Friction Is A Civic Virtue
Note: This piece was first published on Medium (June 7, 2026). For more Amor Mundi posts, visit the Medium page (updated weekly) or the official Hannah Arendt Center website, which offers a fuller archive of past writings.
It is comforting to think that AI is merely a tool. We human beings have always used tools to think. Calculators, Computers, and Google help us think faster and more accurately. In Plato’s Phaedrus, Socrates tells the story of Theuth, the Egyptian god who invented writing and offered it to King Thamus as a gift that would improve memory and wisdom. Thamus worried that writing would lead to the end of memory. Relying on writing to remember, we would lose our capacity to remember and become forgetful. It would replace wisdom with the appearance of wisdom and lead us to attend less to the world which now will be available eternally in written form. And yet we know of Thamus’ worries about writing only because Plato wrote it down.
Every technology arrives with this ambiguity; it extends our powers and changes us. The medium, as Marshall McLuhan saw, can become the message. The question is not whether tools change us. They always have. The question is when a medium begins to remake the world in which we think. We need to ask: Can we use AI as a tool without it transforming our world so that it changes who we are?
When Googling became a verb, I worried it would be the end of books. When writing papers, or even when writing books, writers can marshal quotations to prove their points simply by Googling. No need to read Nietzsche or Proust, one can simply Google the citation about Schadenfreude or the madeleine. Writing becomes easier, but the struggle of reading full books is avoided.
AI is an even more powerful tool. We can cure diseases, develop more powerful vaccines, create songs and write seamlessly. When we finish in minutes a polished presentation that used to take hours — and that presentation is actually better than what we would have done ourselves, AI gives us what Pamela Pavliscak calls the “capability glow”: the feeling that we can suddenly do more, think faster, write better, and create more.
But AI’s capability glow has a dark side. There is the feeling of “AI Brain Fry,” which Pavliscak recognizes in coders whose job has changed from writing code to managing artificially intelligent agents. And there is what she calls the “guilt spiral,” when you realize that your bot nailed the difficult email or paper you were writing in seconds — and you worry that maybe “you should have written it yourself.” AI’s unnerving friendliness makes it especially easy to trust.
In a recent essay “On Faith and Freedom of Thought,” the essayist and novelist Marilynne Robinson argues that we have lost our freedom of thought “not because of special interests or conspiracies but because of a diminishment and slacking off that makes us vulnerable to such things.” Her worry is that in the academic world, as in churches and in business as in law, we have sought efficiency over depth. “The consequence,” Robinson writes, is that we “lost rigour.” The laziness and loss of rigour is not born from technology, but it is exacerbated by artificial intelligence.
This past week, I attended the conference “Epistemic Institutions in the Age of AI,” hosted by the Santa Fe Institute. On the way back, I received a text from a friend asking if a video he had received on Instagram was real. It was a video of Hannah Arendt in a college lecture hall leading a class on questions raised by her book Eichmann in Jerusalem. The video offered itself as a 1967 lecture by Arendt, and had been shared by influential public intellectuals. I had never seen a video of such a lecture. What is more, while the words seemed vaguely familiar, I could not recall anywhere that Arendt wrote what she was saying in the lecture. Could it be that there was a previously unknown video of Arendt giving a lecture?
A bit of sleuthing reveals that the video originated from an Instagram and YouTube account that advertises itself as creating fictional, but also “historically and philosophically accurate” AI-generated videos. This particular video was part of a five-part “Masterclass” in which “standing at the chalkboard, Arendt peels back the layers of historical horror to challenge our deepest coping mechanisms.” Whatever that means. The creator of this fake video claims that “Arendt delivers a powerful classroom lecture to break down her most sobering insights and completely upend how we understand morality, complicity, and the true nature of modern evil.”
These descriptions sound, of course, like an AI-generated account of Arendt’s thinking. And the videos as well are not simply a vocalization of Arendt’s words. Arendt speaks words that are paraphrases of various sections of her book stitched together in a seductively accessible way. Instead of reading Arendt, we can watch her teach — or watch a simulation of her teaching. And instead of listening to her words, we get an AI-generated paraphrase. The effort is ease, the opposite of rigor. It is not surprising that behind the account is not a scholar or thinker, but an internet creator and entrepreneur who produces these fake videos about dozens of historical figures.
There is a whole cottage industry of creating fake Hannah Arendt quotations. Two that have received wild popularity on social media are:
“The death of human empathy is one of the earliest and most telling signs of a culture about to fall into barbarism.”
and
“This constant lying is not aimed at making the people believe a lie, but at ensuring that no one believes anything anymore. A people that can no longer distinguish between truth and lies cannot distinguish between right and wrong. And such a people, deprived of the power to think and judge, is, without knowing and willing it, completely subjected to the rule of lies. With such a people, you can do whatever you want.”
Hannah Arendt never wrote either of these quotations. She did, however, say many similar things. Which raises the question: Why would someone create a fake quotation when so many real ones express a similar viewpoint? And, does such an altered quotation matter?
The temptation to simplify Arendt is not new. In Ada Ushpiz’s documentary Vita Activa: The Spirit of Hannah Arendt, many of the quotations projected on screen are silently altered: sentences are reordered, words are changed, and Arendt’s difficult formulations are made easier for a viewing audience. The changes are often not outrageous, although some are. Some may even seem stylistic. But that is precisely why they matter.
Arendt’s words are not incidental to her thought. Her difficulty, her winding sentences, her resistance to slogans are part of what it means to think with her. To smooth her language in the name of accessibility is to risk violating the very spirit of a thinker who insisted on confronting the stubbornness of factual reality.
Arendt herself warned that culture is not preserved when difficult works are rewritten, condensed, digested, or reduced for easy consumption. In her essay “Crisis of Culture,” she argues that cheap editions of Shakespeare do not destroy Shakespeare because the words remain. But when Shakespeare is adapted in condensed or readable editions, when the words themselves are changed, then the object is changed. What circulates is no longer the work, but an image of the work.
The AI Arendt video radicalizes this older problem. It does not merely condense or simplify Arendt’s words; it invents them and puts them in a realistic video spoken by her. It gives us not Arendt, but an Arendt-like performance: fluent, grave, accessible, and false. And because it feels pedagogically effective, because we want to believe that it transmits Arendt’s lesson, it tempts us to excuse the substitution. We allow Arendt to be turned into propaganda, a stand-in for an idea we want to promote rather than a thinker we must struggle to understand.
But that substitution is exactly the danger. Once we accept that made-up words are good enough because they are “philosophically accurate” or serve our purposes, we begin to lose the common world of facts, texts, and accountable speech. The danger in the manipulation of facts is that such manipulation blurs the clear boundary separating fact from lie. As Arendt writes, “the result of a consistent and total substitution of lies for factual truth is not that lies will now be accepted as truth, and the truth be defamed as lies, but that the sense by which we take our bearings in the real world — and the category of truth vs. falsehood is among the mental means to this end — is being destroyed.” The danger, in other words, is a refusal to believe that there is in truth a stable factual world.
The danger of AI deep fakes is not simply that AI produces fake videos. Human beings have always lied, forged, misquoted, manipulated, propagandized, and embellished. The contemporary problem is that AI makes such substitutions seamless. It allows the fake to enter the world with incredible ease and seductiveness. AI fakes are frictionless, they wear the face, voice, and style of the real thing.
The danger is not simply lies and deception, but the disappearance of the boundary between lie and reality. The medium of AI means the medium vanishes into the message and the simulation is received as if it were part of the common world.
In talking about tools, Martin Heidegger makes a helpful distinction. When I’m building a house and I grab a hammer to pound a nail, the hammer is a tool that is absorbed into my work. I don’t attend to the hammer, I just use it. It is handy, ready to be used. But if I miss the nail and hammer my finger, the hammer ceases to be a tool. Suddenly the hammer is not a tool but is an object that is present before me. It is then that I look at the hammer abstracted from its handiness as a tool. I can measure and classify the hammer, treat it as an object present before me, something I can study, know, and control.
This observation of the hammer from the outside is the attitude of modern science. For Heidegger, it flattens the world from one in which we are intimately involved into a world that we study. The scientific view of the world, Heidegger worries, takes us out of the world in which we live, it turns us to observers alienated from the world. We then seek to know and control the world, make it submit to human will. This will to power over things, however, threatens to turn itself against ourselves. We humans also are things, objects to be controlled, studied, improved, and manipulated. This is the danger of the modern technical world order, that we turn even humans into human resources.
Heidegger shows that, ordinarily, we want our tools to be ready-to-hand. We want them to disappear into use. But with AI, such tool-like disappearance is precisely the danger. When AI becomes so handy, so seamless, and so friendly that we stop noticing it, it ceases to be a mere tool. It begins to remake the world in which we think, remember, and judge. AI should not be allowed to vanish into handiness.
The answer to this danger is not to reject AI, as if one could. AI is part of our world. Just as Plato could not reject writing, we must learn to use AI well. To do so, we must learn to resist AI’s aspiration to become invisible. A synthetic Arendt should not glide into the world as if it were an archival fragment. It should announce itself as synthetic every time it appears.
We need laws that require AI-generated creations to be permanently labeled as such. This is not a minor regulatory detail, but a defense of the common world. A label forces a pause. It allows us to stop and think. It tells us that what we are seeing is not evidence, not memory, not fact, but a creation, one put into the world for a reason. It disrupts the handiness of AI in the hope that, in doing so, we allow ourselves to respond to the world that AI is creating for us.
Such pauses matter because attention is one of the most important faculties by which we sustain our human agency and our human reality. At the Santa Fe Institute conference, attention became a recurring theme. Some argued that our response to AI must be to attend to the humanity of others, something that requires an ethic of care. There is much truth in this approach.
But in invoking the need for care in our use of AI, we must be careful not to turn our care for others into an abstraction. Attention is not the same as abstract care. Much of the dream of AI governance is a dream of care at scale: systems that monitor, predict, advise, correct, and optimize human behavior. But care without attention can become a form of domination. It sees people statistically, as populations to be managed, rather than as persons who appear before us, speak in their own voices, and share a world with us.
This is why attention must become a civic virtue. To attend is to resist the smooth substitution of image for person, paraphrase for text, simulation for testimony, aggregate for individual. It is to ask, patiently and sometimes inconveniently: Who is speaking? Are these her words? What world does this claim come from? Who stands behind it?
AI will give us polish, fluency, and the glow of new capability. What we will need to preserve are rough edges: labels, citations, hesitation, eccentricity, difficulty, the stubborn presence of actual words and actual people. In a world of synthetic ease, friction is not a failure of the tool. It is the condition of attention and human judgment.
You might also enjoy my recent essay, “Attentiveness as a Human Act: How the Light Gets In,” and Marilynne Robinson’s interview, “Marilynne Robinson wants America to quit ‘slacking off,’” in FT Magazine.
Learn more about the Hannah Arendt Center for Politics and Humanities at Bard College here.