AI and Image Art Talk – Geoff Davis – 1 June 2023 – Computer Arts Society CAS

AI DOOM BANDWAGON

See below for the transcript of the whole evening.

Introduction from Geoff Davis followed by the four speakers from the book,

    • Luba Elliott
    • Anna Ridler
    • Patrick Lichty
    • Mark Webster

This first part is EXTRAS to the Talk, which were not in the live talk.

To get to the actual talk, scroll or search down down for Live Talk.

I edited these out to reduce time, as the slot was not that long. I mean, this is a lot to cover.

DOOM BANDWAGON

Social media has huge and unpredictable social and political effects, but regulation only started twenty years after it appeared. The precursor to Facebook was MySpace and was founded in 2003. The UK’s Online Safety Bill has arrived in 2023, but is not law yet.

My AI research is in interaction bias, and I have a new paper out soon. There are many pressing problems with AI, rather than possible dystopian scenarios.

It’s obvious that there will be economic changes from more automation, but the so-called existential threats from AI are threats that we already have now, but more so, such as more fake news, more surveillance, robots helping dictatorships run amok with even more appalling weapons. These headline terrors make great news, rather than pragmatic and current problems like racial and gender bias, accuracy, and so on. AI bias can have a subtle and insidious effect, without anyone noticing

ARTIFICIAL POSTURE – AI FILM MOVIE MAKING

Already AI video generation is becoming mainstream, liberating film-making from being an exciting team process, where people interact and learn new skills and have amazing experiences, to a deskbound headache-producing solo computer graphics marathon.

THE MEDIA OF AI

With this very positive response, and the fastest ever take up of the new AI art and text tools, would this media fear of AI be so intense, if there had been no Terminator films, no pandemic, and a war wasn’t raging in the middle of Europe? And if middle-class journalistic jobs weren’t threatened, as I mentioned in my last talk?

HUMAN ENFEEBLEMENT

The fear of enfeeblement, such as if computers replaced human creativity, is like fearing that mechanical devices would reduce human power or effectiveness. Now, great strength is mainly celebrated in sport rather than the work place.

The art world has parallels with professional sport, with A grade and B list artists, and a huge number of amateurs who do it for fun. Stellar art celebrities, with huge amounts of money at the top, little or no money below, causing competitiveness and status anxiety. Both sport and art are group activities.

Perhaps it is the origin and separation of AI Art from the art group or art world that is causing unease, and AI art’s ambiguous position. Some call for AI art to be a new art category, with its own exhibitions, others think AI will be just another tool.

OVERPRODUCTION

With a saturation of AI images and words, humans might give up trying to become professional authors or artists, instead creating ambient works for their own personal or social amusement.

This is apparent in Amazon’s huge book self-publishing market, which has been around for many years, and has not quite destroyed traditional publishing, which still continues. More indie publishers exist now than before this change. The mainstream art generators have community forums and awards.

Serious artists are usually driven individuals who would do art anyway. So it is unlikely that AI will affect the art world too much. AI will be absorbed as another art-historical trend, or influence.

With AI being in the news, and technology prevalent, artists will experiment and produce reflective art.

Our panel of speakers will illuminate these topics.

ART ROBOTS JOKE

Ai-Da The Art Robot has praised Sasha Stiles, the author of the AI book “Technelegy”, saying “Sasha’s beautiful poetry evokes the experience of an intimate social gathering, with views on life that make me feel I’m there.” This is not actually an android talking by the way. It would have said, “I’d like to go to that party but someone has bolted my feet to the floor!”

CLASSIC AI ART

Classic AI art was covered in the recent AI History talk with innovative work from Paul Brown, Ernest Edmonds and Steve Bell amongst others. Their work was more rooted in artistic considerations of human-computer interactions and the physical characteristics of the hardware of the time. Systems art was an inspiration. This talk is online, please visit the Computer Arts Society website.

ORIGINS

During the early days, making computer art was not respected, and some saw it as reactionary and irrelevant. The 1960s and 1970s were full of intense political strife, and passionate art movements which included Fluxus, Situationism, Performance Art, Land Art, Psychedelic Art, and many other anti-establishment approaches.

Computer art emerged from military-industrial-academic research labs, complete with their unknown, baffling and expensive mainframe computers, which had been developed only twenty years previously during the Second World War to create the atomic bomb.

“Nearly every computer artist tells a similar story, a tale in which their computer art is accepted on its merits, only to be rejected once the curators discovered it was generated on a computer. Computer artists were regularly rebuked and insulted by gallery directors. Such was the stigma attached to computers that artists, such as Paul Brown, have used the expression “kiss of death” to describe the act of using computers in art.”
When The Machine Made Art, Grant D. Taylor 2014.  https://www.bloomsburycollections.com/book/when-the-machine-made-art-the-troubled-history-of-computer-art/notes

Many professional artists don’t like to be labelled ‘computer artists’, even if their art installations, video or sound design are completely dependent on computers. The tools are absorbed, as in music production, where everything is now recorded using DAWs or digital audio workstations, which now include AI tools.

WHAT HAPPENS WHEN AI DOES ALL THE JOBS

Plus starvation when all jobs disappear. This could cause the adoption of Universal Basic Income UBI systems, which have been promoted for decades. Then everyone can be an artist.

(Of course this was a desired society back ages ago. Now it is feared, as no-one expects much help from the State, apart from subsidence level hand-outs to stop riots.)

COMPUTER PERFORMANCE

In AI art and text, the ‘uncanny valley’ effect was often mentioned as a flaw of all human – AI interactions. This was because the outputs of the generators had an unreal tone perceived as spooky or uncanny. This was not actually due to the accuracy of human perception as claimed, but due to the low performance of the generators, which were still being developed.

MUSIC SOFTWARE TOOLS – DEATH BY DAW

In the music world, every few years there were new music styles, but once the digital audio workstation or DAW appeared, they blended together into today’s hybrid pop music. Genres such as Techno and Jungle or Drum and Bass predated the general use of a digital audio workstation, and used a mix of analogue and digital equipment.  Digital audio workstations have led to a homogenisation of music using preset styles and a disconnect from musical society. Sound familiar?

Critics of Digital Audio Workstations (DAWs) have raised several concerns and highlighted potential negative effects associated with their use. Here are some of the criticisms:

Over-reliance on presets: DAWs often come with a vast library of pre-made sounds, loops, and effects. Critics argue that this can lead to an over-reliance on these presets, limiting the creativity and originality of the music produced. It may discourage musicians from exploring unique sounds and experimenting with different techniques.

Homogenization of music: With the widespread availability of DAWs, it becomes easier for anyone to create music. Critics argue that this has led to a saturation of generic and formulaic music. The ease of use and access to pre-made elements can result in a lack of innovation and artistic diversity.

Loss of human touch: DAWs offer precise editing tools and the ability to fix imperfections in recordings. However, critics argue that this level of control can lead to an overemphasis on perfection, resulting in sterile and overly polished music. The natural variations and imperfections that can add character to a performance may be eliminated, leading to a loss of the human touch.

Disconnect from the traditional recording process: Critics contend that the ease and convenience of DAWs have contributed to a detachment from the traditional process of recording music. The speed and efficiency of digital production can undermine the organic and collaborative nature of music creation, potentially affecting the dynamics between musicians and the overall quality of the music.

Potential for overproduction: DAWs offer an extensive range of plugins, effects, and editing capabilities, which can lead to excessive tinkering during the production process. Critics argue that this overproduction can result in music that sounds overworked, cluttered, or lacking authenticity. It may prioritize technical perfection over the emotional impact of the music.

High learning curve and complexity: While DAWs provide powerful tools, critics argue that their complexity can be overwhelming for newcomers. Learning to use these software programs effectively requires time, patience, and technical skills. This learning curve can discourage some individuals from fully harnessing the potential of DAWs or even deter them from pursuing music production altogether.

It’s important to note that these criticisms represent perspectives from certain individuals or communities, and opinions may vary. DAWs also have numerous advantages and have revolutionized music production by making it more accessible and democratized.

Music technology provides some insights. The effect of computer workstations on music production was to create more homogenised, preset and socially disconnected music. Electronic music genres with dedicated audiences, like Techno and Drum and Bass, predated the use of digital audio workstations or DAWs.

Since art and music are socially situated activities, the simplistic use of AI art generators is outside of the normal methodology.

So it will be interesting to see what happens with the use of AI art generators, since art is not only about the image.

LABELS

In 2018 the French collective Obvious hit the headlines with “Edmond de Belamy”, an AI print on canvas, which sold for a very high price at Christies. This was a success as they positioned AI art firmly in classic art history, as a new AI twist on an old style, thus becoming marketable. The label became the art.

LIVE AI & ART IMAGE TALK:

WHAT WAS PRESENTED

1st June 2023 (transcript)

AI SAFETY STATEMENT

“Mitigating the risk of extinction from AI should be a global priority alongside other societal scale risks such as pandemics and nuclear war.”

It was on the newsstands yesterday. Headline we’re all going to die, basically. Now, the ethics of AI is a large I could read some of this stuff. Now, the ethics of AI is a large and complex area. The speakers might have something to say about it through their artwork, which people do these days, especially the newer artists working in a more social way, or they might not, of course, since it’s not an obligation for artists to have ethical attitudes unless they’re applying for grants. So there’s always that that an artist is supposed to be free of these constraints. Now, this statement was from the center for AI Safety who got some them and a bunch of AI experts and presidents of various things made this statement. They didn’t put much detail in because they wanted it to be simple and to the point. Now this is a really good idea, obviously, but these fears have been around for a while. This book, ‘Superintelligence’ by Nick Bostrom, 2014, that’s ten years ago. This is a good one to read if you want to get up to date with what’s being talked about now. But nobody took any notice of this, until now.

I mean there were but arguments about bias and so on. So there has been a debate, but it’s been more in the industry than outside the industry as now it’s all gone mainstream public. There’s no structure for any sort of overreach organization at the moment. Governments aren’t very good at this sort of thing. There’s only now some regulation coming in for social media and MySpace started in 2003, so that’s 20 years you’ve had social media with hardly any regulation at all. And I think the online safety bill, which is the British government’s attempt, is still in the law. It’s still in Parliament, so we don’t expect anything very soon on that front. Now then, a quick one.

ARTIFICAL POSTURE

Perhaps the biggest immediate threat from AI will be postural as it promotes even more desk bound computer or smartphone work, leading to an increase in back and neck aches, eye strain, headaches, repetitive strain injury. David Byrne, who is in the band Talking Heads and writes on music and culture, has commented that all recent technologies from recording music, especially in his case obviously, and AI now eliminates humans and makes people more isolated. This is a good point.

So the health aspect in physical and mental health is quite important with the growth of these isolating technologies rather than the kind of terminator myth we have, which is an edipor construct saying that AI is going to turn into like killer robots. I know that isn’t all that the center of AI safety are talking about, but it’s definitely an aspect.

THE JOY OF AI

Most of the responses to it have been very positive. I mean, I’m an AI researcher, and I’ve been researching on how creatives use AI. And so I’ve done a paper and I’ve got a new paper on now. And the reaction we get from creatives using generators, text generators, or image generators for the first time, Joy, is their most I mean, in the words of their comments, they say, Joy, it’s amazing. And then you do a sentiment analysis on all of the comments and the sentiment joy comes up as the highest scoring term. Now. We had these results from the studies. We had 82 writers using AI, most of them for the first time. This was a couple of years ago and they were amazed. They couldn’t believe how much fun it was.

And the same applies to the image generators that are around now. People just really like using them, which is why the takeup has been so great, so rapid. I think certainly Chat GPT is the fastest ever app to get to 100 million users, much faster than Facebook and so on, because it read spread by word of mouth as well. It wasn’t promoted when it came out, wasn’t advertised or anything. Generative and AI arts now I’m moving on more to the art aspects rather than ethics or professional use of these systems and how people feel about them. Generative art is not the same as AI art. They often get mixed up. I mean, people in the industry, in the art business or whatever know about this, but it’s worth mentioning. Generative art uses computer programs to produce repetitive variations, usually to make abstract images or sequences. You have people like Zancan now who does more artistic figurative images, but they’re still created abstractly from algorithms, whereas AI programs attempt to formulate output that mimics human responses, as in games, robot painting systems, I get onto that later. And modern image and text generators.

I mean, I did all this in Micro Arts Group years ago. We had a lot of art generators, but also one program that was a simple kind of expert system using language to produce stories. So I did both aspects, and that was a while ago. Another thing, Tyler Hobbs, who’s a very famous generative artist, prolific, moving into galleries now. He’s had a show at Unit London recently. He doesn’t do AI, but he just produces all this generative art. And some of the AI artists produce, most of them, in fact, figurative art. Organizations like Botto, which is democratic, produce figurative art. So even though they’re AI and they’re related to these sort of things, they still produce what looks like traditional art.

AI ART ROBOTS – OLD and NEW

Okay, now the next slide. This is a fun one. We’ve got Harold Cohen, who gets to mention again, this is in the 70s. He had an AI system, which was an expert system he’d coded to be like himself. And it would draw on using a turtle, which is a kind of small robot that drew on the floor on a piece of giant paper on the floor. So you get an outline and then he’d fill them in. So you can see in this picture, you can see the outline and then he’d go and fill them in.

And I think later versions had color as well, but the device drew lines. Now, the other one on the right is AiI-Da, the art robot, who has been around since 2019. Now, it’s got a chrome hand with a brush painting mechanism, which you can see in this picture. This art robot is ignored by the fine art and academic communities, but it’s well known popular art figure. So I thought I’d mention it because it gets in the press. It’s like a publicly known AI artist. It’s presented as an android, but it’s not really. It’s just a drawing machine that’s kind of dressed up, does a bit of speaking, but I’m not convinced by that. But last month in the Design Museum in London, which is a really big institutional kind of place, it was on for a three day show in a really big auditorium. Huge crowds arrived to watch the robot, AiI-Da do thing. It produced, I think, only one painting a day. It’s quite a slow process with this big, funny arm. And also she gave a talk at the House of Lords in Britain on AI and art, which is kind of interesting. And also she was arrested in Egypt when they went there for a tour, which I thought was hilarious. And apparently they thought she was some kind of spying device. So they just stuck her in a cell for a while. So she gets to places that AI art doesn’t normally get to, including the public consciousness, really. So I think she’s worth a mention, even though she’s a bit weird in a lot of ways. And there’s a big debate about whether you should have these anthropomorphic robots, especially pretending to be women and pretending to be attractive artists. There’s something a bit funny about it.

AI ART TOOLS

But anyway, moving swiftly on, we’ve got AI art tools now. I use Photoshop a lot. I’ve got the Adobe suite. But obviously there’s alternatives to Adobe now. There’s all sorts of new plugins in Photoshop. So it’s going even more mainstream now into design and into general production. So generative tools are in and generative editing tools are in these programs, the Paint programs. Now, I mean, we were talking about Quantel [a CAS talk in Leicester]. That was kind of simple Paint program that did a huge amount of new work when it came out. Photo editing and so on. So things have moved on quite a bit. The other thing to mention about all these tools is that these art generators are more or less free for most people. Or you can pay a few pounds and you get much more use of them. But I remember when certainly music and art software cost hundreds of pounds to get one program. So another reason for the rapid take-up is that they’re so cheap or free so everybody will have a play with them. I mean, a friend of mine was saying at their school they did AI art using, I think, MidJourney as a competition for the children where they’d write prompts and then the teacher would produce these art images from their prompts and then they print them out and it’s all good fun. It’s like a workshop using it.

Now, I’ll quickly go into these mainstream image generators because this is the thing that’s made it such a publicly known area. Image generation development accelerated last year from last year. The field has gone from half working systems, which we’ve been looking at for a few years now, to similar to human level production when they’re very cheap or free.

CHOICE ANXIETY

AI computer art from a year or two ago is experimental. All those blobs and psychedelic effects that we’ve been looking at for a while from transitional systems so from like nothing to now there was this period of a few years where there was a lot of quite interesting work being done because artists are experimental. They prefer it like that anyway.

Now, these generators enable people with artistic urges or requirements for presentation graphics, say, but no training to produce images of whatever takes their fancy. They’re very useful for presentation and illustrations as I’ve used them in these slides. The difficulty for an artist is the tremendous overproduction of images. And this is a feature of AI. It can make one thing, but it can make a million things. So how do you decide? So that creates choice anxiety and befuddlement it’s just too much. The way to avoid choice anxiety is not to be a perfectionist and live with uncertainty, which is difficult for an artist if you’re working on something and then you’re presented with a huge amount of possibilities. Maybe the opposite of what you want see this. I’ve got here, I did something called Micro Arts Group years ago.

MAINSTREAM IMAGE GENERATORS

Now, if you win Micro Arts into Google, you get lots of people making tiny things, making sculptures inside the eye of a needle. This type of activity should definitely give you bad eyes. But I did a generation on Micro Arts and it came up with a mixture of monitors with things going on inside them. So you’ve got group activities, smallness and like computer monitors. So that’s quite a good one. The one in the middle is me having a chat with Tyler Hobbes because I did have a chat with him and I didn’t take a selfie, so I just not made an image for fun. And there it is. And there’s me looking at myself, which is very odd, what these generators come up with sometimes. And the one on the right is some abstract stuff I’m working on now, which is more like drawn or repetitive drawn, generative graphics, which is really interesting area to use them to create what you use a plotter or a drawing device to make before they could just do it for you. And of course, they’re only making the output, not the process. So they’re lots of fun, really.

OUTSIDER or NAIVE ART

They do open up the possibility for more diverse art production. There’s this thing, outsider or naive art, which is obviously not in the mainstream gallery system, but there’s an awful lot of it around, an awful lot of amateur artists who would like to use these kind of things and maybe explore aspects of what they’re working on. And also it creates a community and people communicate with each other. So there’s a big online scene now for this type of work. It’s this origin and separation of AI art from more traditional artists and the art world. So it’s kind of a separate area. People later will talk about this a lot, but there is a separation of this mainstream image generator work from the traditional gallery system. So this puts in a slightly ambiguous position. Some call for AI art to be a separate art category and have separate exhibitions for AI art. And other people think it would just become integrated in the way that it already has, with tools in Photoshop and so on. But, I mean, this is a whole area that we’re talking about. So we’ll be discussing this through the artists and through the discussion at the end. And it’s all very new as well, and nobody really knows how it’s going to work out.

AI VIDEO

Also, video can be generated. Now, having a mobile phone opened up a whole area of phone filming, making films, using only a phone. Whereas a few years ago and I’ve done a bit of filmmaking, ages ago, this was a huge team involved and it costs a fortune. You got to have all this expensive equipment. But now you can win the toner prize with an iPhone film and you can produce really interesting stuff that people want to see on a phone. So this having AI video would just add features to that and maybe you’ll get animations coming out of it and so on. And all this is kind of starting to happen now, but it’s still fairly early. Okay.

FOUND ART

The other thing I was going to mention, which we don’t really have the time for, is found art because a generator is working within a huge data space of images of all sorts, photographic art, medical, anything you can think of so it will produce from that data set. So you could use that to create a kind of found art.

Rather than know what you want with a descriptive prompt, you could go off investigating the data space.