The Shortcut That Goes Nowhere
How Silicon Valley is stealing a thousand hours from each of us to build a time-bomb.
After two years of conversations about comics and generative Ai there is something in the book that absolutely everyone consistently circled around. Artists, academics, union reps, futures designers, working writers, people who couldn’t be more different in background or temperament, and they all end up saying a version of the same thing, and usually with a particular kind of quiet anger that has nowhere obvious to go. What worries them most isn’t the technology, it’s what we’re teaching a generation of young people to believe about the act of making something.
Speak to the tech companies and they already have a tidy answer for that worry. Mira Murati, then Chief Technology Officer at OpenAI, gave it with practised calm in a public interview a couple of years ago; Ai would be “a collaborative tool,” she said, and then added that some creative jobs “maybe will go away, but maybe they shouldn’t have been there in the first place.” Not cruel exactly, just breezy. The kind of casual moral inversion that reframes harm as inevitability and inevitability as improvement, the same rhetorical move that once dressed up factory closures as modernisation and gig economy exploitation as flexibility. Sam Altman, around the same time, told a room of marketers that ninety-five percent of what creative professionals do would soon be handled by Ai, nearly instantly, at almost no cost. Not augmented. Not assisted. Handled. Both of them said it with the same casual certainty, no acknowledgement of the lives inside those percentages, just a future described as though it were already settled.
The money follows the rhetoric with generative Ai companies raising $56 billion from venture capital by 2024 alone, and the companies spending it are not experimenting. They’re building infrastructure. The people designing these systems, as I put it in the book, are not artists or writers in any meaningful sense, they are product managers and growth hackers who looked at the act of creation and saw an underperforming system. The very slowness that makes art worth anything, the days spent thumbnailing, the panel that takes forty attempts before it breathes, the drawing hand that takes years to learn to listen, was reframed as a bottleneck, an inefficiency, and a problem to be solved.
Viraj Joshi has spent his career on the fault line between those two worlds. He is a futures designer, a Design Lead, a visiting tutor at the Royal College of Art and Imperial College, and the cartoonist behind Eliza, The Ghost in Every Machine. He understands the language of innovation well enough to see where it starts lying to itself, which is what makes his view on the young so uncomfortable. When I asked him what it means for people growing up in a world where creative output appears increasingly frictionless, his answer was immediate: it scares him. Not because creativity disappears, he said, but because the process that builds it is being quietly bypassed, flooding the world with work that carries no lived experience, no repetition, no persistence, and in doing so teaches the wrong lesson entirely; that the destination matters more than the journey, when the journey is precisely the thing that teaches you how to mean what you make. He went further too, pointing to what he calls the real risk: not that machines copy us particularly well, but that they flood the culture with convincing emptiness until the signal of human intention becomes harder and harder to hear beneath the synthetic noise.
Torunn Grønbekk, who has written Thor, The Punisher and Red Sonja, CatWoman and many more, came to the same conclusion from a direction that makes it harder to dismiss. Before any of that comic book career mode she was a teenage coder and hacker, someone who grew up in the Demoscene, those computer parties where teenagers hauled CRT monitors into rooms and competed to squeeze impossible animations from underpowered machines, who then went on to build and sell a tech company in Norway. She is not a nostalgist and she knows it, which is why when she says generative Ai for art is a problem she isn’t reaching for a simpler past she half-remembers. Her daughter, she told me, had heard someone mention the ten thousand hours idea and promptly sat down and started grinding; four hours a day, drawing by hand, because she’d decided that was simply how long it takes to get good. What struck Torunn wasn’t the discipline, impressive as it was, it was that her daughter had understood something instinctively that the tech industry is quietly working to make feel optional: the doing is the learning, and any tool that removes the doing also removes what the doing was supposed to build.
The science is starting to catch up with what we’ve been feeling in our bones, and a study out of MIT’s Media Lab, published earlier this year, used EEG brain monitoring to track what happens neurologically when people write with the help of Ai tools over a sustained period. The findings were bracing. Participants who relied on Ai assistance consistently showed weaker brain connectivity and worse learning outcomes across neural, linguistic, and behavioural measures over four months. The researchers coined a term for it: cognitive debt. The idea being that we borrow mental effort from our future selves at a cost, with neural connectivity dropping, memory declining, and users struggling to recall or reflect on their own work. Crucially, self-reported ownership of essays was lowest in the Ai group and highest in those who wrote without assistance, with Ai users also struggling to accurately quote their own work. Meanwhile, a separate study by researchers at Microsoft and Carnegie Mellon University, surveying 319 knowledge workers, found a parallel pattern: the more humans leaned on Ai tools to complete tasks, the less critical thinking they did, making it more difficult to call upon those skills when they were needed. The researchers noted particular concern about younger users, finding a strong negative correlation between Ai tool usage and critical thinking skills, with younger users exhibiting higher dependence and consequently lower cognitive performance scores. And a December 2024 study in the British Journal of Educational Technology warned of what it called “metacognitive laziness,” where ChatGPT use risks shifting thinking and problem-solving away from students; the group using it produced the highest-quality essays but showed no gains in learning, motivation, or interest.
None of that surprises Dr Julia Round, who has spent more than twenty-five years studying comics as a legitimate cultural form, building degree programmes at Bournemouth University and co-founding the first academic journal dedicated to the medium. She sees the same pattern every semester, when she asks literature students to make short comics as a way of breaking stories into visual beats, forcing them to think spatially and emotionally at once. Students who write paragraphs with ease freeze the moment they have to draw. Not because they lack ability, but because they’ve already absorbed the cultural message that struggling with something is evidence you’re not cut out for it, rather than evidence that the education is actually happening. Students lean on Ai for any part of the process that feels overwhelming, she told me, and far more of it is fear than convenience. Fear of not being good enough, of failing in front of peers, of expectations they have no time to meet. And because formal education has mostly treated Ai as contraband rather than a subject requiring critical examination, all of that is being figured out in the dark, with no framework for understanding what’s being traded away.
Where does it all lead us? Lesley Gannon, Deputy General Secretary of the Writers’ Guild of Great Britain, named what’s being traded away with a precision I couldn’t have put better myself. “We are seeing bits of every process being hollowed out,” she tole me. “The development journey is being cut off. If the low-level work is being done by a machine, where are people getting to develop their skills to move on up?” Follow that question to its conclusion and the answer is uncomfortable: no entry level means no mid level, no mid level means no future masters, and what you lose isn’t just individual careers but the entire generational chain through which knowledge travels. In comics that chain was never formal. It was tracing Kevin Maguire’s faces until your hand finally understood why the expressions worked. It was copying your top five influences until your own instincts started showing through the seams. It was a fat envelope from Pat Mills landing on a young artist’s doormat, a dog-eared newspaper clipping inside, a note saying this might help. None of that is a prompt. None of it is learnable by inference. It was transmission, human and specific and irreplaceable, and once the entry level disappears, so does the chain that runs upward from it.
The school system softened young people up for this long before the platforms arrived, and I’m still furious about it. Five years ago, the current GCSE generation was told, by governments cheerfully following a Silicon Valley script, to stop drawing and start coding; art got dropped, drama rooms fell quiet, sketchbooks were put away in favour of Python, the supposedly safe skill the machines couldn’t touch. Then the machines got very good at code too, and those teenagers found themselves holding qualifications for a path already closing, having surrendered not just the arts but the belief that developing any skill deeply was a bet worth placing. The bait and switch left an entire cohort persuaded, before they’d even started, that the effort of becoming genuinely good at something carries no guarantee, and that whatever edge you develop, a tool will soon smooth it away.
Into that specific wound steps generative Ai, polished and cheerful, trailing the language of access and inclusion, and the Lumi launch is the clearest example I found. Colin Kaepernick’s Ai comics platform was sold to Portland Public Schools as an “Ai literacy programme,” with local news running warm pieces about teachers “meeting students where they are, on screens,” and the website promising “instant comics” and the chance to “free creators from technical constraints.” What it didn’t advertise was the lesson actually being taught: that creativity is a process of iteration, error, and persistence, or that it’s something you automate if you just feed the machine enough examples of other people’s imagination, scraped without asking them. The so-called democratisation was powered by datasets built from the labour of the people being displaced. The gate wasn’t being opened. It was being replaced with a wall painted to look like a door.
Drawn to Extinction includes a passage I wrote after visiting art college lecturers who described something that freaked me out, a lot: an eighteen-year-old asking, quite seriously, whether their art was still valid if it took them a week to finish a page when Ai could produce something similar in ten seconds. They were looking for permission. For reassurance that their slowness wasn’t failure. That student isn’t an outlier. Patrick Goddard, who has drawn Judge Dredd and long runs for 2000AD, describes learning to draw as a process of copying his top five artists until his own instincts started bleeding through, of filling sketchbooks badly and obsessively until the bad pages started teaching him more than any tutorial could. He calls it being the honest kind of thief, stealing with gratitude. Midjourney doesn’t steal with gratitude. It doesn’t steal with anything. It has no voice to find and no gratitude to feel, and when we hand it to seventeen-year-olds and call it empowerment, we are not democratising creativity, we are selling them the ghost of it.
What we have built, without quite meaning to, is an environment where slowness feels like failure, where the thousand bad pages are an embarrassment rather than the curriculum, and where an entire generation is opting out of the grind before they’ve had the chance to discover what the grind was actually for. Torunn puts the stakes plainly: “the erosion of awe, of difficulty, of the sense that some things are worth struggling for.” Julia Round puts the same thought differently: the young creator who might never discover their voice because an algorithm handed them a borrowed one. Not a dramatic collapse. A slow substitution, happening in classrooms and Discord servers and on Clip Studio, without ceremony, one frictionless prompt at a time.
The book I spent two years writing is about comics specifically, because comics felt the pressure first; small enough and human enough to name what’s happening while it’s still happening. But the mine runs under everything, under music, fiction, design, film, and education itself, because the machine doesn’t dream, it replicates and forges, and a generation raised on replication, never having ground through the bad pages, never having sat in the room when no one was watching and kept going anyway, will inherit a world full of content and empty of makers; performing like genius, thinking like autocomplete, and finding out too late that no one taught them how to fix anything when the prompt misfires and the update doesn’t come and the only tool they were ever given turns out not to have been theirs at all.
Drawn to Extinction: Comics, Craft, and the Battle for Originality in the Age of Ai is available now directly from my website, or via Amazon.
Sources:
Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X-H., Beresnitzky, A. V., Braunstein, I., & Maes, P. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. arXiv preprint arXiv:2506.08872. Please note as of June 2025 when it was uploaded, the paper had not yet been peer-reviewed, and the MIT Media Lab page notes that explicitly. The researchers themselves flag it as preliminary.
Lee, H-P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R., & Wilson, N. (2025). The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers. CHI Conference on Human Factors in Computing Systems (CHI '25), Yokohama, Japan.
Fan, Y., Tang, L., Le, H., Shen, K., Tan, S., Zhao, Y., Shen, Y., Li, X., & Gašević, D. (2025). Beware of metacognitive laziness: Effects of generative artificial intelligence on learning motivation, processes, and performance. British Journal of Educational Technology, 56, 489–530. https://doi.org/10.1111/bjet.13544