And Woman Created Picasso!

Finally a woman turns the tables on the old Spanish cubist. Plus a detailed look at Stable Diffusion from an art history perspective.

And Woman Created Picasso, by Michael Howard and Stable Diffusion

This is my first attempt to delve in depth into the world of AI art, to create some works, to apply principles of art criticism, to consider what this will mean for working artists, and to discuss some of the problems in the AI art community, including the objectification of women, and the increasing corporatization leading to censorship. Let’s get started!

NOTE: If you don’t want to hear about Picasso, Griselda Pollack, Doctor Who, or Gadamer’s hermeneutical aesthetics, but only about AI art (with examples), you can scroll down to PART 2 – EXPLAINING THE INEXPLICABLE. If you just want to view around a dozen images I created using Stable Diffusion, you can scroll all the way down to the GALLERY (slideshow) at the end.

PART 1 – ART HISTORY AND AESTHETICS

In my series “Was Picasso Spiritual?” (Part 1, Part 2, Part 3) I tried to get at what’s spiritual in the work of Picasso. In Part 2, I discussed his many portraits of women — those from his neoclassical period which appeal to our traditional sense of beauty, those from his Weeping Woman series which depict psychological torment, and many others in between. I sought a balance between admiring his work and recognizing that he was an inheritor of a system in which male artists used the female body as raw material for their visions of what art could be.

Picasso never seemed to question the setup whereby a woman’s function was to pose for men. Art historian and feminist Griselda Pollock addresses this issue in the 1989 TV miniseries Art of the Western World, Episode 14. She shows us the work of women artists like Mary Cassatt and Suzanne Valadon, and highlights the disparity between how women see themselves, and how they are portrayed by men:

She ends the segment with this potent quote:

From Manet to Picasso, from “Olympia” to the “Demoiselles d’Avignon,” male avant-garde artists have staked their claims as ambitious modernists on the bodies of women. The major paintings of European modernism are surprisingly often paintings of the female nude. The power men enjoy makes women available for artistic experimentation and colonization. In many ways, modern art rejected the Humanist traditions of Western art, but it never abandoned the female nude. Instead, it used women’s bodies for its most extreme innovations and fantasies. On their bodies, the implication of distanced creation is there for all to see.

Pollack’s phrase “distanced creation” is one we will return to. But does her critique imply that a huge proportion of art created in the twentieth century should be thrown out on grounds of political incorrectness? I faced a similar question when dissecting the Whovian universe. In “Survival, Friday The 13th, Doctor Who, and Black Cats,” I discussed the plight of female Companions:

At least, having complex female characters who drive the series is progress for Doctor Who. Things weren’t always that way. Back in the 1970s women wanted challenging roles, and though actresses auditioning for Companion were often promised same, their roles frequently degenerated into screaming and being rescued by the Doctor. This was even true of Sophie Aldred’s immediate predecessor, Bonnie Langford, who played Melanie Bush. Though Andrew Cartmel originally had high hopes for bringing out Mel’s complex side (she was supposed to be a computer programmer), that complexity never emerged and Mel became instead the paradigmatic shrill screamer, able to shatter glass at thirty paces.

The character of Ace marked a genuine turning point for the show, and the final (26th) season was all about seeing Ace develop emotionally. “Survival” was written by Rona Munro, whose interest in feminist theory was not so heavy-handed as to spoil the story as entertainment. The strong female Companion is one good feature of “late period classic” that survived the regeneration.

A trollish commenter wrote (in part):

More uninformed, ignorant nonsense about the “sexism” in classic Who. Have you ever actually seen any, or are you just parroting the usual bollocks? … Never mind, go back to your usual brainwashed feminazi learned-by-rote gobshite.

I responded (in part):

I’ve seen a lot of Doctor Who classic and new series, and like them both. I admit I find some episodes from the new series more emotionally moving because Companions like Amy Pond have a back story that makes them more compelling characters. But I also appreciate that Liz Sladen from the classic period took a limited role and invested it with her own energy, originality, and chemistry. The great thing about Doctor Who with its rich history is that everyone can have their favourite Doctor, favourite Companion, favourite period, and favourite story.

While most Companions seem happy with the time they spent on Doctor Who, many did leave because the role was too confining. I’ve seen [the film] An Adventure in Space and Time, which makes the point that when Verity Lambert launched Doctor Who in 1963, the Beeb was still very much a boy’s club.

Things were different 40 or 50 years ago, so we can appreciate classic Doctor Who in its own right, without subjecting it to a yardstick of political correctness. But it does no harm to point out the ways it has evolved. The Doctor will always be the lead character, but the stories are stronger when the Companion has a carefully written back story and a greater degree of emotional complexity. If believing that makes me a feminazi, then somebody send me a t-shirt.

(Of course, things have continued to evolve, and in 2017 we got the first female Doctor in the person of Jodie Whittaker.)

My answer, then, to the enduring question of whether we should throw out a huge proportion of art (or pop culture) which reflects the sexism, racism, or colonialism of its time, is “no.” We should try to immerse ourselves in it, experience it, enter into dialogue with it, and be transformed by it. This will lead to a deep understanding in which we are also conscious of the ways that art of another period or culture is based on different assumptions than those we personally embrace, or those currently in vogue. The same principle can also be applied to AI art.

In taking this view I am influenced by Gadamer’s hermeneutical aesthetics, which I tried to explore in a (hopefully) fun way in “Art and Hermeneutics” (Part 1, Part 2). In Part 1, I wrote:

Hermeneutics, simply defined, is “the art and discipline of interpretation.” In art criticism, hermeneutics is not so much a single theory as a way of approaching art. This approach stresses entering into dialogue, striving to understand a work rather than standing coldly aloof from it and making iconoclastic pronouncements.
…
The art we seek to understand may be from another time or have different cultural roots, so in entering into friendly dialogue with it, we may discover the limits of our own knowledge. Hermeneutics is concerned with how we know what we think we know and what cultural assumptions we bring to the table. The dialogue between a spectator and a work of art may occur over a great historical and cultural distance. We can try to see a cave painting through the eyes of its creator, or we can view it through the lens of modernity; or we can look at it both ways — moving backwards and forwards in time to have a more fulfilling and illumining experience (no TARDIS required).

These different views from different cultural and historical perspectives are sometimes called “horizons.” When we delve deep into a work of art, seeing it from different perspectives (both ours and other people’s), the net resulting view is sometimes called a “fusion of horizons.”
…
One might say that hermeneutics has two different but complementary functions: One is to help ensure that people’s interpretations of art are not merely whimsical, anecdotal, or based on personal or cultural bias. This a limiting function. The other is to foster a depthful connection with art based on dialogue, ideally leading to a fusion of horizons which comprises understanding. This is an expansive function. Still, hermeneutics is not a science; Gadamer said in a 1978 lecture that it’s a gift, like rhetoric, and that one of its components is empathy.

Another feature of Gadamer’s hermeneutical aesthetics is the idea that both artist and spectator are involved in a form of play which brings people together in the manner of a festival. To understand a work of art is not to come away with a crib sheet summarizing its salient points, but rather to lose oneself in it (along with other spectators, perhaps from different times) and to be transformed by it. But this doesn’t signify an end to the game, since further revelations are always possible.

Though Gadamer is heady stuff, we should not forget that sense of play, dancing across the ages. Playfulness and openness are, in many ways, the opposite of political correctness.

PART 2 – EXPLAINING THE INEXPLICABLE

The earlier quote from Griselda Pollack ends with an intriguing phrase: “distanced creation.” Mary Cassatt has empathy for her subjects; Paul Gaughin does not. Even Whistler, whose most famous painting we quaintly retitle “Whistler’s Mother,” actually named it “Arrangement in Grey and Black No. 1.” This implies that, like many artists to follow, he was primarily concerned with the formal qualities of the painting — less so with its subject.

Arrangement in Grey and Black No. 1, by James Abbott McNeill Whistler

“Distanced creation” might also describe the new AI art which became a popular craze in the latter half of 2022. The image below takes advantage of this new AI technology, a subject of much controversy among artists and non-artists alike:

And Woman Created Picasso, by Michael Howard and Stable Diffusion

Who created this work? One answer is that it is a “co-creation” between human and machine.

Does it have expressive value? My (not unbiased) view is that it does.

What does it mean? As the poets say, it doesn’t mean, but simply is.

The plane where artistic creation takes place can be a very high plane, as can be the plane where we respond to art. On that high plane, it’s possible that we perceive seeming opposites as being in harmony, and that we locate more than one meaning in a work.

When we try to explain a work of art, we have to come down from that high plane which comprises depthful understanding. We can easily get bogged down in dualistic concepts.

Nevertheless, if we want to interpret this unusual image, two very different interpretations come to mind. If we look at it from a feminist political perspective, we can say that it shows a strong, powerful woman who evinces tremendous determination and confidence in her own creative power. She is literally creating Picasso from her own upraised hand (and she has the upper hand). She appears youthful and strong, while Picasso looks tired and sad. Of course, all male artists have been created by women in the sense that they came from a woman’s womb. But here the act of creation is direct and immediate. This is a woman wresting control of art back from Picasso and showing her dominance.

But wait! Are the woman and Picasso the same kinds of creatures? Picasso was a “natural person” (a living, breathing human being). The woman is not a natural person, but rather a synthetic entity created by artificial intelligence. If we take that tack, looking at the image from the point of view of a paranoid science fiction plot, it depicts the AI entity forcefully announcing its dominance, claiming superiority over a celebrated, totemic human artist who was once regarded as the pinnacle of creativity in the preceding era of art, a half-century ago.

The image may be empowering when viewed from a feminist political perspective, or frightening when viewed from a humans vs. machines perspective. Some working artists view the new AI technology as a thief and a job-killer. They could be right.

Winners and Losers

The easy availability of advanced AI tools to penniless artistic types like me, and to the masses in general, is often spoken of in terms of “democratization.” We tend to assume that democratization is a good thing. Would we feel the same way about the democratization of brain surgery? The comparison is unfair, since an untrained person performing brain surgery could easily kill the patient, while an untrained person making pushbutton art could probably do no worse than create something in bad taste. (Perhaps some will argue that I have already done so.)

Pushbutton art as a mass phenomenon creates an art glut (as we can see from the home page of Lexica), and a general dumbing down of artistic standards. But in the hands of someone who has studied art history, AI may be a useful tool — if not for final results, then at least for prototyping and experimentation.

The phenomenon of computers making art which is good enough to fool us — occasionally even surprise or move us — is a revolution of sorts. Like most revolutions, it involves a transfer of power: from professional artists to non-artists or would-be artists, and to the companies which harness this new technology and package it commercially. This is likely to include Google, Microsoft, and Adobe. Perhaps even the much-hated Getty Images will jump on the bandwagon, especially if their current lawsuit against Stable Diffusion fails to yield much joy.

The transfer of power means that AI will be used by some people who have little respect for art and artists, little knowledge of art history, and are instant gratification junkies looking for a fix. Ill use of new technologies is more or less a given, and may for a time overshadow responsible use by more knowledgeable users.

As in any transfer of power where there are winners and losers, battle lines will be drawn. I happen to be one of the winners. I’m not a professional artist, but an amateur digital artist who has always played with tools like Photoshop, Paint Alchemy (remember that?), Metacreations Painter, and Mediachance Dynamic Auto Painter. Stable Diffusion is, in one sense, a natural progression towards better and better approximations of what real (natural media) artists do — the kind of tool digital artists always dreamed of having. But when AI art suddenly gets a whole lot better almost overnight (in part by scraping the works of real artists), this can pose an economic and existential crisis.

Since I don’t make my living as an artist, I won’t be thrown out of work if people use AI to make book covers, posters, corporate flyers and t-shirts, rather than hiring a pro. I sympathize with artists who feel justifiably threatened by this new technology, and by the culture surrounding it, which is a sort of “tech head” culture that may be lacking in warmth and camaraderie, and rubs some artists the wrong way.

The old Honeymooners TV show had a classic routine about the “Chef of the Future.” What will the Art Director of the Future look like? Will he or she know how to draw, paint, or sculpt, and have a strong background in art history? Or will the Art Director of the Future be a priest in the Temple of AI, keeper of the magic incantations which prompt the computer to produce fast, cheap results that satisfy the client?

Will there even be an Art Director? Or will the task of supplying occasional bits of corporate art be farmed out to a low-level employee with very little actual training or experience? At the smaller companies, perhaps so.

My Response to the AI Aesthetic

The art one sees on sites like Lexica is often squeaky clean, technically excellent, but lacking in human vision or emotion. I sometimes want to shake the artists by the collar and say: “Yes, but what do you believe?” I fear the answer would be: “We don’t believe in anything. We’re just making eye candy.” Very good futuristic eye candy, but eye candy nonetheless.

My own experiments with Stable Diffusion have led me in the opposite direction. I like making images based on real people, images with grain, texture, visible brush strokes, and images which evoke some past epoch:

Gertrude Stein with Sousaphone No. 1, by Michael Howard and Stable Diffusion

Cave Painting No. 1, by Michael Howard and Stable Diffusion

Tamsin Greig photographed in the style of Alfred Stieglitz, by Michael Howard and Stable Diffusion (fan art)

Since I’ve owned a cheap print of Picasso’s “The Old Guitarist” for about fifty years and have written about that work, I like asking: What if the old guitarist were a woman, and she was painted in the magnified style of Geogia O’Keeffe?

Old Guitarist in the style of Geogia O’Keeffe, by Michael Howard and Stable Diffusion

Corporate Influence on AI Art — The “Safe Art” Conundrum

Given proper noodling and cajoling, the AI can be persuaded to demonstrate its “talent” for caricature:

Donald Trump in the style of Charles Addams, by Michael Howard and Stable Diffusion

At FirstAmendmentMuseum.org we can trace the venerable history of political cartoons, which are a cherished form of free speech. But the increasing role of venture capitalists in the AI art scene translates into pressure not to offend. Knowing that AI art will face some legal challenges, they want to keep the politicians sweet.

After an initial period of relative freedom, the corporate suits are taking over the means of production, banning users from generating art which targets politicians and celebs — likewise banning any form of nudity, no matter how classical and tasteful. Not that nudity in art should be thusly confined. Some feminist art uses nudity to express sadness, anger, or outrage.

Outraged, by Michael Howard and Stable Diffusion — an image you’re unlikely to create with the Microsoft Bing artbot.

On the politicial front, journalist and Twitter user Eliot Higgins posted a series of AI images showing Donald Trump being arrested, and later appearing in an orange prison jumpsuit. It wasn’t a “deep fake.” He clearly labeled the Tweets so no reasonable person could possibly think the images were real. But he got banned from the Midjourney AI, apparently because AI art is not supposed to be controversial or express strong political views. (Sheesh!)

Eliot Higgins – Trump Midjourney Image 1

Eliot Higgins – Trump Midjourney Image 2

On social media, the going snark was that no one could take the latter image to be real, since it depicts Donald Trump reading a book. (The ex-president is not known as a big reader.)

Given the corporate crackdown, and the increasing number of “pay-to-play” sites, many people are installing the free and open source(ish) Stable Diffusion on their home computers. This tends to be more of a guy thing, since the most readibly compatible systems are so-called “gaming rigs” with high-end nVidia graphics cards. However, easy 1-click installation is possible even on some systems which lack a dedicated GPU. You’ll just have to accept a smaller feature set, and wait longer for images to cook.

Obviously, one benefit of having your own personal install of Stable Diffusion is that there’s no corporate behemoth sitting between you and the AI, dictating that hoodlum politicians are off limits, and all nudity is bad. The Stable Diffusion licence does have usage restrictions, but in the real world, these restrictions seem unlikely to be enforced in the draconian, anti-art, Big Brotherish manner of Midjourney or Microsoft. Still, there are many unanswered questions about AI art, leading to a legal and ethical tangle.

The Contortionist, by Michael Howard and Stable Diffusion

Energy Fields

In Seymour: An Introduction, J. D. Salinger writes that “A fat-faced peony will not show itself to anyone but Issa — not to Buson, not to Shiki, not even to Basho.” This raises a delicate, largely unexplored question about AI art. Does the individual human person writing the prompts, pressing the buttons, and imploring the AI to create something meaningful have some influence on the output? During one long, feverish session, I wrangled this image out of Stable Diffusion:

Two Nuns, by Michael Howard and Stable Diffusion

Would anyone stand an equal chance of mining or curating this image? Or does one’s personal energy field play some role in the outcome? I have no clear answers, but it’s similar to questions arising from various forms of divination, such as Tarot cards or yarrow stalks. Certain seemingly random processes are affected or “stacked” by the influence of the human subject’s energy field. (At least, that is one theory.) I have no idea if this applies to AI art, but it would explain why my AI art doesn’t look quite like anyone else’s (if that’s not just a vain belief on my part).

More on Distanced Creation

I’m not through extracting every last drop of meaning from Griselda Pollock’s striking phrase. There are a number of problems in the AI art community, which tends to be somewhat male-oriented, tends to be dominated by true believers in the wonderfulness of AI art, and tends to recapitulate the scenario whereby male artists use the female form as an object of fantasy and experimentation.

Such experimentation is “distanced,” lacking in empathy. Also, the manner in which the technology itself has evolved means that by its very nature it is “distanced” from the human subjects and human artists whose images were scraped wholesale from the Internet in order to “train” the AI.

I am neither an unalloyed supporter nor unalloyed detractor of AI art. I simply note that we do seem to have reached an inflection point. I suspect there are many people like me who enjoy using this new technology, but are also conscious of the ways it can be abused, and the ways in which it can’t help but have unintended consequences. As Marshall McLuhan explained clearly some decades ago, new technologies change the patterns of human association irrevocably, regadless of how we feel about them. The old saw that it’s how we use them which determines the outcome is far less true than we might imagine. See “Marshall McLuhan Explains Why We’re Blind to How Technology Changes Us” on OpenCulture. Maybe also this.

I learned about AI art by reading, and watching YouTube tutorials. One thing I noticed right away about the tutes is that 95% of them are made by men, and while their technical mastery is often impressive, their taste in art is often questionable. At times, the AI art scene resembles Objectification of Women 2.0, with computer nerds using AI to build their perfect woman, conceived of as possessing super-smooooth skin, superhero contours, and eternal youth. This pushed me in the opposite direction: My “Gertrude Stein with Sousaphone No. 1” (above) is the antithesis of the typical AI art aesthetic. My “Saints and Cats” series often depicts elderly women in religious garb gently cradling their cherished pets:

Saints and Cats Series No. 1, by Michael Howard and Stable Diffusion

Now that’s what I call beauty! (or so sez me).

Stable Diffusion slightly resembles the media app Kodi, in that its open nature means it can easily be co-opted for questionable purposes. End users have the technical ability to “train” their own specialized SD models and upload them to public sites for downloading by others. This is a good feature. BUT… Visiting the home page of some such sites, you would be greeted by a wall of thumbnails showing scantily clad Asian females about the age of schoolgirls, posed suggestively. Mindful of McLuhan, we can observe that the electric light may be used to illuminate night baseball, midnight mass, or strip shows. In fact, it will be used for all these purposes.

New technologies often bring with them a spirit of hucksterism. We may be experiencing a feedback loop whereby the initial demographics of the (male-dominated) AI art community has led some entrepreneurs to market their wares in a sexist manner. Commenting in the Stable Diffusion subreddit, one person wrote:

I have stopped using [service] altogether because of this and, quite honestly, am often embarrassed to browse this sub because of the absolutely ridiculous amount of waifus, semi- or actually NSFW images, and how “photorealism” of young women is considered the target for these models.

The AI art scene has a real issue with this, and it doesn’t bode well for creating an awesome future with lots of people using new tools to create great new things. I’m an open-minded middle-aged man and find myself cringing constantly. I can only imagine what it’s like to open up this subreddit as a woman or someone new to the scene.

In the same thread, someone else wrote about a particular service:

Immediately upon landing on the front page you are scrolling through an endless deluge of sexualized women in skimpy clothes, both real and anime. There are no men to be seen anywhere, nor anything on the artistic spectrum for that matter. It is all models tailored to generate humans, bust shots, sexy poses. … Now how do you think that makes women in tech feel? Or just women in society in general, when they hear about AI and decide to take a gander to see if they could benefit from it? Unwelcoming at best, but let’s not be economical with words I’d be pretty ******* disgusted I think.

People often learn about art history at a university which has taken steps to create an environment that’s welcoming, non-threatening, and supportive of women. I hope I’m right in thinking that no woman could study art at university level without coming into contact with Griselda Pollack’s excellent work, or other scholarly work written from a feminist perspective. While a university setting may not always be the perfect place to learn about art in all its aspects, still, if it’s a somewhat cloistered environment, that tends to redound to the benefit of women, because it’s a safe space (at least, compared to the wilds of the Internet).

Some universities do offer courses in AI art, but given the fast-breaking nature of developments, whatever their faults YouTubers and Redditors often explain the latest breakthroughs quickly and efficiently, even if their style can be off-putting. Still, there’s a potential for culture shock. I can easily imagine a woman who studied art at university feeling bewildered by this hucksterish how-to, wondering what the hell it has to do with art as she understands it:

O brave new world, That has such people in’t.

The case for idiots replacing artists is made (unintentionally) in this video:

Yes, you too can sell “beater” images on “Itsy”! Ka-ching!

There’s also a pretty large creep factor to YouTubes which use AI to create slideshows of women with fully developed bodies, but what look like children’s faces. Ditto for the endless parade of “Corset Queens” and “Night-Wear Beauties.”

That AI art can be used in furtherance of get-rich-quick schemes, the displacement of real artists, and the objectification of women doesn’t prove that it’s a net negative. All new technologies are subject to abuse, but reining in the abuses without clamping down on freedom of expression poses a thorny problem, one which (paradoxically) AI still isn’t clever enough to solve. There’s a McLuhanistic sense in which we don’t use the tools — the tools use us.

I don’t want to over-problematize AI art, or indulge in elitism. This video shows Philip Joseph Adams and his mum Diana engaging in high hilarity as they play with Stable Diffusion on their (high end) home PC. After yukking it up over two silkie chickens riding a motorcycle, and Donald Trump and Joe Biden wearing tutus, they arrive at this image of “a pink blossom tree growing next to a snow-capped mountain in the sunset”:

Diana Adams shows off her Stable Diffusion creation. Screenshot courtesy OSFirstTimer: https://www.youtube.com/watch?v=DmtSLqhLcV4

Diana seems genuinely delighted to see her fantasy image come into being. This is holodeck stuff that does improve the quality of life for millions of people.

Conclusion

I want to close by saying that I do like creating images with the help of AI. My experience has been that it takes time and effort to make images which reflect classical ideals of beauty, and where the figures come to life with elements of human expression:

Woman of Antiquity No. 1, by Michael Howard and Stable Diffusion

In a sense, what we’re doing is training computers to reflect back images of ourselves — that is to say, of human civilization. When that training is sufficiently vast and well-tuned, we start breaking the complexity barrier and begin seeing results which can surprise us.

Plinth of Saint John Coltrane, by Michael Howard and Stable Diffusion

I dedicate this post to Sri Chinmoy, who drew hundreds, and thousands, and millions of birds, always using natural media.

A very happy April 13th to all those who revere Sri Chinmoy as I do!

---

GALLERY (slideshow) — Works created by Michael Howard with the help of Stable Diffusion

This slideshow requires JavaScript.

* * *