Rob Horning
LightNews LightNews
robhorning.bsky.social
Rob Horning
@robhorning.bsky.social
4.7K followers 260 following 520 posts
robhorning.substack.com
Posts Replies Media Videos
robhorning.bsky.social
data-collection and programming are not sufficient to run an economy because economies should not be focused on optimizing for one particular goal newleftreview.org/issues/ii153...
This is where the limits of technical reasoning become clear. Determination of the optimal solution set is a problem for technical reasoning to solve, but choice among the options in that set is not, since from a technical perspective, these options are equivalent in rank, no matter how different they are in content. Hopes that advances in computational power might overcome any need for social and ecological choice—making it possible to digitally plan the economy—fundamentally mischaracterize the problem, once we recognize the existence of multiple, non-compensatory social goals. The challenge is not simply processing more data. It is deciding what matters—negotiating trade-offs and choosing among possible futures. Reason, experience and collective deliberation will inform these choices but never entirely resolve them.
January 2, 2026 at 6:56 PM
robhorning.bsky.social
“slop” as the new “nonplace” maxread.substack.com/p/what-is-sl...
One idea I’ve been circling around is that “slop” is that which is “fully optimized” to its domain to the point of texturelessness or characterlessness. “Slop” in this sense is anything designed to be as easy as possible to produce, sell, and consume, but it’s particularly slop at the point where all or most other players in the same space adopt the same strategies
December 19, 2025 at 6:34 PM
robhorning.bsky.social
the curse of streaming media: "the ability to cheaply deliver content is matched by its inability to produce anything new"
www.unemployednegativity.com/2025/12/livi...
The idea of AI produced art or music is in some sense of culmination of the dead weight of existing intellectual production over the living. One of the striking things of the streaming model, of movies, television, and music, is that its ability to cheaply deliver content is almost matched by its inability to produce anything new. Spotify can make the entire discography of dead artists available at the click of a button, but it cannot pay or support living artists; Netflix can buy up Warner Brothers, Disney can buy everything else, but the cost of this is an undermining the entire film industry by shifting distribution to streaming. In some sense AI solves a problem that the streaming services has created. Spotify may turn out to be better at creating the basis for AI generated songs than promoting living artists.
December 18, 2025 at 11:11 PM
robhorning.bsky.social
pursuing the dream of a life lived as pure "data creation," with the ultimate goal of being completely manipulatable by the data processors who have a plan for your life
hypervisible.blacksky.app Hypervisible @hypervisible.blacksky.app · 18d
Tech companies are actually pretty explicit about this: The point is to have a device that you can’t escape, one that always has access to you and makes you a constant source of data beyond what a phone does. Surveillance-first devices.
In 2025, wearables made a hard pivot to AI
If you want to put AI on someone 24/7, wearables are the best bet.
www.theverge.com
December 17, 2025 at 4:48 PM
robhorning.bsky.social
one of the dumbest things I've ever read
www.wired.com/story/people...
Petter Ruddwall knows the idea of AIs becoming sentient and seeking to get high with code-based “drugs” seems “stupid.” But the Swedish creative director couldn’t get it out of his head.
December 17, 2025 at 4:12 PM
robhorning.bsky.social
these findings suggest that people should always understand that interacting with chatbots is like talking to an advertisement, not talking to some supergenius
In two large peer-reviewed studies, AI chatbots shifted voters’ views by a substantial margin, far more than traditional political advertising tends to do.
December 16, 2025 at 7:49 PM
robhorning.bsky.social
this reminds me of theories about advertising, and how much we seem to want to believe that we are intrinsically hard to persuade and have something like "firm, established beliefs" when the evidence usually suggests otherwise
www.technologyreview.com/2025/12/05/1...
In the coming years, we will see the rise of AI that can personalize arguments, test what works, and quietly reshape political views at scale. That shift—from imitation to active persuasion—should worry us deeply.
December 16, 2025 at 7:35 PM
robhorning.bsky.social
when something that seems to belong to you (a device, your identity etc.) can be modified remotely at will by outside forces, you're always already expropriated
December 12, 2025 at 5:31 PM
robhorning.bsky.social
any device that is "always connected" is either working against us or forcing us to work unpaid for someone else
December 12, 2025 at 5:29 PM
robhorning.bsky.social
it's almost as if the purpose of such devices is to incapacitate you and make you dependent
December 12, 2025 at 5:01 PM
robhorning.bsky.social
i.e. how algorithmically sorted social media already works
sambiddle.com Sam Biddle @sambiddle.com · 25d
We need an app where you just input an opinion you hold and it generates two synthetic podcast experts explaining why you're right
maxtani.bsky.social max tani @maxtani.bsky.social · 25d
The Washington Post is launching a personalized AI podcast, saying users will be able to "shape their own briefing, select their topics, set their lengths, pick their hosts and soon even ask questions using our Ask The Post AI technology."
December 10, 2025 at 6:18 PM
robhorning.bsky.social
this gets at the question of whether one can close-read a generated text or image; the contingencies in them indicate something about the training data, but there seems to be nothing aesthetic in it www.tandfonline.com/doi/full/10....
In twentieth-century theory, however, the accidental, unintended, seemingly insignificant photographic detail has become a defining feature of the photographic as an esthetic phenomenon, most notably for Barthes (Citation1981). The famous ‘punctum’, the almost accidental detail that cannot easily be subsumed under the more obvious culturally coded meaning of the photograph, is analyzed by Barthes as a kind of excess of contingency that transcends any legible caption, written description and ultimately any intended meaning. As visual details that were never prompted, the strange distortions in AI-generated images also create an excess of visual contingency beyond the reproducible patterns, stereotypes and clichés that the model was trained to reproduce, thus transcending any intended meaning. In contrast to Barthes’s punctum, however, the contingency manifested in the impossible and distorted spaces of AI image generation is not the contingency of the material world, but the contingency of the statistical process that generates these images.
December 8, 2025 at 9:58 PM
robhorning.bsky.social
feel like an ethnography of the people who get paid by the piece to "rank" LLM outputs would also explain a lot about the kind of prose the models are optimized to generate
maxread.substack.com/p/will-ai-wr...
much of what we dislike about A.I.-generated text from a formal perspective--it’s generally cautious, inoffensive, anodyne, predictable, neutral, unmemorable and goes down smooth--is a product not of some inherent L.L.M. “voice” but of the training and fine-tuning processes imposed by A.I. companies, which are incentivized to make their chatbots sound as annoying and bland as possible.
December 5, 2025 at 9:10 PM
robhorning.bsky.social
not clear if the loneliness becomes visible to the chatbot user, though it certainly becomes visible to tech companies, who have incentive to perpetuate it
datasociety.net/points/all-t...
Being digitally alone, then, is not merely a social condition but a method of relating to oneself. In conversation with a chatbot, loneliness becomes visible as this existential condition is translated into text and archived, interaction after interaction. For some, this aids in the unfurling of patterns they could not otherwise name; it might surface feelings they would rather keep private. Technology mediates a fundamental, familiar act: talking to oneself via another. And what begins as a private exchange soon reveals itself as a social system, reflecting both the self as well as the architectures of isolation that shape us all.
December 3, 2025 at 10:10 PM
robhorning.bsky.social
also think this fantasy of mastery expresses itself as an opportunity for total surrender, to expect nothing from yourself
robinjames.bsky.social Robin James @robinjames.bsky.social · Nov 30
"Generative AI platforms center the idea that one has written, or created, by interacting with a structure they did not build and do not understand" - this I think is key -- less about interpassivity & 100% about the master/slave dialectic: the point is not to create but to be a master.
November 30, 2025 at 4:32 PM
Reposted by Rob Horning
robinjames.bsky.social
OK so this concept of 'interpassivity', which this author dates to the 90s, is exactly what Attali wrote about in 1977's NOISE as "repetition", or stockpiling hordes of use-time that you never actually use

mail.cyberneticforests.com/from-interac...
From Interactive to Interpassive
Where AI Art Meets Cognitive Offloading There is the joke about AI: we wanted robots to do our dishes so we could have time to make art, but we got robots that make art while we do the dishes. We can...
mail.cyberneticforests.com
November 30, 2025 at 2:39 PM
robhorning.bsky.social
interesting that Flusser invented gooning
What I have tried to put into words here is both a feverishly involved and a passionate state of mind, something like a synthesis of what absorbs people in artistic and scientific creativity, in political activism, in revolutionary proclamations, in chess and roulette, in the stock market, and in libidinous dreams. It is a state of mind that does not intensify and then fall away, as in an orgasm, but that maintains itself at its orgiastic climax without interruption through a lifetime. For this state of mind is not physical but cerebral. Images are steering the telematic society in this direction: toward a continuous cerebral orgasm.
November 14, 2025 at 3:30 PM
robhorning.bsky.social
I wrote something about generative music and reading Vilém Flusser robhorning.substack.com/p/the-songs-...
The songs remain the same
Into the universe of posthistorical music
robhorning.substack.com
November 14, 2025 at 12:26 AM
robhorning.bsky.social
seeing LLMs as "ideology that can pretend to be a person" helps make it clear what "talking to chatbots" accomplishes: Althusserian "interpellation"
hagenblix.bsky.social Hagen Blix @hagenblix.bsky.social · Nov 11
I do, of course, have a paragraph from "Why We Fear AI" for this
An LLM is an abstraction over the status quo (or rather, the text of the status quo) that has, at its core, no other aim than producing more text that is just like the text already in existence. One result is that the dominant ideas, the dominant textual patterns of our time, are recast as “intelligence” per se. We might call the complex mass of concepts and ideas, the channels into which ideas flow often and easily, or the boundaries of what is reasonable to debate in so-called “polite society”—the dominant ideology of an age. An age’s dominant ideology tends to presuppose that the power of that age is natural; the hallmark of ideology, in this sense, is that it is easy to recognize only when it’s past its time. We may no longer think that an almighty god appointed a king to rule us, that women are naturally unfit to vote, or that sex outside of a heterosexual marriage is immoral. But at various times, to think otherwise (to think of replacing the monarchy with a republic, to affirm that women are people, or being in favor of sexual or gay liberation) was the far more difficult, even dangerous thought—if not to have, then at the very least to put in writing. Ideology, of course, is never complete, other thoughts are always possible. But ideology is easy to think, it’s what our thoughts are already marinated in. And hence, ideology will be dominant in the textual patterns of an age. Seen in this light, what is called “intelligence” in a language model, is really that—ideology as a thing that can pretend to be a person.
November 11, 2025 at 9:11 PM
robhorning.bsky.social
this seems right, and always makes me wonder what the "suckers" get out of being on "sucker lists," why they enjoy being on them. Is it just some combination of masochism and getting paid attention to? Confirmation bias as a service? maxread.substack.com/p/prediction...
One way of thinking about Twitter, especially in the Elon Musk era, is as the latest and most sophisticated version of the mailing lists that have powered the conservative movement, and the various scams and cons and schemes that constitute it, since the 1960s--in other words, as a huge and immediately accessible collection of suckers, waiting to have their avidity and paranoia exploited for cash. The predictions markets here are happy to play the same role with heavy X.com users as crypto scams and meme-stock hustlers did on the same site a few years ago, or as cash-for-gold ads did with Fox News viewers a decade before that, or as direct-mail get-rich-quick schemes did with Goldwater voters deep in prehistory.¹
November 7, 2025 at 7:10 PM
robhorning.bsky.social
regardless of what else they convey in a specific image, image generators offer images that tell those people who want to hear it that "effort/craft is a sham and a waste of time"
November 4, 2025 at 7:20 PM
robhorning.bsky.social
seems like one of the main use cases for genAI is to eliminate the joy that other people take in thinking and actually doing things www.thecut.com/article/woul...
he’s watched the group spiral into the depths of AI paranoia as images of phony crochet projects and impossible-to-execute patterns flooded the feed
November 4, 2025 at 6:02 PM
Reposted by Rob Horning
hypervisible.blacksky.app
Sora 2 is not being “misused” when people use it to produce racist, transphobic, misogynistic, & otherwise vile images—it exists to make it easier to put these images out into the world. Calling it a “misuse” is a grave misunderstanding of what these companies are up to, & lets OpenAI off the hook.
November 4, 2025 at 3:22 PM
Reposted by Rob Horning
bildoperationen.bsky.social
However, the problem with these images is not just their genericness. It's the deeply populist idea that politics can be reduced to its immediately visible effects: politics is not judged by how it affects people's concrete daily lives, but rather by its aesthetics—by what image it produces
4/
Another AI generated image of a cityscape - now with vertical forest on the skyscapers and the Green Party logo in the corner
October 23, 2025 at 11:37 AM
robhorning.bsky.social
hope that is the actual cover
October 22, 2025 at 7:01 PM