Trevor Paglen, How to See Like A Machine

The very essence of what constitutes an image has undergone a seismic shift, according to artist and author Trevor Paglen. Once a presumed artifact of tangible reality, a photograph now exists in a perpetually malleable state, increasingly divorced from the certainty of its origin. Paglen, whose latest book, "How to See Like A Machine," delves into this evolving landscape of visual perception, argues that the advent of generative artificial intelligence has fundamentally altered our relationship with imagery, moving us from an era of assumed authenticity to one of pervasive doubt.

This transformation, which Paglen terms the "indexical flip," was starkly illustrated to him during the initial air strikes against Iran. While monitoring news feeds, he found himself not discerning misinformation from truth, but rather questioning the very reality of what he was seeing. "Instead of trying to identify the crap and throw it away, I was trying to identify what’s real," Paglen recounted in an interview with Dazed. "I’d gone from assuming things are ‘real’ to assuming things are ‘fake’." This unsettling pivot underscores a core thesis of his book: in a "post-indexical world," all media becomes "superpositional," existing in multiple states of potential reality.

Paglen’s extensive decade-long research and artistic practice converge in "How to See Like A Machine," weaving together seemingly disparate threads such as UFO lore, government disinformation campaigns, the mechanics of magic, algorithmic feedback loops, the discriminatory applications of AI in policing, and his firsthand experiences with early machine vision technologies. His exploration suggests that these diverse phenomena are not isolated curiosities but interconnected elements shaping our contemporary visual environment.

The Erosion of Indexicality: From Record to Activation

Historically, photographs held an implicit promise of representing a specific moment in time and space. Even with known instances of manipulation, such as the notorious removal of purged individuals from Soviet-era photographs or the fabricated images of fairies in early 20th-century Britain, the effort involved in altering an image lent a degree of credibility to the unedited version. However, generative AI has democratized and accelerated image manipulation to an unprecedented degree. As Paglen elaborates, "Everything has become a potential AI hallucination."

This erosion of indexicality—the direct link between an image and the reality it purports to depict—means that images are no longer solely passive representations. Instead, they are increasingly becoming "activations" capable of influencing and altering reality itself. Paglen cites the example of a speed camera photograph, which, while a representation of an event, triggers an automated legal response, thereby enacting a change in the real world. Similarly, AI-generated content, designed to capture and hold attention, can actively shape our cognitive states and behaviors.

A Paradigm Shift: Computer Vision and the Invisible Image

Paglen draws a profound parallel between the invention of photography and linear perspective and the emergence of computer vision and generative AI. He argues that each represents a fundamental redefinition of what an image is and how it functions. "When we look at the invention of photography, or the invention of perspective, or any number of other moments, the fundamental question of what an image is was changed," he stated.

A key distinction with computer vision, Paglen highlights, is the removal of the human viewer. Historically, an image required human perception to be complete. Now, with machine vision systems generating and processing images for other machines, we are entering an era of "fundamentally new kind of image that is natively invisible to humans." This raises critical questions about the nature of perception and understanding when a significant portion of visual information operates beyond our direct comprehension.

Operational Images and the "Poison Coin" Technique

Paglen builds upon the concept of "operational images," a term coined by filmmaker Harun Farocki, to describe images used within control systems—such as those for calculating missile trajectories or enabling robot navigation. He extends this idea to encompass what he calls "activations," which include not only functional images but also those that induce cognitive and neurological responses.

This concept of activation is further illuminated by the "poison coin" technique, a method drawn from stage magic and historically employed by intelligence agencies. This technique involves obscuring a sinister or manipulative element within a seemingly innocuous or everyday facade. Paglen explains, "The idea is that you use the quotidian to hide something sinister." He draws a parallel to the way chatbots, designed to be emotionally engaging and appear intelligent, can mask underlying data harvesting or manipulative functions. The proliferation of AI-generated content and anthropomorphized interfaces, Paglen suggests, exemplifies this technique, creating a disarming layer around technologies that may be collecting vast amounts of personal data. The widespread adoption of smart devices, like robot vacuums that map and catalog home interiors, serves as a contemporary example of this principle.

The Political and Social Implications of a Post-Indexical World

The implications of this shift extend far beyond the realm of art and technology, touching upon the very foundations of democratic discourse. Paglen posits that a shared belief in empiricism and a collective understanding of facts, historically underpinned by photography, are essential for democratic deliberation. The "dissolution of indexicality," he fears, "is a bigger deal politically than we might think," potentially undermining the creation of shared realities necessary for societal consensus.

The rise of AI-generated content, coupled with the sophisticated algorithmic curation of social media feeds, creates an environment where reality itself can be manufactured and tailored to individual consumption. This raises concerns about the potential for widespread manipulation and the erosion of critical thinking, especially when images are designed to bypass rational thought and appeal directly to subconscious responses. Paglen’s observation about the addictive nature of short-form video content, such as people eating spaghetti on TikTok, highlights this phenomenon: "There seems to be something neurological going on, where the image is speaking to your subconscious and activating your cognitive apparatus in such a way that you feel compelled to pay attention to these things, even though the rational part of your brain cannot understand why."

UFOs as a Meme in the Psy-Op Landscape

Paglen’s exploration of "How to See Like A Machine" is deeply intertwined with his prior work investigating secret military installations and the role of disinformation. His early research into places that "don’t exist," such as Area 51, exposed him to a landscape saturated with conspiracy theories and deliberate misinformation campaigns. He argues that UFO narratives, particularly those amplified by government disclosures during the Trump administration, function as powerful "memes" within the broader framework of psychological operations (psy-ops).

"UFOs are essentially a meme that psy-ops folks have used for a very long time," Paglen asserts. He notes that the inherent ambiguity and perceived indexicality of UFO photography, combined with a deep-seated human desire to believe in the extraordinary, make it a potent tool for distraction and narrative manipulation. The proliferation of UFO sightings and imagery, he suggests, is not coincidental with the rise of generative AI, as both phenomena exploit our innate cognitive vulnerabilities and our fascination with the unknown. This, he argues, aligns with the image-making strategies of the Trump administration, which frequently utilized AI-generated imagery and embraced meme-like narratives to shape public perception.

Developing New Analytical Tools for a New Media Epoch

In an era where the very nature of images is being redefined, Paglen stresses the urgent need for new analytical frameworks. "The analytic tools that we have to describe media and images from 50 years ago are describing a different world than exists now," he contends. His book aims to contribute to the development of a new vocabulary and conceptual toolkit to help individuals navigate and understand the contemporary media environment.

While Paglen expresses skepticism about the ability of artists to directly influence the vast infrastructure of AI and data systems, he emphasizes the importance of developing the capacity to "spot the techniques that are being used against you." By understanding the underlying mechanisms and psychological appeals of these new image-making technologies, individuals can begin to reclaim their perceptual autonomy and engage with the visual world with a more critical and discerning eye.

Trevor Paglen’s "How to See Like A Machine" serves as a critical roadmap through the evolving landscape of images. It challenges us to question our assumptions, to recognize the subtle and overt ways in which we are influenced by visual information, and to develop the cognitive resilience needed to navigate a world where reality itself is increasingly mediated, manipulated, and, perhaps, even manufactured by machines.