Everyone seems to be clamoring for AI literacy these days—how to prompt effectively, how to spot AI-generated content, how to integrate these tools into workflows. I have been critical of this phenomena, see my post on pencil literacy and a new definition of literacy for the AI age.

But there’s been surprisingly little discussion about what generative AI means for visual literacy, our ability to interpret, create, and think critically about visual information. Perhaps this gap exists because visual literacy has always lived somewhat in the margins of educational discourse, overshadowed by more traditional literacies.

This oversight feels significant, especially as AI systems become increasingly sophisticated at both analyzing and generating visual content. So when I was recently invited by my friend Maria Avgerinou to write an article for the Journal of Visual Literacy exploring this topic, it felt like addressing a crucial blind spot (no pun intended) in our current conversations about AI’s impact.

The resulting article, “Brains without minds: Musings on visual literacy and GenAI,” (citation and link below) explores the complex relationship between human visual literacy and Generative Artificial Intelligence. Through experiments with multimodal systems like ChatGPT-4, DALL-E, and Claude, I examined both the remarkable capabilities and fundamental limitations of how these systems process visual information.

Mishra, P. (2025). Brains without minds: Musings on visual literacy and GenAI. Journal of Visual Literacy. https://doi.org/10.1080/1051144X.2025.2538006

In this paper I discuss a fascinating paradox: these AI systems demonstrate sophisticated interpretive abilities, decoding cultural symbolism in street art and extracting precise historical details from photographs, yet struggle with other seemingly simpler tasks. This led me to characterize GenAI as having “brains without minds” and “eyes without hands:” systems that process patterns without truly understanding their referents, lacking the embodied experience that accompanies human visual learning and creation. This, however, doesn’t preclude creative partnership. The article explores human-AI collaboration through my typographical design experiments, pointing toward an expanded conception of visual literacy, a skill increasingly vital for navigating the ethical challenges of AI-generated content in our world of deepfakes and synthetic media.

Coming Full Circle (some historical context)

Writing this piece was a genuine pleasure, but not just because of the research itself. This invitation felt like coming full circle—a chance to revisit foundational interests that trace all the way back to my high school years in India, spending hours at the British Council and USIS libraries, reading books on visual perception and how we make sense of the world. That sparked an interest in art and art history, and led to a burgeoning interest in film, which was fed by weekends at the Shakuntalam theater in Pragati Maidan (sadly now closed) watching the best films the world had to offer. Somewhere along the way I got introduced to Douglas Hofstadter and his work on AI (and yes, I have been interested in AI since the 1980’s) and somewhat tangentially to Scott Kim and ambigrams. I create ambigrams even today and my interest in optical illusions led to creating, what I believe, is an original oscillating impossible figure illusion, and maybe most importantly, to the optical illusion New Year’s videos our family has created over the years.

In short, it led to a lifelong fascination with visual cognition, optical illusions, and the visual arts.

As it happens, when I was a graduate student at the University of Illinois, I got a chance, as part of a graduate seminar, to write a paper, about scientific illustrations and how we make sense of them. I then adapted that class-paper into an article (The role of abstraction in scientific illustration) that was published back in 1999 in the Journal of Visual Literacy. This paper allowed me to bring together my diverse interests: the history and psychology of art and science, optical illusions, visual perception, and scientific education. It was perhaps the first time, professionally speaking, where all these seemingly disparate threads came together into a coherent whole.

Mishra, P. (1999/2004). The role of abstraction in scientific illustration: Implications for pedagogy. Journal of Visual Literacy. 19(2), 139-158. Reprinted in C. Handa (Ed.). Visual rhetoric in a digital world: A critical sourcebook. (pp. 177-194). Boston, MA: Bedford/St. Martin’s Press.

I’m quite proud of of this article, not just because it was cited (especially for work somewhat outside my area of expertise), but because it synthesized research literature across domains that typically don’t interact with each other. The paper was later selected for inclusion in a sourcebook on visual rhetoric, and I remain grateful to JVL for taking a chance on this interdisciplinary work.

For a time, I was actively engaged with the visual literacy community, attending conferences and following the field. But as often happens in academic life, other work took precedence, though my fascination with visual representation never truly waned.

So imagine my delight when I was invited to return to these questions through the lens of today’s generative AI technologies. I’m grateful to see this new work now in print, representing both a return to my roots and an exploration of new frontiers in visual literacy.