No, AI bros didn’t invent mathematical art.

AI-generated “art” fanatics have a new claim to support the supremacy of their format: that mathematics is only now being introduced into artistic practices. That they have essentially invented math-based artwork and pioneered it in a way that has made human creativity obsolete.  

To quote:  

“I think l've got it (maybe) why art-Karens are mad about Al. 

See, all this time they used to say their art is something they found in their soul-an unspoken, unspeakable urge (it's just chemical reactions btw) that found expression. But now it turns out art can be made with math: linear algebra, differential calculus, statistics, and probability. This takes away the mystical quality of art, the one thing that they thought couldn't be quantified, the je ne sais quoi. 

And it makes them mad.” 

This idea should not be given currency. The truth is that mathematics and art have been deeply intertwined for centuries. Long before computational algorithms were “creating” images, artists and mathematicians alike were fascinated by geometric forms, patterns, and proportions.  

In the Renaissance, painters like Piero della Francesca studied perspective using mathematical principles. Francesca even wrote treatises on the subject, such as De Prospectiva Pingendi, exploring how mathematics could aid visual realism. His studies on perspective, light, and geometry advanced artistic techniques for capturing 3D spaces on flat canvases. 

Centuries later, new movements like Cubism intentionally played with geometric shapes and multiple perspectives in their paintings. Pablo Picasso, for example, incorporated mathematical ideas like four-dimensional space into his groundbreaking works. This wasn’t mathematics being accidentally employed - Picasso deliberately used mathematical concepts to convey new forms of visual space beyond the traditional Renaissance perspective. Islamic art has long featured intricate tessellations and symmetries, showing how mathematics can generate aesthetically pleasing decorations. 

In the 20th century, mathematical structures became even more explicitly referenced in art. Op art pioneers like Victor Vasarely created vivid illusions using carefully plotted lines and curves. His 1965 Zebra painting relies on meticulous permutations of angles to produce its eye-popping zebra stripes. Here, sensitively tuned mathematical variations drive the entire artistic effect.  

Developments like fractals, chaos theory, and computational algorithms have also inspired artists for decades before AI. Digital artists coded algorithms to mimic nonlinear systems and generate abstract organic patterns long before Dall-E and Midjourney came along. Even Jackson Pollock’s iconic drip paintings have been found to contain fractal patterns underlying their chaotic splatters, reflecting the way nature builds complexity from iterative mathematical rules. 

So, while AI may be a new development – and perhaps a new frontier - for mathematically-generated art, the creative use of numbers, shapes, and data in art stretches back hundreds of years. The algorithms may be making waves, but Picasso and Pollock paved their way. Mathematics has stimulated human creativity for centuries—so don’t be misled into thinking it all started with AI. 

There is something profoundly different and unsettling about the new wave of AI art systems (and their fanatical proponents) rejecting the longer historical trajectory of mathematical art. Works like Midjourney's pastiches and Dall-E's surrealist mashups point towards a future that may end up looking more homogenized, prompted variations be damned. 

My concern is that these systems lean overwhelmingly on existing artistic corpora to build their models. By scraping millions of images from the Internet, they learn textures, shapes, styles and combinations that already exist. They then remix these in novel ways according to their training objectives. But the raw materials are still just recombining what humans have already created previously. 

Will AI ever advance to the point of pioneering wholly new styles and artistic techniques the way visionaries like Picasso, Pollock, and Vasarely did in their eras? Or will the output of the art machines hit an upper limit on creativity by perpetually neo-conceptually recycling the existing paradigm shifts that now educate their models? 

Mathematics may underlie both historical and contemporary generative art, but we have yet to see if AI can spark increments as radical as perspective drawing, abstract expressionism, or op art. For now, AI art’s appeal lies in its ease of output and its uncanny variability more than groundbreaking creativity. While mathematical art has long pushed conceptual boundaries, algorithms remain within the boundaries and limitations of their training distributions. 

This difference also translates into how much humans control the artistic process. With everything from Islamic tiling to Renaissance anatomies, the artist carefully constructed each effect around their intent rather than just prompting a black box system. But with AI systems now running autonomously 24/7, once they are released, human guidance over outcomes becomes minimal, no matter what mathematical architecture is involved. 

The issue is not whether mathematics enables art - that connection is old news. What's at stake is whether AI can help take human artistic innovation and freedom to new levels or whether it risks confining progress by over-recycling our existing image banks. Only time will tell if algorithms open doors beyond where non-computational approaches have already taken art over history’s long trajectory. For now, the jury is still out – but mathematics alone does not automatically imply artistic progress. Unlike the perpetual reinvention of painting, poetry, or music over centuries, algorithms remix variations on human prompts rather than pioneering new stylistic grammar.  

The arrogance of AI art absolutists is distasteful, to say the least. Mocking “art Karens” who critique their systems through a humanist lens shows a flippant attitude towards the nature of their treasured generative algorithms. Dismissing artists as mystics or romantics just because mathematical constructs can produce compelling images ignores centuries of visionaries who deftly integrated numerical ideas into groundbreaking innovation. Unlike AI, pioneers like these understood cultural context and actively advanced artistic traditions rather than just probabilistically recycling them. AI currently creates captivating but often contextless pastiches precisely because it cannot replicate the layers of meaning evolved by sentient artists over millennia.  

And if the sentient prompt-pasters at the helm of the AI “art” revolution cannot understand that meaning and context either, their movement is doomed to failure.

Subscribe to Joan's Index

I publish a weekly email, featuring one tech essay, + 3 curations, recommendations and ideas. Always free, never ad-supported.

Email Address

Sign Up

We respect your privacy.

Thank you!

@Westenberg logo
Subscribe to @Westenberg and never miss a post.