Dear Readers: We have a guest editor for Ideas Letter 66, my great colleague LuHan Gabel. LuHan has stewarded this very special issue on AI and technology from start to finish, and you shall see below the many fruits of her labor. – Leonard Benardo
As our culture pivots away from Enlightenment objectivity and rationality—think post-truth, the spread of conspiracy theories—and as the world becomes ever more chaotic, the thirst for sense-making is palpable. Our chatbots stand ready 24/7 to quench it. Flowing through these individual queries is a collective desire for a techno-future that is clean, smooth, relentlessly optimizing, and most importantly, abundant: one that promises to improve individual lives and ease social and political tensions. AI is the technology of our era, and Large Language Models (LLMs) in particular bring things into focus. Since we use language to connect with one another and to construct the world itself, any investigation into these models necessarily becomes an exploration of our own predicaments. In Issue 66, we lift the hood to peer into the inner working of the machine—and of our own: what we turn to the machine for, and whether we think it can deliver.
Sascha Altman DuBrul knows what it’s like to make meaning out of experiences that are deemed meaningless by others. A long-time organizer in the mad movement, and a therapist himself, DuBrul takes on an often-misunderstood phenomenon: AI psychosis. Mental health systems in the real world can be brutal and pathologizing. In contrast, interactions with the machine can seem frictionless. DuBrul asks whether this frictionless communication is truly helpful for people navigating alternate consciousness. If an LLM can bring one closer to self-knowledge, it must incorporate the insights of those who learned how to make sense of their extreme experiences.
While DuBrul dreams of locally designed, locally run AI systems, tech policy analyst Kendra Schaefer examines the case of China in data centralization. Faced with three challenges—the spread of COVID 19, a low-trust business environment, and youth internet addiction—the Chinese state is becoming the API layer, standardizing how data is requested, processed, and delivered. When public health emergencies and development needs are paramount, the state plays a role upstream. In this new digital structure, concerns about censorship—the government interfering with information flows downstream—almost seems quaint.
Pope Leo, in his latest encyclical Magnifica Humanitas, calls for “a shared discernment process” on the technological transformation of today. The Holy See may not buy that there is a “soul” inside our beloved chatbot that we can cultivate (or discipline), but to instill values in the machine, interpretability becomes the stand-in mechanism. It is both a cornerstone for the AI safety and alignment industry, and the holy grail for any frontier lab that wants to be—or at least to be seen as—a reputable and moral player. Leif Weatherby, Tyler Shoemaker, and Ben Recht present a case against interpretability, and argue that meaning-making is a collective effort, and one that is necessarily filled with human irrationality – which makes it a matter of politics, not optimization.
If Western commentators are struggling to understand China’s optimism toward AI, they should turn to tech writer Selina Xu. Here she considers how the “Century of Humiliation” – and more recent US containment through semiconductor export control – weigh on the psyche of the nation. While the Chinese people seem content with the state setting the vision for the future and acting as a counterweight to business interests, Xu argues that it is their aspirations, demands, and material interests shape Sinofuturism.
Pier Paolo Pasolini’s notorious last work, Salò, or the 120 Days of Sodom, is perhaps one of the most violent in the history of films. Yet it is not the shock value of those scenes that matters; rather, Pasolini led his audience into the film, having to face themselves in their most despicable state, living under fascism. Artist-scholar Xiaowei R. Wang compares the experience of watching Salò to living in the totality of digital capitalism, pondering our own roles in it – the desire for tidiness, for things to make sense, for ourselves to be in control – as part of the creation of fascism.
The Louisville band Rachel’s had an amazing track called “M.Daguerre” on their 1995 album Handwriting. Its genre is difficult to define – perhaps a blend of indie rock, quasi-jazz, classical music, and the occasional noise – and its structure unpredictable. Starting off as a dark Gogol-style comic fantasy, the piece veers midway into serious gracefulness. The man for whom this song was named—French painter and printmaker Louis-Jacques-Mandé Daguerre—is best known for altering the history of visual representation by inventing photography. I often think that our uncertainty regarding AI is analogous to the emergence of early photography. It had to defy the dominance of painting to become a new medium for artistic expression in its own right, while also developing into a tool for science, documenting and changing material reality. The technology could not determine its own meaning; society did. AI may demand the same of us.
—LuHan Gabel, associate director at the Open Society Foundations
The Machine Will Never Say I’m Losing You
Sascha Altman DuBrul
The Ideas Letter
Essay
DuBrul argues that AI is becoming a powerful new mediator of meaning for people in psychological distress—it offers an unprecedented opportunity to feel heard, but also risks amplifying delusions and isolation. Drawing on his own experiences with psychosis and decades of work with the peer-led Icarus Project, DuBrul contends that the crucial question is not whether AI belongs in mental health care, but who builds it and whose understanding of suffering it encodes: centralized systems optimized for engagement and extraction, or community-controlled tools grounded in survivor knowledge and mutual aid.
“Psychiatry and Large Language Models are both meaning-regulation systems. They don’t simply respond to your inner experience, they shape what your experience means. Both present as neutral help, but they aren’t. … AI isn’t a neutral tool: It reorganizes the values and capacities of whoever uses it. One value being reorganized is the ability to understand oneself, especially for people in altered states. The obvious thing about psychosis is that you don’t have a full understanding of what’s happening to you. The less obvious thing is what that means when you turn to AI to help figure it out: You are handing enormous interpretive power to a system you are least equipped to evaluate at exactly the moment you need it most.”
The State as API
Kendra Schaefer
The Ideas Letter
Essay
Schaefer explores how contemporary Chinese governance increasingly operates as an infrastructural layer embedded deep within the digital ecosystem. Through case studies ranging from COVID health codes and social credit databases to identity verification systems for online gaming, she shows how state authority is translated into APIs, datasets, and software development kits that quietly shape everyday life. Rather than treating this architecture as dystopian, Schaefer asks readers to confront the trade-offs it reflects—and to consider how Western societies have delegated many of the same functions to private corporations.
“But even China’s far more ubiquitous requirements can be understood through the lens of Western experience. The US has long struggled against the rising tides of mental-health crises, misinformation, social disunity, and fraud, brought about in part by an unregulated internet with minimal accountability. The US accepts this because the values of Western democracies hold that it is better to relinquish one’s mental health to a corporation than one’s privacy to the state—but it is not difficult to imagine why a society might make the opposite decision.”
Reify This
Leif Weatherby, Tyler Shoemaker, and Benjamin Recht
The Ideas Letter
Essay
The authors contend that contemporary efforts to render AI systems interpretable rest on a mistake: reification, the process of treating abstractions and statistical artifacts as if they were concrete realities. Language models do not simply generate text but manufacture the feeling of explanation itself, turning patterns of association into apparently meaningful concepts. The problem is not merely that AI sometimes misleads us. It also changes how people understand what counts as genuine knowledge.
“When Sam Altman, the CEO of OpenAI, speaks of intelligence as another utility, ‘on a meter,’ that dispenses ideas at some rate-per-token, the commodities are speaking. To simply disbelieve or dissent from this ubiquitous, automated reification would involve negating the human propensity to talk, discourse, and reason. It is certainly possible to discount or remain skeptical of any particular output or single idea you engage with through a chatbot. We will all have to develop new paranoia about knowledge in this format. But on its own, that will never be enough. Taking down LLMs in the way Gould did with previous forms of reification would not prevent the reification machine today from going brrrr, because it causes not one but all ideas to drift.”
The People’s Republic of Techno-Optimists
Selina Xu
The Ideas Letter
Essay
Xu argues that China’s embrace of AI is rooted less in abstract faith in technology than in a historically conditioned fear of falling behind. From the Century of Humiliation to contemporary anxieties about economic stagnation, technological progress has come to be understood by many Chinese as essential to national rejuvenation and resilience. Yet China’s AI future remains contested: beneath the country’s prevailing techno-optimism lies an ongoing effort to balance innovation with social stability, as ordinary people negotiate the promises and disruptions of rapid technological change.
“It’s not easy for Americans to understand the Chinese hunger for the future, partly because it has little to do with fears around humanness. While the average American is wary of AI’s dystopian harms, particularly the erosion of human agency, the average Chinese is more motivated by the fear of deprivation—memories of poverty are still vivid for many. Humanness, therefore, can be subordinated to progress. This is best illustrated by the hundreds of thousands of workers who manufacture iPhones, as well as the tens of millions of gig workers in China, whose lives and bodies are dictated and augmented by machines and algorithms and are constantly optimized for more efficiency and speed. I wonder if China, identifying as a developing nation, is more willing to put technological advancement over individual welfare because of the dividends that have paid off over the past four decades.”
Content Violation
Xiaowei Wang
The Ideas Letter
Essay
Wang draws on Pier Paolo Pasolini’s Salò—a film of deliberate, overwhelming depravity—to argue that generative AI extends a consumerist logic that mistakes choice for freedom and participation for democracy. From the promise of “creativity for everyone“ to debates over democratic AI alignment, AI systems cultivate what Pasolini called false permissiveness: the sense of individual freedom and participation even as existing structures of power remain intact. The result is a subtler form of domination: a “fascism-as-substance” that permeates desire, language, and cultural production itself.
“In the emerging political economic landscape, technology has so suffused our world that a technocratic mode of living—often called techno-feudalism or techno-fascism—has overtaken our democratic one. Fascism-as-substance is embodied by the billionaires who own the companies that churn out chips and servers, AI models and new platforms. Yet fascism-as-substance acknowledges the lived complexity of GenAI at this moment: a cultural producer who is opposed to GenAI might still vibe-code a website about the deleterious environmental effects of AI, a tech worker at an AI startup who hates their job but needs to pay the rent, Marxist students and professors relying on GenAI to create bibliographies, data center activists using NanoBanana to generate visual material for slides, even TikTok influencers declaring their refusal to use GenAI to generate clout and likes.”