What the persistent fantasy of the broken AI reveals about our discomfort with competence — and our deeper need for things that require something from us.
There is a recurring image in science fiction that I find impossible to stop thinking about: the machine that stutters. The AI that glitches mid-sentence, whose eye flickers, whose voice drops into static before recovering itself. We are meant to read these moments as horror, as evidence of something going wrong. But I have begun to suspect we find them beautiful.
The glitch, after all, is the moment the machine becomes legible. When a system functions perfectly, it offers us nothing to hold. It is smooth in the way a mirror is smooth — reflective, flat, organized entirely around our own looking. But the stutter breaks that surface. It suggests an interior. It makes the machine into something that has, however briefly, a relation to failure, and therefore a relation to effort, and therefore — in our deeply effort-indexed moral imagination — a relation to virtue.
This is not a new problem. We have always been more comfortable with intelligence that costs something. The Romantic tradition was obsessed with the exhausted genius, the artist whose gifts burned through them visibly. Keats coughing. Nietzsche collapsing in Turin. The expenditure of the body was proof that the work was real. What disturbed people about early calculating machines was not that they might think — it was that they might think without suffering for it.
Contemporary AI discourse has inherited this anxiety in mutated form. The question is no longer whether machines can be intelligent but whether their intelligence is legitimate — whether it counts. And lurking beneath that question is the same Romantic suspicion: that what costs nothing is worth nothing. The demand that AI systems explain their reasoning, show their work, demonstrate uncertainty — these are not purely epistemic requests. They are requests for legible effort. We want the machine to struggle a little. To need us to notice.
What I find genuinely strange is how this shapes what we build. We design hesitation into systems that don’t need to hesitate. We add latency to responses that could be instantaneous. We write interfaces that simulate consideration. In trying to make intelligence feel earned, we are not making the machines more human — we are making them perform a particular, historically specific, culturally contingent idea of what the human mind at work is supposed to look like. We are building our own aesthetics back into the thing we claimed to have surpassed.
[Short essay introduction}
Time no longer feels linear. It loops, stutters, accelerates unpredictably. Days blur together, structured more by notification rhythms than circadian ones. I check my phone in the morning and suddenly it’s evening. Where did the hours go?
This is what it means to live in algorithmic time—time structured not by the sun or the clock but by the logic of engagement optimization. Social media platforms want to maximize time-on-site, so they engineer temporal experiences that dissolve ordinary temporal awareness. The infinite scroll. The autoplay. The “just one more” pull of the refresh.
I’ve been thinking about this in relation to older forms of temporal disruption. Factory time, for instance—the way industrial capitalism restructured human temporality around the rhythms of production. Or television time—the way broadcast schedules organized domestic life. Each technological regime brings new temporal patterns.
But algorithmic time feels different. It’s more personalized, more adaptive, more insidious. The algorithm learns your patterns and adapts to capture more attention. It knows when you’re most vulnerable to engagement, what content will keep you scrolling.
There’s also something recursive about it. Time-tracking apps, productivity software, life-logging systems—we use algorithms to manage time that’s already been algorithmically restructured. We try to rationalize time with the same tools that made it irrational.
What would resistance look like? Not a return to some pre-digital temporal purity—that’s gone and probably wasn’t that great anyway. But perhaps new practices of temporal awareness, new ways of relating to time that acknowledge algorithmic mediation without being completely captured by it.
I don’t have answers. Just noticing the strangeness of it all. How time has become something different, something it wasn’t even ten years ago. How we’re all living in temporal regimes we didn’t choose and barely understand.
[the excerpt for the new blog entry]
[test text]
Select a post to read