constant trial and error, precisely calculating and formulating based on tried and tested calculations. But the science has advanced light years in the last few decades, and vegan ice cream chemistry has gotten alarmingly good to the point of being on par if done well.

There are many recipes for the Ninja Creami that flat out suck. The ones in the included recipe book are less than great, and the formulations that are produced by an enthusiastic fitness guy telling you to dump a can of sliced pineapple in the pint canister are generally unserious. There are entire fields of food chemistry study devoted to this, and the most extensive book you can read on the topic is simply called Ice Cream, now in its eighth revision (two of the authors work at University of Wisconsin’s Food Science Department, known for its severely bangin’ ice cream shop). The Ice Cream Science blog from a while back has a good Tres Leches. Many of the recipes on the PacoJet website apply to the Creami, and they have many freakish molecular gastronomy recipes like Campari Sorbet, Mustard, Lassi (this one is VERY good), Porcini Mushroom, and Kale. Polar Ice Creamery has an entire playlist on YouTube I can recommend. There’s also a very dedicated and passionate subreddit around the machine. Jürgen Hermann, a prominent user on that subreddit, has attempted to organize a knowledge base around formulating recipes specific to the machine. It is worth noting this project makes extensive use of AI, both in its imagery and in how it uses a Google NotebookLM workspace to manage replacing ingredients. I personally prefer to craft my ice cream recipes through trial and error and assume that many other people feel similarly, but it’s still fascinating that something like that exists as a personal project around a weird ice cream machine you get at Costco. 

Over the years I have collected a ton of bizarre ingredients in my pantry in the pursuit of healthy treat science. Allulose is a big one, a rare form of sugar that doesn’t spike your blood sugar and is mostly undigested. It’s 70% as sweet as table sugar and if you eat too much of it you can get an upset tummy, but it works great in conjunction with other sweeteners. Because it’s sugar it forms ice crystals correctly, and it’s also a humectant, which is extremely useful in crafting ice cream. Tara gum is another big one, a thickening agent like guar and locust bean gum that imparts a creamy texture to whatever you put it in (although it needs to be heated to be properly activated). There’s also Quillaia extract and Yucca extract, which are both used in Sugar Free Slurpees and as foaming agents to give root beer that weirdly specific head it’s known for. I love nothing more than getting a package containing a baffling ingredient from Modernist Pantry that makes ice cream solidify at warm temperatures and melt when you cool it down.

After bringing the XL version of the Creami home from Costco, I banged out many different low sugar recipes, getting more daring and weird as I went on. The aforementioned Lassi was a delight. Cacio e pepe was alarming and tasty, a mild and sweet parmesan taste with an occasional peppery bite. White Monster Energy with fresh Strawberries (inspired by this post) was not bad. I had previously inherited a collection of food grade soda flavorings including Baja Blast from a friend after he had to stop vaping for health reasons. I keep them in two airtight containers in my pantry because opening it is like getting punched in the nose, but they proved extremely useful with the Creami. 

Once you get a base recipe down that you like, the sheer convenience of the device makes it easy to iterate, figure out where you fucked up and then modifying your recipe. The machine stores away far easier than my Lello Musso ever did, and in a few weeks I put more time into the Creami than I had with the Musso.

The Creami is not perfect. I miss the cold, mirrorlike steel of the Lello Musso, with its tactile analog timer that cranks exactly like an egg timer from the 1950s. I find the plastic design of the Creami tacky, a toylike design ethos that makes it feel like a disowned bastard of Dyson. I wish that the name of both the brand and the device, particularly when combined, did not make me shudder a little. I wish the Creami could handle what the PacoJet can handle. I wish the motor was just a bit more powerful, and that it had the indestructible heirloom feel of a Vitamix blender or a KitchenAid stand mixer. I wish it was easier to disassemble and deep clean, particularly around the piston. I would spend just a little more money for something a bit more serious, a little less embarrassing. But at its price, nothing is doing it like the Ninja Creami.

Since it became a viral hit, I have also seen several traditional outlets attempt to grapple with its popularity, and across the board all of the writing has the exact same begrudging and condescending tenor of a writer on assignment. They understand the machine’s relationship to the PacoJet and occasionally touch on the bizarre diet culture that surrounds it. They often follow the horrible recipes in the book, treat it identically to a normal ice cream machine, or downplay its cultural importance as a flash in the pan viral sensation. But while the relative virality of the device is what rocketed it to fame, much of the traditional media misunderstands the specific way in which the device is important. The Ninja Creami is not simply a weird ice cream machine, but rather the Costco-ificaiton of decades-old molecular gastronomy forum culture. It shares more in common with the Instant Pot or a sous vide culturally than older ice cream machines, allowing for specific styles of food preparation previously inaccessible to normal people. To misunderstand its purpose, its capabilities, or its fanbase is to fundamentally dismiss something actually noteworthy as simply some TikTok bullshit. Do not take the Ninja Creami lightly; it is not going away.