It’s become a meme at this point: the way we used to talk to ChatGPT with flowing paragraphs and proper punctuation when it first came out, versus how we now just bark incomplete, pithy sentences at it and expect it to get it. Worse, it usually does.

And it’s absolutely true. Back in the day (like, two years ago), when engaging with AI assistants, we would craft each prompt with the care and solicitude of a dwarven smith in Moria, using expansive notes and carefully laying out clauses to get the best possible responses.

Well do I remember the smug satisfaction of always saying please and thank you to them, confident that when the AI-pocalypse did happen, the machines would remember that courtesy and spare me obliteration (or grant it, depending on the state of the world) as mercy.

Today, you barely need to open a new thread, with personalized intelligence (now branded by Google, used by literally every tech company) remembering your conversations and contextualizing their output no matter how scattershot your input was to deliver an answer.

(“hey GPT, what’s that idea for a story that struck me when we were discussing which historical events a Song of Ice and Fire draws from. Not the municipal uses of Greek fire, the other one”).


Never mind the accuracy or veracity of said output (that’s a different story), but the fact that it can remember and respond in a highly customized manner really lowers the cognitive requirements on the part of the user.

And the bigger thing: it reduces friction.

It’s easy, breezy, no need to brain squeezy. Ask, and ye shall receive. And soon, you won’t have to even ask. It’ll know what you want before you want it.

AI companies didn’t come up with this. They simply learned from the best in the game: Google and Meta specifically, and social media and tech platforms in general.

And from a technical technological standpoint, it’s really incredible. From a human agency and ability lens: it’s fucking horrifying.

When YouTube has just the right video queued up below the one you’re watching, when Instagram has Reels cascading naturally as you swipe by each one in a sort of overarching motif, when Spotify spits out one banger after the other, it’s not just algorithms aligning, or even “Wow, I use this app way too much.” There’s something more insidious going on under the surface you’re searching, scrolling, streaming, shopping on.

That smooth lack of friction is easing away decisioning, and chipping away at more than just attention. It’s paring away your need to think, smoothly, easily, neatly.

That’s the thing about these seemingly smooth surfaces: the better these systems get at guessing what we’ll want next, the less practice we get at deciding for ourselves. Every time the next video, the next reel, the next answer just appears, we skip a tiny bit of work that we used to do: figuring out what we’re in the mood for, searching through things, weighing options, and finally choosing what worked best in that time and space in your life. None of that feels like much in the moment, but over thousands of micro-decisions a day, you’re basically outsourcing the muscle of wanting to whatever sits behind the feed.

We’re losing the ability to make decisions. Why bother, when an AI will do it for you, be it a Claude from Anthropic or the Al(gorithm) from whatever platforms you use most.

It’s happening to people my age and older, who were lucky enough to grow up in the analog-age and see that transition to digital media and communication. For kids who are growing up today sans the inconvenience and scratchy modem burrs of connecting to and engaging with news, entertainment, education, and other people, I wonder where their critical thinking is going to come from. And whether companies even want them to have it.

What really unnerves me is that a lot of the people I know and care about and who are having or trying to have kids are the good ones: present, thoughtful, and I’m sure they will try their best to ration screens instead of treating them like nannies, and spark their kid’s curiosity in the real world. And yet I’m not sure how much that matters when the default environment is this plugged-in, this optimised, this relentless.

You can be the most switched-on parent in the world and you’re still fighting an always-on attention machine that lives in your kid’s pocket and in everyone else’s. It starts to feel less like a question of “good” or “bad” parenting and more like the Ben from Captain Fantastic problem: you can take your kids off-grid, but the grid is still where the world (and their lives) will happen.

And that’s the thing, Dear Reader. It is going to happen.

While recently writing about how Meta looks poised to overtake Google for the first time ever in ad revenues, what staggered me was the sheer amounts of money being generated and almost solely due to the ability of platforms to retain user engagement, even at the cost of said user’s attention spans, cognition, and just plain stopping to think.

Google has pulled in a cool 1.3 trillion dollars from just ads since 2020 to the time of writing, while second place “loser” Meta has pulled in a mere 800-plus billion dollars in the same amount. That’s a lot of money for basically getting people to gawp at the magic rectangles now that run lives.

At that scale, it stops feeling like a media business and starts looking more like an extraction industry, except the resource isn’t oil or lithium, it’s our ability to sit with a thought before something else rushes in to replace it. It’s the friction that helps us slow down and actually think.

Those trillions are the running tally of every moment we let a feed decide what comes next, every time we tap whatever’s put under our thumb instead of asking what we actually want. The product isn’t the ad impressions or the targeting sophistication; it’s the gradual outsourcing of choice itself to companies whose balance sheets get healthier the less we notice it happening.

Maybe that’s the real joke behind that meme about how we talk to ChatGPT now: we laugh at how lazy we’ve gotten with our prompts while the rest of our lives quietly start to look the same. Everything smoothed over, everything suggested, everything one tap away so we never have to sit in the gap between wanting something and getting it.

Maybe the only thing we can do, as a species is to start noticing how often we’re not actually choosing at all.

The feeds, the AI assistants, the recommendation engines are all brilliant machines for reducing friction, and we’re grateful for the convenience right up until the moment we realise we’ve outsourced the part of ourselves that decides what’s worth wanting in the first place. If we don’t want that to become irreversible, the work won’t be in building a prettier interface or a smarter algorithm.

It’ll be in accepting a little more friction, in deliberately asking, “What would I want if no one had already suggested it to me?”, and then, every once in a while, actually going out and finding it. That’s the small, stubborn act of re‑asserting agency that no platform can safely monetise, and that, for now, still belongs entirely to us

All this is a long way of saying that if we’re not careful, we don’t just become the audience for the meme, we become the punchline, and that’s a gut punch.