The Brooklyn Rail

Public discourse around generative AI is as chaotic as our political present, rife with instability, hearsay, and conflicting information. Will AI ultimately supplant the workforce? What role is it playing in military operations? Is AI already sentient—or will it be? Even our own sense of visual perception can’t be trusted: AI-generated images and memes are deployed deceptively, even nefariously, by our current administration and those well beyond. With each technological advent comes a wave of public and institutional suspicion and so-called “slow adoption.” With AI, however, a slowness to adapt is perilous, a fact that has made people not only afraid, but outright paranoid around the implications of its use—especially knowing that what we see cannot necessarily be believed. Artist Trevor Paglen’s new book, How to See Like a Machine: Images After AI, could be accused of stoking this fire of public paranoia—if what he was writing weren’t true.

Paglen introduces himself on a self-effacing note: Opening the narrative in his mid-2010’s Berlin studio, he recounts his own disbelief at the capabilities of the computer vision software framework he and his team were developing at the time. “I figured it might turn into a Photoshop plugin one day,” he writes. “One that no self-respecting artist would ever use.” He was wrong.

How to See Like a Machine aptly traces a true paradigm shift in visual culture. The methods and models that have served the field so well for so many years (semiotics, for example) fall down in the face of machine vision. The book includes essays written over the course of the past fifteen or so years, a time period that has coincided with Paglen’s rather meteoric rise as an artist (he was named a MacArthur Fellow in 2018 and is a recipient of the 2026 LG Guggenheim Award).

First published in the New Inquiry in 2016, the book’s first essay, “Invisible Images (Your Pictures Are Looking At You),” presents a provocative thesis that haunts the rest: human-centric visual culture, as Paglen identifies it, has given way to invisible, insipid, and capitalist-driven computer vision, whose images and their cultural implications require new forms of negotiation and resistance. Whereas classical notions of visual culture imply relationships between people—as in, people produce images; people consume images—computers produce images that are invisible to people and parsable only by other machines. If this fact feels anxiety-inducing, that’s because it is: Machine-machine vision is used to produce power in the form of hegemonic surveillance systems that govern everything from our everyday shopping habits to our government’s policing practices.

In the essays that follow, Paglen works to show exactly how machine vision works, and how it impacts us not only as individuals, but as a public. He traces a wide range of histories and practices, from neuroscience and medicine to art history and military psyops, all of which have colluded in various ways to enable and inform the technologies that have so greatly impacted our engagement with images. Paglen’s academic background is apparent here, as his work is impeccably researched, so much so that the essays sometimes feel lost in their own details.

Authoritative as it is, that doesn’t mean the book feels apace with the sometimes-glib, lightning-fast discourse that currently surrounds generative AI. In his introduction, Paglen concedes that some of his essays are rife with outdated examples given the timespan during which they were written. While this works against him in a way—How to See Like a Machine is publishing just as AI-engineered images are dominating the narrative around the United States’s current war in Iran—it also proves just how ahead of the discourse Paglen has been.

Indeed, one could read How to See Like a Machine as a kind of companion text to the conversation currently surrounding, for example, the viral video campaign generated by Iranian agency Explosive News that has dominated the war narrative thus far, wherein Lego-themed images of world leaders have accumulated millions of views online. Suddenly, the three-part essay on military psyops that dominates the middle of the book doesn’t seem so paranoid after all.

Who is How to See Like a Machine for? Paglen has joined a cohort of artists who have written around the subject in recent years, with artist Hito Steyerl and her 2025 book Medium Hot: Images in the Age of Heat, which follows on researcher Kate Crawford’s 2021 Atlas of AI, chief amongst them. Collectively, their work feels like an opportunity to slow down and delve more specifically into the impact of generative AI on visual culture. While it may feel like his book is written for this cohort of insiders, all of whom are cited in the book, Paglen works hard to generate a story line that is clear and accessible. Our entire society, after all, must learn how to “see” and negotiate with the machine-generated representations Paglen describes. With How to See Like a Machine, Paglen works to make the invisible not only visible, but legible.