In the realm of generative AI, the discourse surrounding its cultural implications is vast and varied. Some envision a dystopian "dead internet" where bots dominate content creation, while others hope for a brighter future marked by universal basic income that allows for greater creative freedom. However, artist Trevor Paglen has been examining the effects of generative AI on culture for over a decade, long before large language models became a household term. His latest work, How to See Like a Machine: Images After AI, encapsulates his insights, arguing that our understanding of images is trapped in an outdated framework.
Traditionally, the interpretation of images has been rooted in a semiotic and human-centric perspective, viewing them as mere representations or metaphors. In contrast, Paglen introduces a new operational paradigm where images are created by machines for machines, aiming to shape reality instead of merely reflecting it. He emphasizes the importance of asking not just "What does this image say?" but rather "What does this image do?"
Paglen acknowledges that this line of inquiry is not novel; media theorists and artists have explored similar themes over the past fifty years. However, he identifies two significant shifts in our visual culture: the advent of computer vision and the rise of generative AI. The former transforms visual data into mathematical abstractions, while the latter manipulates our relationship with reality itself.
He coins the term "machine realism" to describe how machine-readable images can perform functions far beyond traditional imagery. For instance, surveillance systems in grocery stores and commercial trucks utilize machine vision to monitor behavior, thus normalizing surveillance in our daily lives. This evolution signifies a shift towards a reality where images are not just seen but actively engage with us, adapting based on our interactions.
Paglen's book synthesizes his earlier articles and presentations, including his project ImageNet Roulette, which highlighted biases in facial recognition technology. He discusses the implications of cognitive warfare, where technology is used to manipulate human perception, tracing its roots back to historical covert operations. In this context, he posits that today's psyops are automated and scalable, often going unnoticed in our interconnected culture.
In the concluding chapter, "Archives of the Future," Paglen argues that post-AI, images hold a quasi-indexical status akin to UFO photographs, where their truthfulness is ambiguous. This reflects a broader shift in our media landscape, moving from linear to networked and now to AI-driven paradigms, prompting a need for enhanced media literacy.
Ultimately, How to See Like a Machine serves as a crucial exploration of our current media environment. Paglen's insights encourage us to recognize the evolving role of images in our lives and the potential for human agency amidst these technological transformations. As we navigate this new terrain, the ability to critically engage with the images we encounter will be essential for shaping a future where technology serves humanity rather than the other way around.