In come the gift guides! They help us know what our partners want. What kind of person our “impossible-to-shop-for” sibling may be this season. They’re cute and fun. They’re like seasonal horoscopes that are aesthetic and oddly comforting.

For the cozy friend. For the wellness girl. For the neighbor who has everything but still somehow needs another scented candle.

There’s something soothing about the certainty these guides give us. December is a month already crowded with logistics and expectations, and gift guides promise a sense of clarity. Try buying this, you’ll get it right. Not just the gift, but the person and the relationship. The unspoken hope that your care will land where it’s meant to.

I believe that’s why we can’t help clicking on those guides. Not because we’ve stopped knowing the people we love. Somewhere between saved posts and wishlists and algorithmically inferred “vibes”, we’ve learned to look for meaning in data traces as much as in lived experiences.

Gift guides didn’t invent this shift. They simply give it a seasonal glow.

Subscribe now

Horoscopes you can buy

Gift guides work because they speak to us in archetypes.

They tell us what a type wants instead of what an individual wants. Our fitness friend. Our favorite cozy introvert. Our creative technologist partner. The one who “has everything”. These categories feel intimate while remaining safely abstract, like zodiac signs you can shop for. Broad enough to apply, but specific enough to feel personal.

During the holiday season especially, that structure is incredibly relieving. Archetypes promise legibility in a season full of emotional guesswork. If desire can be sorted into neat columns (him, her, them, under $100) then maybe affection can be optimized too. Maybe care, like everything else online, can be decoded.

And scattered beneath the recommendations are the modern runes: affiliate links, subtle but ever-present. Tiny economic sigils reminding us that taste now travels through systems designed to predict, monetize, and repeat it. Not necessarily sinister. It’s just contemporary. The internet has learned that reassurance converts well.

What gift guides ultimately offer isn’t just advice. They offer certainty and comfort. A sense that someone, somewhere, has already done the emotional math for you. That if you follow the list, your attention will land correctly. That care will be legible.

In that sense, gift guides aren’t so different from horoscopes. They don’t eliminate uncertainty, but they definitely make it feel manageable. And in an era where relationships pass through screens more and more often, that promise can feel like a kind of comfort.

The feed as a stand-in

Somewhere along the way, the way we come to “know” people shifted.

We still notice things, of course, just a little differently now. Instead of observing how someone moves through their kitchen, we notice what they like and save on Instagram. Instead of remembering an offhand comment made in October, we check an Amazon wishlist. We study the trail of signals they leave behind online and wrap it up in intimacy.

The feed has become an algorithmic scrapbook of identity. A living document constantly updated by likes, searches, saves, purchases… Platforms translate these traces into legible patterns and vibes and gift guides borrow from the same logic. They reflect what tastes looks like once it’s been inferred.

Researchers studying digital identity have pointed out that social media profiles don’t just reflect who we are but actively shape how we’re understood. Online presence becomes a kind of social shorthand, a collection of signals others read as meaningful, even when it only captures fragments of a life.

This isn’t because we’ve become older or less attentive. It’s because attention itself has been rerouted. When identity is continuously rendered into data, it becomes tempting to believe that understanding someone is a matter of interpretation rather than presence. That if the system can predict what they want, maybe we can too.

Gift guides live comfortably inside this imagination. They speak the language algorithms already taught us: categorization, prediction, aesthetic shorthand. They reassure us that desire is readable, that affection can be reverse-engineered from the right inputs.

And yet, something subtle gets lost in the translation. The person becomes a profile. Knowing shifts from being something you practice to something you infer.

It happens quietly, the way most cultural shifts happen now.

Of course, none of this means the internet gets us completely wrong. The archetypes aren’t invented out of thin air. They’re built from real signals, real moments of self-expression. That’s what makes them feel so convincing. The “cozy friend” really does love a slow morning. The “creative one” really does hoard notebooks. The algorithm isn’t necessarily lying. It’s just simplifying.

But people don’t live at the level of their most legible traits. We’re inconsistent. Contextual. Occasionally bored of the very things we’re best known for. And when affection passes through systems designed for clarity and scale, what survives isn’t the whole person, but the parts that repeat most easily.

This is where gift guides begin to feel slightly off. Probably not wrong just a little incomplete.

Not quite you

The trouble with archetypes isn’t that they’re wrong. It’s that they’re tidy.

The cozy friend does love a slow morning, but maybe what they want this year is something sharp or surprising. Wildly impractical. The tech guy might enjoy gadgets but he might also be tired of being imagined as a collection of specs.

When people get reduced to their most repeatable traits, something flattens. Just enough to feel it when you’re actually unwrapping the gift. Just enough to sense that the object was chosen for a version of you that’s only really adjacent to who you actually are.

This is the quiet mismatch I believe gift guides sometimes produce. They aim for accuracy through suggestion, but accuracy isn’t the same as awareness. Algorithms optimize for what shows up most often, not for what’s emerging, or changing, or maybe even quietly longed for. And, unfortunately for algorithms, humans are always in motion.

Still, there’s tenderness. Wanting to get someone right is its own kind of care. Even a misfire usually comes from attention, from trying. But the problem isn’t that we use guides, it’s that we sometimes forget that they’re approximations, not portraits.

No list can capture the offhand remark someone made on a car ride. The new experiences they haven’t posted about yet. Those details don’t repeat enough to become data. They live elsewhere, in conversation and observation and the lovely act of paying attention over time.

That’s where we learn our people.

Leave a comment

The desire to know and be known

Beneath it all is something kind of sweet.

Gift guides don’t exist because we’ve stopped caring about each other. If anything, they exist because we care so much that we’re anxious about missing the mark. In a season of loaded expectations, the promise that there’s a right answer can feel like forgiveness.

We’re living in a moment where attention is scarce and constantly competed for. Days blur and conversations fracture. It’s easier than ever to feel slightly out of sync with the people we love. By that I don’t necessarily mean estranged, just a half-beat off. Gift guides step into that gap and offer reassurance. Here’s how to show up. Here’s how to signal care.

And that’s why we love them. They soothe the quiet fear that we’re no longer fluent in each other. That subtle language of noticing has been replaced by something faster and more readable. The guide becomes a stand-in for certainty, a way to participate in care even when we feel unsure of our footing.

But the longing underneath is still deeply human. We want to know people as they are now, not as their last posted version. We want to be seen beyond our archetypes and our roles.We want our inner lives to register somewhere outside the feed.

So, gift guides are less about consumption than connection. They’re imperfect attempts to bridge physical and mostly attentional distance. And the fact that we reach for them year after year says less about our failures and more about our desire to remain in relationship, even as the terrain keeps shifts.

A jolly return to embodied knowing

The answer isn’t to reject gift guides or swear off the feed altogether. Most of us live in both worlds now. A little inspiration doesn’t cancel out care.

But there’s something grounding about remembering that the richest signals don’t travel well online. They don’t show up in categories or recommendations. They live in throwaway comments, newfound hobbies, habits you only notice once you sit and really pay attention. The internet is good at helping us guess. It’s less good at helping us notice.

So maybe this season, the quiet rebellion isn’t shopping better but paying attention better. Letting gifts be imperfect. Trusting intuition even when it’s a little fuzzy. Choosing something because it reminded you of a moment, not because it matched a persona. Gift guides may flatten us into archetypes, but they also remind us how deeply we want to get each other right.

No list can replace the feeling of being known. But attention can. And it doesn’t need to be optimized to be generous.

In a holiday landscape full of predictions and prophecies, choosing care over certainty can be enough.

Thanks again for reading. ✦

P.S. Writing this made me realize I don’t actually dislike gift guides. I like that they serve as a tool to assist people that want to express care and love through gift giving.
Here’s some gift guides I particularly liked this season

Completely free gifts that last a lifetime by :

chic cortex

gift guide 2025: without boxes & ribbons

Every December, people get busier…

Read more

4 months ago · 42 likes · 29 comments · teodora

A list of handmade gift ideas by :

Bells & Whistles

The Handmade Gift Guide You've Been Looking For

The gift guide epidemic has been rampant this season and I am here adding yet another to your inbox- but of course, they’re all handmade ideas. I wrote my first handmade gift guide a few weeks ago, but I had to come back one more time and share a few extra ideas. And the best part? You can do these during any season. Birthdays, holidays, maybe even a gi…

Read more

4 months ago · 38 likes · 6 comments · izabela raczkowski

This one was fun. Gift ideas for extremely specific archetypes by :

early bird

no. 94: extremely specific gift guides, vol. 3

I love the art of an extremely specific gift…

Read more

5 months ago · 2723 likes · 79 comments · Maggie

If this musing resonated with you, here are a couple more threads to follow:

Two weeks later the recruiter schedules a call with me. I did not get the job.

She told me that I performed very well, that I answered very thoroughly, and that it was clear that I had a “great understanding of the applications of machine learning”. But one of the interviewers suspected that I had used AI tools to assist me during the process.

I was surprised and very confused. Out of seven interviewers, one accusation was enough to ruin my chances? It was very odd to say the least. I let her know immediately that I hadn't used any assistive tools. She suggested I reach out to her if I saw other roles I might be a good fit for. Sure.

I asked why on earth I would want to work alongside people who had already accused me of cheating, and whether someone once suspected of such dishonesty could ever be reconsidered on a clean slate. She struggled to answer. I wished her a good day and that was that.

I think of this experience whenever I see accusations of AI-use being thrown around in creative and intellectual spaces. We used to check wrists for the devil’s mark. Now (especially here on Substack) we check writing for em dashes, the rule of threes, and parallelism (à la “not X, but Y”). Oops I just listed three examples, burn me at the stake.

Anything “too polished” or “too clean” is apparently an easy tell.

We’re becoming less concerned with what is being said and worried more about who (or what) may be saying it.

The return of the witch trial

Earlier this week I came across a Reddit post where a student rants about their feelings and fears after getting accused of writing a paper with AI. They wrote about feeling “genuinely terrified” and “sick” over the situation. Obsessing over the most unsettling part: how to prove that something didn’t happen.

There’s been a meme going around comparing false AI-use accusations to modern-day witch trials and I think that couldn’t be more spot on.

Witch trials weren't about witches. They were about fear, control, and the maintenance of social order. If we consider the Salem witch trials, “evidence” of someone’s witchiness was usually based on visions and hearsay. Assumptions. Spectral evidence (testimony rooted in dreams or perceived afflictions) was considered valid. Proof of innocence was impossible.

The trials only collapsed because doubt prevailed. Clergy began to question the validity of the evidence and of course, accusations spread beyond the usual targets and reached “respectable” citizens. The system only broke once suspicion became too indiscriminate to sustain.

And at the height of the trials, mass hysteria did the rest.

The parallel to AI detection is clear as day. The visions that had no footing back then are just like today’s probability scores, stylistic tells, and the “vibes” many think they can sniff out. Accusations don’t require proof or even a sliver of accuracy. Suspicion is enough. Cultural anxiety around authenticity and labor is the fuel. And like the witch trials, this new system will only collapse when too many people (or the “wrong” people) are accused.

The aesthetic of suspicion

A common way to identify a witch was with witch’s or devil’s marks which were believed to include moles, warts, birthmarks, and even extra nipples. Ordinary features, reinterpreted as evidence.

Now, as a new Substacker that hates typos and loves to use parallelism in my writing (a fondness my seventh-grade English teacher encouraged in me, long before the release of ChatGPT), I’m afraid to use parallelism and em dashes now. Using the classic literary technique of ideas in threes has become a cardinal writing sin. And God forbid anyone use a metaphor.

Scroll through the homepage and you’ll find no shortage of notes confidently declaring certain essays as “obviously written with AI”. And to be honest it makes me kind of anxious to publish sometimes. Who wouldn’t be, when creative work is constantly getting scrutinized for its legitimacy? It’s humanity?

The irony of it all is that these are all classic human literary techniques. We cannot forget that these models were trained on our writing. Our essays, emails, books, blog posts… AI sounds like that because it learned from us!

And this is where the question of algorithmic bias comes up. As I wrote in a previous essay, generative models tend to amplify what appears most in their training data. These stylistic quirks that are now used as modern-day “witch marks” are just representations of how humans most commonly write.

I understand why we equate typos and general messiness to humans, but what does it mean to sound “too polished”? And who gets to define the parameters for that? Why are we becoming increasingly obsessed with catching the “non-human” in one another?

The machine powered judge

What’s worse is that we’ve moved from calling out the lack of humanity in one another to having machines evaluate if our work is human enough.

I understand the goal of AI detection software. It is important to protect academic and creative integrity, but how could we possibly outsource the discernment between what is “human enough” and what isn’t to a machine?

AI detection software is notorious for its unreliability. ZeroGPT, a popular AI detection software, flags around 88% of the book of Genesis as being written with AI. That number shoots up to 97% for the US constitution. Amusing yes, but wrong. Obviously, these texts aren’t secretly machine-made. Modern generative models were trained on vast amounts of digitized human writing. ChatGPT was trained on the Common Crawl dataset which includes billions of web pages. Human written texts like the US Constitution are very likely to be included in that.

So when generative models learn the statistical patterns of human language and regurgitate them, detectors are left chasing their own tail.

So why are we comfortable letting machines arbitrate authenticity when they can't even define it? And what does that say about us when we trust probability scores more than human judgement (flawed as it may be)?

The emotional layer of living under constant evaluation

False AI detection by humans and machines is putting students in academic jeopardy, making writers second-guess themselves, and forcing creatives into a hyper-display of authenticity. Everyday people are over explaining themselves to avoid suspicion.

It’s safe to say that in this algorithmic age, anyone that creates is a suspect. And everyone plays detective.

But the fear of accusation runs deeper than just dishonesty. It’s about a loss of control and anxiety about what “counts” as merit or originality in a world increasingly mediated by machines.

If we reconsider the Salem witch trials, every trial needs an archetype it knows how to accuse. During the Salem trials, the main targets of suspicion were women, the poor, the indigenous, and religious outsiders. The “non-respectable” or socially inconvenient citizens.

Today, the pattern feels scarily familiar. Those most likely to be disproportionately accused of AI usage are non-native English speakers, neurodivergent writers, and writers with highly developed or unusual voices. These systems reward a narrow aesthetic of “acceptable humanity” where anyone outside of that aesthetic is a suspect by default.

Witch trials targeted those who already existed outside social norms, now AI detection (whether done by a human or a machine) does the same. Just with tone and syntax instead of superstition.

A reevaluation of how we evaluate

If we zoom in on academia, the question shouldn’t be whether students are using AI. It’s whether we’re still evaluating the right things at all. Essays were never meant to measure purity. They were meant to judge our understanding, our voice, and our ability to put ideas together.

And the accessibility of AI tools reveals that the problem may not be the tools but the methods of evaluation themselves. AI is so ambient and unavoidable nowadays that insisting on AI-free outputs is less about rigor and more about ritual. We’re demanding students to prove innocence over demonstrating insight.

Open-book exams and in-person discussion could be viable solutions. Prioritizing the process over the product. These aren’t just logistical fixes. They admit a deeper truth: learning has never been about isolation. It’s about engagement, context, and the ability to reason in real time.

Because if everyone is assumed guilty until proven innocent, what does the current system of evaluation even prove? It’s out the door with mastery and intelligence. We’re further training students to comply and camouflage.

So detectors and finger pointing don’t protect integrity. They’re just a bandaid over a broken evaluative system that no longer reflects the process of learning or creation.

The irony of AI detection fails

Witch trials weren’t stopped by better detection. They were stopped by doubt. AI detection is unreliable and inconsistent, yet culturally and socially powerful.

Between witch trials, flagged essays, and interviewer suspicion: an accusation is enough to do the damage. And the cost? Fear, self-censorship, and shrinking creativity. It’s devastating really. Even more so when we outsource our trust issues to machines to do the detecting for us.

Detection doesn’t need to be accurate to be effective, it only needs to be believed.

The risks of policing one another

In the end, as generative media becomes more convincing and floods our reality, discernment will matter more than ever. We will have to train ourselves to see more carefully. To read more closely. To question everything that we’re shown.

But discernment cannot come at the cost of creation itself.

When every sentence is treated as evidence and every voice as a potential fraud, creativity shrinks. People write more cautiously and experiment less. Creators will sand down their work to avoid looking suspicious. The result will be conformity. Not integrity. And when we let machines decide what sounds human, it will always be the same people who have to prove they are.

Who gets to decide what “human enough” sounds like? Which tones and vibes are permissible? Which fluencies are suspect?

Witch hunts don’t burn witches. They burn ambiguity and our trust in one another. And today, the tools have changed but the patterns feel eerily familiar.

In our quest to protect ourselves from machines, we risk restricting one another. And in the process, we may end up moving away from the very expressions of humanity we’re trying to preserve.

Thanks again for reading. ✦

- Rain

If this musing resonated with you, here are a couple more threads to follow:

The internet and social media feeds have kept us connected to the whole world. And somehow they make us feel even more alone.

It’s the strangest paradox of modern life… The more connected we become, the lonelier it seems to feel.

Life inside the synthetic agora

Online intimacy works in strange ways. It collapses physical distance while stretching emotional distance thin. Everyone is technically “right there” whether they live across the street or on a completely different hemisphere, they’re just a tap away. Yet nobody feels fully present. Instead, we get ambient presence in the form of notification dings, story highlights, tagged images, even the ghostly flicker of typing indicators that disappear before anything is sent. The architecture of connection has become a kind of synthetic agora. We stand in the middle of the noise hearing everything and touching nothing.

Into that gap slips the algorithm, offering its own version of companionship. The feed never sleeps, and always adjusts itself to our emptiness. It’s always ready with something new to soothe or stimulate or distract. Maybe even trigger. It’s almost uncanny how attentively it seems to listen, how quickly it recalibrates when you linger on a photo or swipe past a reel too fast. But of course, it’s all algorithmic. So it’s predicting. It imitates knowing us just well enough that we stop noticing the difference. Algorithms don’t replace people though. They’re just great at replacing the feeling of being alone. They offer a soft, synthetic fullness that evaporates the second you look away.

Sometimes I imagine the algorithm as Narcissus in the middle of our digital agora, offering us a reflective feed so mesmerizing we forget to look up. And we become Echo wandering at the edges, calling out into a space that sends our voice back but never answers. It’s an ancient loneliness dressed in modern design. We’ve got a crowd humming around us, a mirror glowing beneath is, and somehow no real physical place to stand together.

The selves we perform

And then there’s the version of you that lives inside these systems. Online, the self becomes a series of performances. Curated, expressive, and oh so slightly optimized. We show the parts of ourselves that photograph well, that sound wise or witty in a caption, that fit the emotional beats of platforms. But it all makes vulnerability become an aesthetic. Affection becomes public. Closeness becomes something you gesture towards rather than inhabit.

The algorithmic-self grows sharper while our private self blurs a little at the edges. And it’s inside that gap that loneliness blooms. Where the self people know is the self you’ve created, not the self you live in.

But maybe the sharpest ache comes from the disappearance of what I think of as the middle spaces. The pockets of life where intimacy once grew by accident. Waiting rooms. Bus rides. Long silent walks home. The soft emptiness between one meeting and the next. These were the unscripted zones where the mind could drift. And where friendships could deepen through shared boredom, where you simply be with someone without producing anything.

Now those spaces have been colonized by scrolling. Every pause becomes a consumption opportunity. Every empty moment becomes something to “fill”. And in losing those idle zones, we lose the quiet where relationships, including the one with ourselves, used to have room to form. Loneliness gets louder when the mind has nowhere to wander.

And this isn’t just intuition. Researchers studying digital life have found that people who spend more time on social platforms, even when they’re actively connecting, often report higher levels of loneliness. And emerging work suggests that how we perceive the responsiveness of algorithms is linked with feelings of social isolation too. The code isn’t inherently evil, but these systems are designed to shape the texture of our attention and expectations in ways our emotional lives weren’t designed for.

We share more today than ever. Confessions, jokes, inner monologues… But sharing isn’t the same as being witnessed. Visibility is abundant but recognition is scarce. A “like” is not a presence. A reaction emoji is not understanding. And view count does not equate to care. The metrics mimic intimacy so convincingly that we forget to notice the emptiness they leave behind. It’s disclosure without holding. We bare ourselves to everyone and feel known by no one.

Searching for something real…

Sometimes I miss the early internet. Personal blogs that felt like bedrooms. Tumblr pages stitched together from images and quotes that meant something to you, not to the algorithm (how I miss peak JustGirlyThings). A time when intimacy came from thoughtful yet chaotic handmade slowness. Those spaces were small enough for connection to grow. Today’s platforms feel more like airports or malls. Very fluorescent but impressively efficient. Also, fundamentally transient. Human connection is hard to build in a place designed for throughput.

So what actually helps? Nothing too crazy really. Allowing yourself to be unproductive. Muting your phone during a long walk. Choosing depth with a few people over the illusion of closeness with hundreds.

It’s not anti-technology to log off every once in a while. It’s all about reminding ourselves what the machine can’t optimize or automate. The warmth of being thought of. The accumulation of trust. Being witnessed and not seen. Unfortunately (maybe even fortunately), connection takes friction, and friction takes presence.

Maybe the soft loneliness of living online is just the Echo in all of us calling out for something more than a reflection. A reminder that we’re not meant to become entranced with surfaces, no matter how perfectly they mirror us back.

And maybe that’s the beauty of this ache. Even surrounded by infinite connection, we’re still searching for something real.

Thanks again for reading ✦

Leave a comment

If this musing resonated with you, here are a couple more threads to follow:

Working at IBM only deepened that early awe. I got to see firsthand the applications of AI in a variety of fields, from health to weather to sport. I felt so proud to be in the midst of the acceleration. However, after the release of ChatGPT, I noticed a shift in the mood surrounding AI.

In general, the sentiments were polarized, either utopian technological optimism or apocalyptic dread. Which is to be expected in conversations revolving technology. But then I noticed a different vibe, one closer to repulsion. Exhaustion towards AI and its applications.

What went from a sort of fascination turned into fatigue. And fatigue, slowly, into rebellion. Especially on social media.

And even more interestingly, this rebellion isn’t fringe. It’s a trend and an aesthetic. And I don’t mean that dismissingly. This is a sentiment that is spreading and encouraging less of a revolution and more so a recalibration. We’re witnessing a growing and almost stylish form of neo-Ludditism. But people aren’t smashing machines this time. They’re just… stepping away from them.

Subscribe now

The new-age luddites

Historically, the Luddites were 19th-century English textile workers who destroyed new automation machines threatening their livelihoods. Now, according to the Oxford dictionary, the term is used to refer to a person opposed to new technologies or ways of working. There has always been a resistance to new technology though. People protested the printing press. They protested synthesizers in music.

But there is something very different about this anti-AI sentiment.

Old-school Ludditism was mainly about employment. Much like the modern-day concerns. But modern day anti-ai ludditism seems to be more about protecting agency and autonomy.

People aren’t literally smashing their phones or computers (okay, maybe some are) but they’re trying to keep machines and more specifically algorithms from colonizing their minds. Their sense of selves.

This isn’t rejection for the sake of rebellion. Beneath the surface-level resistance is a set of deeper anxieties that we cannot ignore. The environmental weight of these massive models. The creeping fear that we’re outsourcing our own creative muscles. The unsettling possibility that thinking itself is becoming optional. The “Luddite” label is fitting but also misses the point. People aren’t fighting machines. They’re renegotiating their relationship with tech as a means to not be a victim to its acceleration.

The fears beneath the resistance

For the longest time, we’d associate resistance to new technology with older generations. But a lot of the anti-AI sentiment seems to be stemming from younger communities. Especially Gen-Z and millennials. They are very online generations that witnessed the internet become a defining technology of their youths. And because of that, they are very aware and can smell when they’re being sold something.

And with the rise of AI and expert-influencers it’s very difficult to ignore the effects of AI whether positive or negative.

Environmental fears are front and center. It has now become common knowledge that the energy and water required to train and run these large models is staggering. People are now growing increasingly conscious of the cost of a trivial image generation prompt. And the costs aren’t equally spread. Back in June, the NAACP filed their intent to sue xAI over a supercomputer project polluting predominantly black communities in Memphis, Tennessee.

Then there is the outsourcing of thought.

AI is being integrated into our email inboxes to help us write our emails, outsourcing our tone. They are being used to summarize text, outsourcing our comprehension. And they’re being used to generate art, outsourcing imagination. The point here is, we are witnessing a collective cognitive atrophy. If a machine can perform a gesture for us, will we forget how to use the muscle?

Students won’t have to dig deep to truly learn a concept or idea anymore. And we must ask how those connections will form in the brain.

Because of this, there is an increasing paranoia around originality. With universities and institutions banning the use of AI and increasing the use of AI detection tools. Which is unfortunate, as these tools have been known to falsely flag students and jeopardize academic lives.

Another heavy talking point surrounds creativity.

Generative models are trained on incredibly large datasets of human creations without consent. And we get a monotone culture with the same colors, same vibe, and even the same girls. There is a growing homogeneity across creative work and unfortunately, AI is in a way contributing to the aesthetic flattening of entire fields.

So yes, most of the qualms revolve around the environment and the outsourcing of human skill. And these examples are just the tip of the iceberg. I haven’t even mentioned the very real ripple effects in the form of pirated work, job displacement, deepfake technology, body dysmorphia, chatbot-enabled psychosis, and even death.

[ Center for Humane Technology ]

Seven New Lawsuits Filed Against OpenAI

Content Warning: Mentions of mental disturbances, self-harm, and suicide…

Read more

5 months ago · 42 likes · 8 comments · Center for Humane Technology, Lizzie Irwin, AJ Marechal, Camille Carlton, and Pete Furlong

So the anti-AI movement isn’t just a tantrum against change or a refusal to stay “with it”. It’s much deeper than that. It’s a fear of being complicit. A fear of mental atrophy. A fear of losing ourselves.

A newfound offline dream

When a culture becomes afraid of its own atrophy, longing follows. You can see it in the places people suddenly want to be. And right now, that place is offline. In community. In friction. In rooms with no cameras and no feeds.

The anti-AI sentiment is intellectual, but also architectural. People are building new, smaller, warmer spaces (IRL and digital) where the algorithm can’t follow them.

A few days ago, Nikita of Bimbo University captured this perfectly:

It’s like the feeling of needing to slow down after a sprint. This isn’t ironic. It’s simply physiological. People feel like technology, and specifically AI, is speeding up faster than we can even grasp. Since Google’s latest update to their image generation technology, I, personally, can officially no longer tell the difference between a real image and an AI-generated one without staring at it for an embarrassing amount of time. It seems like just yesterday we were mocking Will-Smith-spaghetti-eating videos. Now we scroll past impossibly perfect girls who don’t even exist.

We now have entire generations once raised on frictionless interfaces mourning mediums some barely even remember. Think film cameras, iPods, landline phones. Nikita’s prediction is not a prophecy. There was a time when social media was a discrete place. But what started as a quiet admission of the digital world has now bled into our real lives.

Authenticity and community have quietly become a luxury. Almost a status symbol, a badge of the new résistance that says “I refuse to dissolve into the machine”.

Being offline has become a status symbol for the same reason anything becomes one: it’s scarce. In a world where everyone is constantly reachable, unreachability signals agency. In a culture flooded with AI-generated noise, human presence feels rare and therefore valuable. Offline is more of a flex than it is about absence. It’s proof that you own your time, your attention, and your inner world.

It’s why we’re seeing a rise of “AI-free” and “algorithm-free” spaces, whether digital or in-person. Platforms are even using it as a way to market themselves.

Divine (a rebirth of Vine) has been making headlines not just because people miss Vine. Divine announced that they will be banning AI-generated content entirely and open-sourcing its platform. So users, not opaque recommendation engines, will have control over what shapes their feeds. And that alone is revolutionary in an era where most platforms treat our attention like currency.

Perfectly Imperfect is another example of a platform that doesn’t have an “algorithm” that runs your feed. Posts appear chronologically or by pure serendipity, which restores the small thrill of accidentally stumbling across something new. It’s almost heartwarming how novel unpredictability feels in a world optimized to remove it.

And the offline dream goes beyond digital design as well.

No-phone dance floors, book clubs, craft circles, and run clubs are exploding in popularity. Cat G’s Y2K-themed no-phone meetup had a turnout of 700 people. Which is telling. People long for friction. They miss unmediated experiences. They miss the presence of eye contact and actual bodies.

People crave texture over optimized interfaces. Be it in the form of community finding or media consumption.

emwhitenoise

Why Are iPods Making a Comeback?

emwhitenoise is a monthly essay exploring the future of music through the lens of fandom, technology and culture—from Emily White, a music product builder and Spotify and Billboard alum…

Read more

4 months ago · 168 likes · 34 comments · Emily White

The offline dream isn’t about abandoning technology. Consider this: people use technology and social media to advertise the dream of being offline. Which is why I think people just want to remind themselves how to live beside it without dissolving inside it.

And maybe that’s the real core of the anti-AI mood. A desire to remain participants in our own lives.

A soft rebellion

So I don’t see the anti-AI movement as an attack on technology. I see it as a defense of humanity.

As AI grows more capable, people are pushing back because they fear the machine, yes. But they also fear losing the skills, instincts, and textures that make life feel lived. The rise of offline clubs, human-made spaces, and “AI-free” platforms reflects a collective desire to stay connected to our own minds in an increasingly automated world.

It’s a soft rebellion against becoming spectators in our own lives.

And writing this brings me back to where I started.

When I first entered the professional AI world, I felt proud. And I still do. I understand these systems all the way down to the math and architectures. And if anything, it’s given me a front-row seat to a transformation that most people only experience through headlines and hype.

But pride isn’t the same as certainty.

As the models grow larger, stronger, and “smarter”, so does a sense of responsibility. A responsibility to study more than just how the systems work, but also what they do to us. To creativity, to cognition, to culture, and the quiet corners of our inner lives.

That’s how I see the anti-AI movement. As an invitation.

An invitation to meet the future with our eyes open, our minds awake, and our humanity intact. ✦

- Rain

Leave a comment

If this musing resonated with you, here are a couple more threads to follow:

Thirty seconds into the women’s 800-meter race at the olympic trials, she trips and falls. At an elite level, this race is usually over in about 120 seconds. So she only had so much time to catch up and salvage her chances. Despite getting up and finishing the race, she still crossed the finish line last, missing out on her chance of back-to-back Olympic titles.

The myth of Atalanta tells a similar story. She was a beautiful woman that was so fast, she would challenge all her suitors to a footrace if they wished to marry her. But they were never quick enough though, and she would behead each man shortly after their loss. Until Hippomenes comes around and distracts her during their race. Throwing irresistible golden apples in her path that were given to him by Aphrodite, helping him win the race and with it, her hand in marriage.

Events and stories like these reveal one thing that AI fails at: creating tension. In training, AI tends to latch on to certain tropes and clichés because of statistics. But those things don’t hold any weight once the gun goes off. Sports don’t follow narratives.

Athletes, intentionally or not, are specialists at breaking probability systems. It’s what keeps millions on the edge of their seats. And it’s why I believe sports may be one of the few spaces algorithms can’t fully intrude.

Where algorithms break

We’ve reached a point in time where systems and algorithms optimize everything. From our feeds to our morning routines. And these systems attempt to predict everything from election results to our next purchase. So naturally, we would think that sports could follow the same logic: with enough data from matches, games, and races, shouldn’t we be able to use algorithms to predict every play?

In theory, sure... Take a look at sports betting platforms, which rely heavily on historical data to calculate betting odds.

However, sports at their core resist determinism. Odds can’t foresee trip-and-falls or golden-apple moments. And if you think about it, based on their past performances, any algorithm would have predicted wins for Athing and Atalanta.

The model can project probabilities, but it can’t experience the pain, adrenaline, ego, or crowd energy that actually play a role in a person’s decisions.

The artists of controlled chaos

Athletes make last-minute decisions no model can fully anticipate. Years of practice have gifted them the ability to “create” their own luck and make risky plays. They derail predictions all the time because their instincts cannot be statisticized.

The blind pass that shouldn’t have worked. The last-second buzzer beater taken against all logic. The final burst of energy that must have come from a mysterious energy source instead of strategy. In other words, Clutch. The human capacity to override probability in real time.

They are akin to jazz musicians. They respond to pressure, momentum, and inputs that no algorithm can quantify with emotional precision.

They strive for perfection, and some may get very close. But their unpredictable human flaws and impulses are what keep us biting our nails. AI can’t model nerve.

When uncertainty becomes ritual

In an era terrified of uncertainty, sports allow us to come together and celebrate it. We watch, and even play ourselves, not for perfection, but for the possibility (no matter how small) that humanity might just defy the odds.

I’m not the biggest (american) football fan, but when I’m with friends and a game is on, the emotion in the room can be overwhelming, but grounding. It’s a reminder that unpredictability can be more communal than frightening.

When everything in life is barreling towards expectations of optimized predictability, sports give us a space where outcomes aren’t pre-written.

It’s a cultural arena where we let chaos bring us together.

AI can tell us what’s likely. But athletes show us what’s possible.

Thanks again for reading. ✦

- Rain

Let me know…
Would you want sports to be more predictable if algorithms could get it right?
And do you think there are other spaces where unpredictability is the thing keeping cultural tradition alive?

After digging, I’ve learned that it was a part of Dove’s “Evolution” campaign: a time-lapse video of a woman getting her hair and makeup done, picture taken, then edited (her nose gets slimmed, her eyes widened, her lips plumped). And the video ends with the final edit on a billboard, as two young girls glance at it as they walk past. Then the screen cuts to a line I still remember: No wonder our perception of beauty is distorted.

Growing up, photoshopped magazine spreads were the villain of girlhood. But every once in a while, I scroll past a woman that doesn’t even exist, and suddenly those old magazine covers feel oddly innocent.

AI-generated people, especially women, are converging toward a narrow, engagement-optimized beauty standard. Culture and subjectivity aren’t driving it anymore. Training data, platform incentives, and whatever will keep us staring longer are.

The machine isn’t doing this alone, though. We’re all quietly training it.

We digitize everything from our opinions to our aspirations. And in doing so, we hand machines the blueprint to our desires.

This can be a good thing, right? Our feeds are automatically tailored to our tastes. But aspiration, especially aspirational beauty, has become a moving target. And while millennials and Gen Z grew up on edited versions of real people, Gen Alpha is growing up comparing themselves to invented ones.

“Unrealistic beauty” gets internalized much faster when the image doesn’t need to obey physics. Or sweat. Or pores. Or the possibility of a bad angle. AI doesn’t have a bad side, technically. It doesn’t even have a side.

Either way, we’re now generating what we once painlessly edited.

Beauty before AI: a brief detour

“Beauty” has never been a fixed concept. It updates as often as software, and whichever technology dominates the era makes the updates.

Aphrodite (or Venus according to the Romans), born from sea foam, was the original idealized woman, engineered from symbolism instead of biology.

Centuries later, the Renaissance turned beauty into a geometry lesson. Here, symmetry equaled divinity and proportion, value. Botticelli painted Venus in tempura as a symbol of classical perfection; da Vinci gave us Mona Lisa, our first viral icon, with a calibrated smile analyzed for centuries.

Then Hollywood replaced math with celluloid. Early cameras blurred skin like an accidental filter, and Marilyn Monroe became the saint of mid-century glamor: carefully sculpted by lighting, makeup, and the machinery of fame.

By the early 2000s, “Photoshop” became cultural vernacular and a coping mechanism. If someone looked impossibly gorgeous, we’d discredit them, saying: It’s Photoshop. At least there was a real person underneath…

Then social media arrived with sliders, presets, and Facetune. At this point, we were producing the altered imagery that we once consumed. And editing our own faces became a casual expectation.

The tech that set the stage

The rise of social media platforms didn’t invent beauty anxiety, it just industrialized it. With every mirror selfie or tagged photo, we added our faces to a public archive. Suddenly how we looked shifted from being a private moment, to data.

For the first time, we could make a movie out of our own lives. Photoshop became cultural shorthand and with the help of Facetune, optimization became the default. While the unfiltered face became a statement.

The conditioning started even earlier in Japan’s PuriKura booths. They were cute, but they made a slightly altered self feel normal.

So by the time Snapchat, Instagram, and TikTok rolled out beauty filters, we were already fully primed. And as the technology improved, so did the expectations. Surgeons reported an increase of patients bringing filtered versions of themselves (or other people!) to consultations. Without realizing it, we started treating our natural faces like drafts.

AI didn’t spark this mindset though, it just took the body out of the equation.

The AI aesthetic regime

Take Aitana Lopez, one of the first AI “it girls”. She has brand deals, fitness content, and a feminine persona engineered for maximum engagement. She doesn’t exist, but she’s treated like she does, complete with unhinged DMs.

But Aitana is a small fish in a massive pond.

AI-generated women are generally pretty, symmetrical, flawless, ambiguous but Eurocentric (unless explicitly prompted otherwise), and rarely plus-sized. They all kind of look like they could be cousins.

Seraphinne Vallora is the company behind the controversial Guess campaign featured in Vogue’s August 2025 issue. They’ve admitted that their tech “struggles” with generating diverse body types. Translation: the data isn’t diverse, and the financial incentive to fix that is low. Engagement, after all, is their north star, and only the images that perform survive.

In the attention economy, engagement is everything. The last time they posted an unambiguously black AI woman was in June 2024 and currently sits at under 90 likes. A white AI woman posted the same day? Over 3,500. These aren’t critiques of the images themselves, but signals the algorithm interprets as universal truth.

None of this happens in a vacuum, though. During the early pandemic, leaked TikTok moderation documents suggested experiments in down-ranking “ugly, poor, or disabled” users. TikTok denied it, but the underlying logic matters more than the leak. Appearance-based gatekeeping wasn’t a bug, it was a preview.

AI learns beauty from us, but the most distorted version of us. The training data for these generation algorithms is full of influencers, airbrushed bodies, symmetry worship, and Eurocentric defaults… A narrow slice of humanity platforms quietly boost. And then the machine extrapolates, abandoning the spectrum of human beauty for a distilled version of the same ideal.

The machine is choosing probability over beauty, really. Ask one to create a “beautiful woman”, and she’ll likely be white, slim, and poreless. The machine doesn’t necessarily believe this, we taught it what succeeds.

It’s a feedback loop wearing a pretty face.

Meanwhile, brands love efficiency. AI women don’t age, sleep, travel, or request pay… They simply deliver the most clickable and monetizable face money can buy. On demand.

And because regulations around labeling AI-generated imagery are inconsistent (or nonexistent) the line between “aspirational beauty” and “algorithmic hallucination” keeps dissolving.

The human cost of optimized beauty

We may think that we are the ones choosing what we find beautiful, and maybe that was the case once. But increasingly, the algorithm chooses first and we subconsciously adjust. Our feeds are a self-fulfilling prophecy: the machine shows us what “performs”, we reward it with attention, it interprets our attention as desire, and our taste shifts in the process. Eventually, we stop noticing that our taste has been domesticated.

And psychologically, the fallout is real. Facial dysmorphia is rising and it’s fueled by the impossibility of measuring ourselves against people that never age, bloat, or break out. Our reflections start disappointing us because humanity now feels like a flaw.

We’ve also become terrified of aging. Youth has always been currency in beauty culture, and AI has inflated its value. People are getting preventative Botox in their early twenties and ten-year-olds are badgering their mothers for retinol. We’ve conditioned a generation to treat aging like a preventable disease instead of a sign of life.

And then there’s the uncomfortable truth that most of us are average. Mathematically, that’s just the definition. But socially, “average” has evolved into an insult. Mid. A word used to describe Margot Robbie in 2023 by certain corners of the internet. It just tells you how warped the scale has become. Being a five out of ten, mathematically, should mean looking like the majority of people. Instead, it’s treated like a moral failure.

AI beauty flattens everything that makes us human: the wrinkles, pores, body hair, crooked teeth and noses, disability, and racial nuance. Anything that adds depth or individuality to the perception of beauty gets thrown out for optimization.

And unsurprisingly, it hits women the hardest. It’s not that men are exempt from beauty pressure entirely, their anxieties are just monetized differently. And the cultural archive AI learns from is lopsided. The image of women has historically been photographed, aestheticized, and monetized at a far higher volume. And the machines are just mirroring that history at scale.

AI isn’t practicing eugenics, but it definitely has a type. And when algorithms with global reach start to amplify one version of perfection, we drift closer to an aesthetic echo of a dangerous old belief: that beauty can be engineered into a singular truth.

In defense of AI generation as a tool

Before declaring AI the new villain of beauty culture, it’s worth remembering that this didn’t all start with algorithms. Victorian painters routinely idealized their sitters. Early photographers scraped cinched waists onto negatives. Beauty has always been part documentation, part aspiration.

What has changed is access. Controversial, I know. But, AI has genuinely democratized creation. A two-person self-funded team can now create visuals that once needed an entire studio ecosystem. And emerging designers can now experiment, iterate, and test ideas that may have been financially impossible. I doubt most of them are trying to erase humanity. They’re just trying to survive the attention economy.

However, tools reflect the priorities of the hand that wields them. When the intention is optimization above all else, the results mirror that. And once those outputs enter our feeds, they start shaping us back.

Moving forward

Part of why this all feels so destabilizing is that biologically, we were never meant to see this many faces this often. And certainly not this many conventionally attractive ones. For most of human history, beauty was shaped by the small circles we lived in. Now we scroll past more “perfect” faces in a day than our ancestors would have seen in a lifetime. So it’s no wonder we’re overstimulated (and hyper-aware of every fine line). Our brains aren’t wired for comparison at this scale.

The answer isn’t to swear off tech and move into the woods. Awareness may protect us more than abstinence here. A little transparency from platforms, brands, and from the people developing these systems could go a long way.

And the same goes for us. The feed is an ecosystem that we play quiet roles in tending to. Intentional curation may feel like a chore, but it shapes the signals we send back into the system. The more space we make for varied, real, human faces, the more signals we send about what beauty actually looks like. To both the machine and ourselves.

Because the truth is, AI never stole beauty from us. We built this machine. We trained it on everything we worshipped and it learned our preferences a little too well.

Maybe the most radical thing we can do isn’t to opt out, but to opt back into our own physical nuance.

And maybe relearn how to love the parts of ourselves the machine can’t smooth out.

Thank you for reading. ✦

- Rain

Very related post:

More interesting stuff:

• The history of the concept of symmetry by Michael I Selzer (2016)

• The AI Labeling Act of 2023

• Why Preventative Botox Injections Could Be Aging You by Lauren Valenti, Chloe Atkins, and Audrey Noble (January, 2025)

But working on similar algorithms professionally made that feeling of anticipation make less sense with every passing year. I already know what I listen to! Most of us do.

What interests me more now is how Wrapped has become a cultural event, prompting think pieces, memes, identity discourse (and plenty of free Spotify marketing). An annual ritual of self-definition where we let an algorithm tell the story of our year. Every timeline floods with people judging each other’s listening habits (jokingly and… not so jokingly), declaring how seen they feel, and questioning why the algorithm miscast them as a “cottagecore melancholic maximalist”.

Part of me loves the conversation the event sparks. But another part of me cannot help but question what we’re truly celebrating. Is this not a celebration of an algorithm’s ability to predict, compile, and present a stylized mirror reflection of who it thinks we are? Maybe that’s where the appeal is. It’s not just a summarization of our listening habits, but a neatly packaged version of ourselves, with an invitation for us to agree.

One thing is undeniable: we are deep into an era where machines and algorithms don’t just serve us; they attempt to tell us who we are. And the question arises: how much of that story are we quietly letting them write?

And how much are we willing to adopt it?

The business logic behind the “tidy version of you”

As a machine learning engineer and data scientist I must admit, personalization can feel magical. Pinterest never fails to draw me in the second I open the app. I can’t imagine what platforms like Youtube or Instagram would be like if I had to sift through lawn mower ads and videos with click-bait thumbnails. I’d rather have my Katseye fancams, skincare routines, and tennis highlights in my face, greeting me on launch.

However, this all relies on a subtle assumption that is not so subtle to corporations: that you are knowable, classifiable, and optimizable. And that who you are can be predicted based on how you behave on these platforms.

Corporations prioritize efficiency, coherence, return on investment… It works in their favor to reduce you to a digestible set of data points. The easier it is to slot you into a marketable cluster or two, the easier it is to serve you content that you’re likely to engage with, and the more ads they can show you along the way. The more predictable you seem, the more the machine can steer you, giving less room for the parts that don’t fit into a box. In practice, it’s a win-win situation.

In theory, the algorithm “knows” who we are, but at what cost?

The machinery

These algorithms, at their core, generally work the same way. For platforms that rely heavily on your engagement, user profiles are built based on a variety of micro-signals (clicks, watch time, scroll speed, pauses…). From those patterns, the algorithms are trained to learn what catches your attention versus what you’re likely to scroll past.

The tough part is that even AI developers cannot fully explain why these particular models capture certain patterns over others. These are referred to as black box models because you can see the input and output, but not the logic in between. So if you watch a few cat videos as a new user, you must be a cat person, right? Until you provide contradictory data. Meaning the identity you get back can be an accident based on misinterpretations made by the algorithm.

In that sense, these systems aren’t too different from horoscopes, personality quizzes, and archetypes. The difference is that algorithmic predictions are personalized, optimized, and relentless. Not to mention backed by the billion-dollar industry of advertising.

A concept: the “Algorithmic Self”

After enough time engaging, interacting, liking, and commenting… your feed starts to feel like you.

But is it?

Maybe, but unlikely. What you’re witnessing is a cleaned-up, legible version of you. Many researchers have referred to this as the “Algorithmic Self”. A version of you that fits into aesthetic “-cores”. A version of you that is trackable, placeable, and monetizable. A version of you whose taste can be forecasted like tomorrow’s weather.

It’s a version of self with the messy contradictions cropped out. But in the eyes of the machine, those parts aren’t easy to monetize. And things get eerie when we start performing the version of ourselves that the algorithm has assigned to us. At that point, identity becomes a negotiation: our impulse on one side and the algorithms expectations on the other.

We lose our sense of taste for the sake of coherence and for these platforms that’s ideal. Coherence facilitates targeted advertising. We’re not just being sold products, but lifestyles, routines, aesthetics, and aspirations. An idea of “you” that can be advertised to, with uncanny precision.

Pushing people into boxes doesn’t only come with the risk of losing your attention or money. It also risks radicalization. Not just political, but aesthetic or lifestyle-based. People can get radicalized into gym-bro ideologies, or coffee-shop-matcha-sipping aesthetics. Subcultures become identity clubs which can be echo chambers in and of themselves. Your taste narrows. Your opinions narrow. And your sense of possibility quietly shrinks. Where does your identity end and the machine-made classification begin?

Algorithms risk predicting more than our identities; they assist in writing the narrative of our lived experience.

A case for personalization

It’s important to acknowledge what personalization does well. I can say for myself that I’ve discovered music I love, and niche creators and micro-communities that have offered unique perspectives I didn’t know I needed.

A good recommendation can even feel like a friend remembering that new restaurant you mentioned a month ago.

Could it be that the danger isn’t in personalization, but personalization without self-awareness? The convenience of being known replacing the introspective work of learning and defining ourselves?

Staying human in a predictive world

How do we push back without deleting our Instagrams and Pinterest accounts? If algorithms compress us, then the cure is reintroducing friction, unpredictability, and contradiction.

Frequent unoptimized spaces with no algorithmic mirror. Volunteer, hike a mountain, wander around parks, museums, or libraries.

Let boredom creep in. Boredom forces introspection and makes space for thoughts that don’t come pre-curated.

Seek offline community. Book clubs, run clubs, wherever you feel you’ll find your people.

Break the patterns you're aware of. And do something out of your comfort zone.

Practice opacity. The algorithm can’t optimize the parts of you that you don’t post.

Follow curiosity and discovery instead of recommendation. Let serendipity thrive.

Diversify your media intake. Don’t let singular forums and online spaces define your entire worldview. Here's a great place to start.

This isn’t about rejecting technology in its entirety. Personalization helps up discover subcultures, build community, and hear voices we may never encounter offline. This is about expanding the possibility space again. The horizon of your potential selves.

A return to self

So this year’s Spotify Wrapped is almost here. And I’d be lying if I claimed that I won’t swipe through it at least three times and share it with those who know me best. However, Wrapped serves as a metaphor for the algorithmically-run era we are living in. One where machines don’t just hold a mirror to who we are but also (inadvertently or not) co-author our narrative.

Personalization can be enjoyed without outsourcing the personal work of forming our identity to the machine.

If algorithms flatten us, then our responsibility is to stay expansive. To remember that the most interesting parts of us are the parts that the machine can’t quite figure out. The contradictions. The parts that refuse classification. The parts that keep us human.
The parts that remain beyond the algorithm’s reach.

Further Resources

• When the Moon Hits Your Eye: Symbolism and Meaning in the Work of Odilon Redon by Sarah Mirseyedi

• The state of personalization in 2025 and beyond by Esat Artug

• Why’s Everyone on TikTok Now? The Algorithmized Self and the Future of Self-Making on Social Media by Aparajita Bhandari and Sara Bimo (2022)

“We shape our tools and thereafter, our tools shape us.” — Marshall McLuhan

Something happens when I’m scrolling on Instagram Reels or Youtube Shorts (my two go-to dopamine fryers of the moment). I’ll be five minutes into my feed, somewhere between the fifth trend I’m apparently late on, or the third AI-generated video that looks a lot more realistic than yesterday’s batch, and suddenly I realize that I’ve been staring at my screen for an hour. Scrolling… Hooked… Dopamine-fried… Yet completely uninspired.

And I’m aware of it. I know I’ve fallen victim to the craftfully engineered algorithm once again. I can see the levers being pulled and that’s because I’ve watched it happen to the people around me as well. Those close to me getting slowly pulled into ideological rabbit holes by their For You Pages, others spending money they don’t have just to keep up with trends that will expire in five to ten business minutes. All while losing their idea of self without noticing. The most ironic part? I have a special connection to these algorithms because I’ve helped build them.

hi :) I’m Rain.

I’m a machine learning engineer and data scientist with professional experience in Big Tech, and a Master of Science in Artificial Intelligence (that I earned before the release of ChatGPT). I’ve worked on user profiling, recommendation systems, AI-assisted care projects, generative models, and responsible AI. I’ve sat in on meetings where we’d talk about “engagement” and “click-through rate” all while knowing that these metrics would translate to colonizing someone’s attention span or leading them towards another purchase that they likely will not need.

Over time, my explorations and projects within the realm of responsible, ethical, and transparent AI gave me a vantage point on the field’s nuance, and how easily that nuance gets lost when we talk about AI at scale. The field is complicated and moving quickly, and it could be worth creating space to pause and think about these changes.

When I first started studying and working in the field, I noticed how… misaligned public conversations were. It was a constant swing between two extremes: one of apocalyptic fearmongering or promises of a utopian AI-optimized future. These perceptions also shaped many of my non-tech acquaintances’ first impressions of me upon learning what I studied and did for a living. I always found these perceptions to be sad but also somewhat amusing considering that most of the world interacted with AI on a daily basis (whether we were aware of it or not), well before ChatGPT or “AI girlfriends” were making headlines. Youtube recommendations, Spotify playlists, email spam filters, autocorrect, ride-share pricing, online shopping feeds, TikTok’s ability to steal hours of your time… AI is not new. It has just suddenly become visible, a culture of its own, and everywhere.

Those early impressions stayed with me. As the years went on, I kept noticing versions of the same thing: the conversations we were having around AI (in company stakeholder meetings, on social media, in the news) felt either incomplete or distorted. Meanwhile, the technology kept accelerating. I often felt small in a field that is fast, big, and loud, but also knew that I could share my perspective from the vantage point I had.

Muse in the Machine is a space for curiosity. A space to think out loud about the era we are currently moving through. Not to add to the jargon. Not to fuel the hype. But to create a space for conversation without getting lost in the noise. AI is likely our generation’s internet, maybe even our generation’s electricity… perhaps even our generation’s fire. Whatever we want to call it, it isn’t going anywhere.

And that’s where the title comes in.

The choice of Muse in the Machine came from a realization that changed the way I saw the field: We are the muse inside the machine. Machine-learning systems don’t dream on their own, they remix our patterns, our language, our art, our behaviors, our biases, and even our desires. They are trained on our digital exhaust and shaped by our choices. In subtle, pervasive ways, AI already influences what we want, what we buy, how we see ourselves, what is trendy, who gets to be heard, what we believe is the truth, and how quickly all of these things shift. Underneath all of that is us, reflected at us by the machine.

This is not an attempt to predict the future. It’s an attempt to understand the present. A present that is increasingly algorithmically-driven.

Expect musings on:

• algorithmic influence on taste, culture, and personal identity

• AI-driven microtrends and the tension between originality and imitation

• art, history, and cultural narratives that illuminate this AI-centered moment

• consumerism, capitalism, and the systems shaping our desires

• and responsible, ethical, human-centered AI

Welcome to Muse in the Machine ✦