Tessera

Critical AI https://read.dukeupress.edu/critical-ai/issue

@criticalai-journal.bsky.social

Critical AI's new issue is out! https://read.dukeupress.edu/critical-ai/issue Email us at criticalai@sas.rutgers.edu Our website and blog: https://criticalai.org/

1454 Posts6481 Follower 1434 Folgt

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 3d ago

Read Critical AI editor Lauren Goodlad's essay in #Representations new #clusteronAI (2026) 175 (1): 80–110. doi.org/10.1525/rep..... The cluster begins with Goodlad's "We Have Always Been Action Theorists" and includes four interdisciplinary responses.

We Have Always Been Action TheoristsToward a Critical Theory of Language for the Era of “Large Language Models”

Scholars of literature and culture understandably place themselves among the world’s premiere experts on matters of language. But they also know that fields like linguistics and communication have the...

doi.org

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last wk.

docs.google.com/presentation... Check out this new teaching resource by @drewuniversity.bsky.social's #JeremeyBlatter and #CriticalAI intern #AJMannino!

AI_Toll_Road_v1.pptx

Open by asking: what does AI cost? Then expand beyond subscription fees. 1 Jeremy Blatter, PhD Robert Fischer Oxnam Associate Professor of Science and Society; Media & Communications Dept. Chair Facul...

docs.google.com

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last wk.

None of us has read this yet but it looks very interesting!

Tim Newman@tnewmsblues.bsky.social · last wk.

The wild projections about an AI jobs apocalypse are what led us to read more of the research in this topic that was making headlines and we created this resource: aijobstory.org There’s a lot in the methodologies that make it unsurprising that the headlines have shifted!

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last wk.

www.wsj.com/finance/big-... Indeed...

Implicit in the analysts’ models is a belief that Big Tech can use AI to unlock rapid productivity gains, allowing revenue to grow much faster than overhead. Bringing in vast amounts of new business historically requires larger sales teams, customer support and administrative capacity. If AI tools fail to drive the hoped-for efficiencies, corporate expenses will scale with revenue, and the projected margin benefits won’t materialize.

ProPublica @propublica.org · last wk.

This video on the rise of Russ Vought is based on scores of interviews, thousands of pages of emails and dozens of hours of videos and recordings of private briefings given by Vought, most of which have not been previously reported.

Russell Vought: The Shadow President

YouTube video by ProPublica

youtube.com

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last wk.

www.nytimes.com/2026/07/27/o...

Opinion | If You’re Over 40, You’re Ready to Use A.I.

A.I. should work like this mystical Jewish practice.

nytimes.com

Critical AI https://read.dukeupress.edu/critical-ai/issue@criticalai-journal.bsky.social · last wk.

We don't wait for kids to turn 16 to hand over cars b/c "cars are evil," or "insist that surgeons train for years" b/c "scalpels are immoral. We build thresholds around powerful things b/c power requires judgment, and judgment can’t be downloaded..." Nice, @nytimes.com @debbiemillman.bsky.social

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last wk.

@wsj.com documents the gradual shift away from peak hype on the AI jobs apocalypse. Just. Not. Happening.

America’s biggest companies say they might need more people after all.

For months, major employers treated hiring as an expensive last resort. Now, a shift is emerging across industries. Companies ranging from railroad giant CSX to Google parent Alphabet GOOGL 2.87%increase; green up pointing triangle have told investors in recent days that they plan to hire to meet growth goals or to seize on emerging technologies.

The push to expand head count, at least modestly, is a reversal from the prevailing corporate messaging during much of the AI era. Major employers largely held back on adding people due to economic uncertainties or a belief that artificial intelligence could shoulder more tasks on the job. But some executives say the costs and limitations of AI now demand that more people be added; others want to hire people back following layoffs.

“We actually need to accelerate hiring a bit. We’re a little bit behind right now,” Booz Allen Hamilton’s Chief Operating Officer Kristine Martin Anderson told investors Friday. “We’re addressing that now.”

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 2w ago

@theverge.com reporter #RobertHart seems to nail: Trump admin regards "science" as a broified start-up (including grift and pay-to-play), not a public good. www.theverge.com/science/9705...

The tech-broification of American science has officially begun

Trump’s ‘Golden Age’ of American science starts with dismantling it, while funneling billions into AI. 

theverge.com

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 2w ago

www.bbc.com/news/article... Google in the poorhouse! "The company's free cash flow, the cash it maintained after paying for operations and investments, came in at negative $5.9bn (£4.3bn) for the first time in at least a decade, according to its past financial records."

Google burning through cash with spiralling AI costs

The company said earlier this year it expected to spend as much as $190bn on AI investments.

bbc.com

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 2w ago

www.chronicle.com/article/is-a... Interesting article in which American Anthropology Assoc prez Carolyn Rouse tells the Chronicle that a report criticizing humanities scholarship was an "embarrassment of embarrassments" that used AI.

Is Anthropology Hopelessly Politicized?

We asked Carolyn M. Rouse, head of the American Anthropological Association. She had a lot to say.

chronicle.com

bityz.bsky.social @bityz.bsky.social · 2w ago

"Because LLMs are trained on math, coding, and science problems—tasks that reward generalizing from just a few examples—they can settle on a hunch too early. And the same instinct that helps LLMs crack logic puzzles also makes them quick to stereotype"

Critical AI https://read.dukeupress.edu/critical-ai/issue@criticalai-journal.bsky.social · 2w ago

Nice coverage from #MITTechnology Review on study showing the persistent tendency to reproduce observed stereotypes and turn correlations into grounds for new biases. This is how classifiers work! www.technologyreview.com/2026/07/20/1...

The models quickly started segregating candidates from different groups into different jobs on the basis of early observations of hiring outcomes. For example, when a model was told an Aima had failed as a doctor, a job considered to require high levels of warmth and competence, it veered away from hiring all Aimas as doctors. Instead, it started hiring Aimas as janitors, which the model classified as being less warm and competent than doctors.

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 2w ago

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 3w ago

This prez fell prey to reflexive "AI first" 'tudes www.chronicle.com/article/how-... This seems right "Clay Shirky [of NYU]... suggested that the best thing" a prez "can do...is to fall back on what it has always been good at: “convene a broad, complicated conversation that has no easy answers."

How ‘Chaos’ Surrounding AI Can Sink a College Presidency

Newly released public records show the extent to which disagreement over AI strategy drove the University of Wisconsin Board of Regents to oust Jay Rothman, the system leader.

chronicle.com

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 3w ago

#CriticalAI4.1, Its early, but you know what I want for Christmas? Media scholar Gavin Mueller's essay on how STS scholar Jathan Sadowski's work can better structure the opposition to corporate tech's expansion of AI! Don't forget to check it out! read.dukeupress.edu/critical-ai/...

Published during a time of increasing anxiety and dissatisfaction around both technology and the economy, The Mechanic and the Luddite deserves a wide readership, even without spelling out concrete alternatives. Its precision with technical details and political economy should appeal to scholarly audiences, and its acerbic zeal and accessible style will likely earn it eager readers beyond the academy. Sadowski has threaded several tricky needles: a book with a wide range that isn't unwieldy; scathing critique that is amusing and energizing rather than depressing; an approachably quippy style that does not compromise the seriousness of its aims. Luddites, seasoned or budding, will want to take a look.

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 3w ago

#CriticalAI4.1, Remember the Titans? No! Remember to check out literary scholar Bruce Robbin's reflection on John Warner's work where he argues that the soulless prose of AI generated writing is reflective of human writing in bureaucratic, impersonal contexts! read.dukeupress.edu/critical-ai/...

What he found in the epitaphs, to his chagrin, was the very poetic diction he despised: a mechanical repetition of conventional words scraped from the common word hoard, clichés by which people tried to express what they thought society required of them. In a sense, the writing on the tombstones resulted from the same pattern-finding that Warner attributes to ChatGPT, even though it was not mediated at tremendous scale through a new technology. As Warner very properly insists, the statistical modeling of internet-size data sets is a wholly nonhuman computational process; but as far as writing is concerned, its likely effect is not wholly inhuman. When commentators use the term “AI slop” they are contending with a massive increase in a sort of bad prose which is now beginning to dominate the internet. The expansion itself marks an effort to game the internet for economic or political motives; but the precedents for the bad prose are all too human. We can be abashed that in just a few years the internet has already become more than half “slop.” But is the question of what to do first and foremost a question of the human?

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 3w ago

www.change.org/p/tobin-enge... Interesting: petition said to be started by 7th grader to eliminate IXL, a supposedly "personalized" K-12 learning tool. "Personalization" is supposedly the superpower here; students say it is demoralizing and counterproductive.

Sign the Petition

Shutdown IXL for Public Use for Educational Purposes

change.org

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 4w ago

#CriticalAI4.1, Be sure to check out literary scholar James Dobson's drawing on of AI interdisciplinarian Nina Beguš' work to explain what humanists must do to effectively make sense of gen AI! read.dukeupress.edu/critical-ai/...

Artificial Humanities has a bold agenda: Nina Beguš wants to establish a new field that will inform the development of current technologies by exporting insights from literature and “the humanistic study of technology at large” (218). Such “artificial humanities” would bridge humanistic critique with an analysis of AI. While Beguš’s goal shares some common terrain with Critical AI, it differs from critical AI studies somewhat (at least as this reviewer understands it) primarily in its rhetorical stance toward generative AI technologies (the topic of a recent two-part special issue in CAI 2, no. 1 and 2, no. 2). Riffing on Bruno Latour's well-known polemic against critique, Beguš at times adopts a postcritical position, writing that “with technology like conversational AI, the time for critique has run out” (2). But Beguš is also not interested in technological approaches to literature, as in the work of computational literary studies. Instead, she seeks to promote the development of technologies that are “informed by the arts and humanities, resulting in fairer, deeper, and more important work on all sides” (2), a goal shared with academic and industry “AI ethics” efforts focused on aligning AI with human values. She envisions a “technologically and scientifically informed knowledge of the humanities used for thinking and building current and novel technologies” (212).

🌶 David Gray Widder @davidthewid.bsky.social · 4w ago

🚨Finally published: How does the US Military enlist AI researchers in war? @siree.sh, Lucy Suchman and I examine 7k US Military grant solicitations for AI research, and show how even "Basic Research" enlists researchers in a warfighting agenda. journals.sagepub.com/doi/10.1177/...

Basic research, lethal effects: Military artificial intelligence research funding as enlistment - David Gray Widder, Sireesh Gururaja, Lucy Suchman, 2026

In the context of unprecedented U.S. Department of Defense (DoD) budgets, this paper examines the recent history of DoD funding for academic research in algorit...

journals.sagepub.com

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 4w ago

#CriticalAI4.1, Reminder to check out historian Amira Moeding's use of Yorick Wilks' work to challenge the idea that AI systems can "understand" language! read.dukeupress.edu/critical-ai/...

What sets Wilks's book apart from other public-facing introductions to AI research (in addition to the beautiful illustrations in the new edition) is the author's philosophical meticulousness around the questions of what different methods and evaluative measures can and cannot achieve. Wilks brings us back to questions at the heart of AI research: What would it mean for a computational system to encode a “representation” of the world, and how can we achieve that goal? If a system lacks such a representation, how can it be said to “know” anything or act “intelligently”?

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 4w ago

#CriticalAI4.1, Don't forget to check out digital studies scholar Kat Brewster's assessment of the heteronormative biases of AI algorithms! read.dukeupress.edu/critical-ai/...

In a moment when public discourse around AI ethics, bias, and algorithmic harm becomes more vital, The Digital Closet is a necessary text. It offers a compelling case for how queerness can be the lens through which we navigate an evolving critical understanding of digital systems. Monea's work widens the circle of critique to include voices and identities too often excluded from computational legibility. As we imagine more democratic, accountable, and inclusive technological futures, The Digital Closet reminds us that the question is not simply who AI systems might see, but also how such systems were designed to see some people and not others.

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 4w ago

#CriticalAI4.1, Remember to check out literary critic Alexander Hartley's critique of corporate AI's aesthetic "hijacking" from the lens of artist Hito Steyerl's work. read.dukeupress.edu/critical-ai/...

These Steyerlian collocations—deft, brash, affronting—succeed on one level: They generate a style that seems able to process, in double time, the absurdities of the wide range of material she discusses. But they sometimes prompt the question of whether her associations are any more than that. The collocation of Stable Diffusion's bizarre limbless renderings of human bodies and the disabled bodies of the 1960s victims of thalidomide poisoning, in “Knuckleporn or Phocomelia,” illustrates the limitations of this style. The smoothness with which Steyerl moves from one subject to the other obscures the essay's lack of engagement with the historical, technological, and social differences that separate them. Steyerl's text reasons that the thalidomide victims are an example of corporate disregard, and the existence of corporate disregard shows the need for regulation; and AI corporations are corporations and thus bound to exhibit corporate disregard, so they need to be regulated. The Stable Diffusion–generated images of limbless children that pepper the essay, however, distract from this simple argument, seeming to suggest a deeper transhistorical or metaphysical equation: Corporations produce mutilated children. This kind of associativeness, which serves to aggregate rather than disaggregate and which hijacks all sorts of disparate phenomena into its dragnet, essentializes the phenomena at stake rather than analyzing them.

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · 4w ago

#CriticalAI4.1, Don't forget to check out roboticist Andrew Hundt's critique of automated solutions to eldercare from the perspective of disability studies! read.dukeupress.edu/critical-ai/...

Wright's book offers roboticists, researchers, corporate managers, and policymakers a sincere invitation to engage with the messy reality and internal contradictions of modern robotics research and development (R&D). Wright tackles one of the biggest motivating claims for robotics research: that an “aging society” necessitates the urgent development of care robots. The global population as a whole currently maintains a broad demographic balance across age groups, despite the growing proportion of older individuals due to longer life spans and declining birth rates (PopulationPyramid.netn.d.a., n.d.b.; United Nations n.d.). Even so, countries like Japan anticipate a surge in the proportion of older people who require assistance with daily activities like eating, cleaning, and hygiene. Proponents of a robotic model of care, which Wright labels “algorithmic care” (56), often emphasize estimates of the strain eldercare could place on budgets and national productivity. In response, they propose robots as a solution, with two separate books bearing the titular message Robots Will Save Japan (Kishi 2011; Nakayama 2006). Roboticists deploy this “aging society” argument so pervasively that a recent study by Brandão (2021) found it to be present in 30 percent of all papers across a range of top robotics publications. By contrast, Wright makes a formidable argument that “despite [the] considerable hype, lofty expectations, and substantial investment, robots alone cannot yet deliver on the promise of solving care crises in Japan or elsewhere” (3).

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last mo.

#CriticalAI4.1, Remember to check out historian Jamie Pietruska's analysis of how "bottom-up approaches" to data demystify it and reveal its political connotations! read.dukeupress.edu/critical-ai/...

This bottom-up approach to data justice is reflected in Everyday Adventures with Unruly Data, which investigates what we might think of as the data epistemologies of the everyday. Melanie Feinberg, an information scientist specializing in classification, takes readers on an engaging tour of a complex “landscape of data” in everyday life that weaves together her daily excursions in Copenhagen and the United States, her research and the scholarly literature on classification and data, and tales from her classroom teaching (2). Her trips to ordinary places like the supermarket, the doctor's office, and the library reveal that, despite widespread faith in the ideal of data as the basic building block of empirical accuracy and scientific certainty, the opposite is true in practice: The data we encounter in reality is messy, human, and ambiguous. In Feinberg's account, this ambiguity is a feature, not a bug. Thus, data demands more nuanced understanding as well as the acceptance that human attempts to tame data's unruliness will never be fully successful. Feinberg overturns the conventional understanding of data as orderly streams of numbers, convincingly demonstrating in case after case that “data is a story that we read and reread; it is not an equation to be solved."

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last mo.

#CriticalAI4.1, Don't forget to check out Russian language scholar Kitt Pribble's critique of automated, LLM-based translation modules! read.dukeupress.edu/critical-ai/...

Accepting the view of language as code means failing to recognize that true language proficiency lies in the capacity to engage meaningfully and intentionally with the target language, with a full understanding of the social and cultural beliefs and practices that language articulates. A language is a living, changing entity that shapes and is shaped by the history of its speakers, taking on new life each time it is adopted by (or forced upon) a new speech community, continuously marked by cross-cultural exchanges and transformed by encounters with new technologies. Syntax, grammar, and morphology bear the imprints of this history, and in studying them, a learner comes to a new way of conceptualizing the world.

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last mo.

#CriticalAI4.1, Reminder to check out media scholars Simon, Laywine & Sinnreich's assessment of gen AI's intersectional failures! read.dukeupress.edu/critical-ai/...

In some ways, gen AI's aggregation of stereotypes recalls the DEI rhetoric that became a prevalent component of promotional culture (Wernick 1991) and branding practices (Asquith and Hearn 2012) within corporate strategic communications.6 These practices were in place when the developers of LLMs strove to “detoxify” these systems and “debias” their outputs. It is possible that these efforts may have leveraged corpuses associated with both DEI programs and the social justice movements they poorly imitate. To be clear, we recognize the substantial difference between social justice discourses and the corporate rhetoric of DEI; but we suggest that when scraped into data sets for fine-tuning or reinforcement learning, such differences are blurred. This hypothesis may help to explain why gen AI outputs commonly echo a language of “pride,” “celebration,” “joy,” and “dignity” without explicit prompting in reference to minoritized communities. No matter whether this language has been deployed as a deliberate debiasing strategy or merely appears in the training data, the autogenerated mimicry of language that was created to foster collective pride and celebration for minoritized social groups has drained the language of the relevant cultural and political contexts. In doing so, it diminishes the critical power of this discourse. In other words, gen AI appropriates the signifiers of social justice and inclusion but does not embed them in an ethos or network of social relations capable of leading toward social justice in action.

Neil Turkewitz @neilturkewitz.bsky.social · last mo.

Google: “enforcement focus should be on outputs.” In plain, non-legal language: We should be able to build whatever we want without accountability. Any enforcement actions should be addressed at our users, not us. It’s in Google’s DNA—they can harm systemically at scale w/o responsibility.

Google says AI training is fair use and copyright should be policed on outputs, not inputs - Music Business Worldwide

The tech giant has published a new policy paper outlining the company’s preferred approach to AI regulation…

musicbusinessworldwide.com

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last mo.

Nice - from over on that other site...with thanks to @neilturkewitz.bsky.social

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last mo.

#CriticalAI4.1, Reminder to check out Molly Crabapple's critique of corporate tech's degradation of creative work by producing "vampiric slop"––AI generated mimicries of human artistry! read.dukeupress.edu/critical-ai/...

While the leaders of AI companies destroy illustrators’ careers and burn through billions in investor cash, they are making fortunes, transferring wealth from artists to Silicon Valley billionaires (Roose 2022). Artists are not the only victims. At the 2023 Perugia Journalism Festival, tech companies sent paid shills to prance around onstage and say that gen AI was the future of journalism (IJF n.d.). Some news writers shared that they were forced to use LLMs while their colleagues were fired. Anyone who questioned the value of replacing human artists and writers with glitchy corporate products was mocked as the equivalent of a horse-and-buggy maker resisting the car. This weak analogy overlooks the fact that automobiles were, for the most part, an improvement in transportation, whereas gen AI's only “improvement” is to the bottom line.

Critical AI https://read.dukeupress.edu/critical-ai/issue @criticalai-journal.bsky.social · last mo.

www.nytimes.com/2026/06/29/u... Surprise, surprise. Kids can easily break through the "safety" features that supposedly make social media apps less addictive and dangerous.

Many Child Safety Features on Social Apps Don’t Work, Report Finds

nytimes.com