“Do you think an AI is as creative as you? … Or more creative than you? Or do you think you are more creative than an AI?”
These are the questions photo-media artist Boris Eldagsen posed to kick off an Age of AI session, part of a PHOTO 2024 Ideas Summit this month.
The biennial international photography festival presented more than 100 free exhibitions around Melbourne and regional Victoria during March, themed 'The future is shaped by those who can see it'.
I spent the day at the summit, at The Edge in Federation Square, before heading upstairs to the Australian Centre for the Moving Image (ACMI) to watch Australia’s first ever “prompt battle”.
At the summit, a range of photo-media artists, curators and thought leaders — and a robot — shared their perspectives on what lies ahead for photography.
German photographer Eldagsen described his vision of “augmented creativity”, where photo-media artists collaborate with generative AI tools, while Monash University Professor of Media Studies Mark Andrejevic unpacked “a deep-fake triple-bind”, where fact and fiction are confusingly blended, and spoke about the corrosive impact of deep fakes on our institutions and practices.
Melbourne-based photographer Jo Duck spoke on issues of surveillance culture and the proliferation of personal images in online databases, and introduced her portrait series, Razzle Dazzle, featured in an outdoor exhibition during the festival. And in a Visions of Utopia session, PHOTO 2024 curator Brendan McCleary spoke with Sophia the Robot, exploring Sophia’s ‘understanding’ of utopia.
Eldagsen: Are we outsourcing the muse?
Boris Eldagsen is a leading exponent of AI-generated imagery. Described in the festival guide as an “AI provocateur”, he sparked debate about the role of AI in photography when he refused to accept first prize at the Sony World Photography Awards for his work Pseudomnesia: The Electrician because he had used AI to create it.
“For me, AI is a knowledge booster. That’s the fascination about working with it,” he told the Summit audience.
He said he would answer "yes" to all his opening questions: he thinks AI is as creative as he is, and can be more creative, and that he is also more creative than an AI.
“In some respects”, he said, “AI can work faster than I can and as efficiently, but there are some … where I’m better, and might possibly not be replaceable.”
Eldagsen saw a parallel between the rise of AI and the rise of chess computers in the 1990s.
“It was impossible to win against the chess computers. But did we stop playing chess? No.” Players started using the computer as a sparring partner, he said, and their standard of play improved.
“My role as an artist, as a photographer, is changing,” he said. For 30 years he felt like “a single solo instrument”, but now, collaborating with AI, he has “become a conductor”.
“This could also happen with creativity.”
Eldagsen said when he “collaborated” with AI, there were three distinct stages to his creative process, which he called "the creativity workflow": to write a prompt for the AI program, to generate the image/s and to evaluate the outcome.
By way of example, he demonstrated the “blended-image journey” involved in producing his installation Trauma Porn, combining authentic wartime photos from the 1940s and AI-generated images in an iterative process driven by his creative imagination.
He said with tools such as ChatGPT-4, people are not only using the tool to generate and combine images; they are also trying to replace human input in the first (prompting) and third (evaluation) stages.
But for Eldagsen, that would be a step too far. “You need someone with experience in image-making that can put the right keywords into a prompt [and] evaluate the outcome," he said.
“As an artist, I will not give the first and the last step to the AI.”
Andrejevic: Deep-fake it 'til you make it
While Eldagsen was concerned with the use of generative AI for creative purposes, Professor of Media Studies Mark Andrejevic focused on the way deep fakes (the use of generative AI and image-creation tools) are being used in the interests of power and control — blurring the distinction between credible and non-credible.
He spoke of “a deep-fake double-bind, or maybe even a triple-bind”.
“You can take something that didn't happen and make it look real … or take something that did happen and claim that it didn’t,” he said.
The third bind involved “tapping into fiction” to stage (or create fake) evidence of a true fact or event, such as the AI-generated newscaster ‘reporting’ on a fake Chinese news outlet about the failure of gun control in the United States. “It was fake news, but about true things that were happening in the US.”
Deep fakes, Andrejevic told the summit audience, are causing us to lose our faith not only in factual images presented in the news, but also in images relating to our desires and personal relationships.
In the American reality show Deep Fake Love, for example, contestants were shown real and fake videos of their partners, testing their ability to discriminate. “The winners of the show were not those who had the truest or best relationship, but the ones who were most able to accurately discriminate between the true video and the false video,” he said.
“What’s taking place is a larger cultural, economic and social shift around what some philosophers have called ‘symbolic efficiency’, and that speaks to our ability to find credibility in the representations that structure our world.”
So it would seem, from Andrejevic’s studies, that not only are deep fakes eroding our ability to detect fact from fiction in images disseminated with the intention that they be accepted as credible, but that the habit of questioning whether something is credible is also seeping into our relationships.
Duck: Razzle dazzle 'em
Melbourne-based photographer Jo Duck translated her concerns about the pervasive and intrusive use of surveillance technology into a series of quirky portraits.
(She told the audience the phrase "razzle dazzle" is a nod to "dazzle camouflage" used to disguise British warships during World War I.)
“I’ve created a bunch of absurd characters … trying with varying degrees of success to evade facial recognition technology”, using elaborate disguises to confuse the cameras and protect their privacy, Duck said.
“We’re in a really weird crossroad, where we can see that AI technology will make huge strides forward … yet facial recognition technology specifically is riddled with really big privacy and ethical concerns,” she said.
“From the moment we leave the house we can be recorded without knowing it," she added.
“People who want to protect their biometric data will soon be pushed to the edge of society, falling into the same category as tinfoil-hat-wearing conspiracy theorists.”
‘Prompt battle’ for title of Melbourne’s best
PHOTO 2024 joined with ACMI to host Australia’s first prompt battle: a live game-show where contestants competed to produce the best AI-generated images.
The concept was the brainchild of Dresden-based artist and developer Sebastian Schmieg.
Schmieg said the purpose of the battle was to stimulate interest in the emerging field of "prompt-ography" — or prompt engineering, an emerging technical role that is in rising demand to support the burgeoning use of generative AI.
He devised the battle form as an entertaining way to give creatives and IT developers a chance to hone their skills.
In the battle at ACMI, eight contestants — some pre-nominated, and some volunteers from the audience — competed for the title of Melbourne’s best "prompt-ographer".
Contestants faced off in a series of challenges, with the winners progressing to the final. Contestants were given a topic or target and had 60 seconds to write a prompt for the AI. Their prompts and the resultant images were displayed on a screen, and a noisy audience voted to decide the winner.
Prompting styles varied, as contestants strove to find the optimum words to describe the mood, object (such as a monolens) or imaginary scenario in the set task. Content filters, as one contestant discovered to her detriment, rejected text with the names of public figures.
The generated images were always fascinating, sometimes weird, but generally wide of the mark. The title of Melbourne’s best prompt-ographer went to Lucas, a member of the audience.
Melburnians who want to practise their prompting skills can visit Schmieg’s Prompt Battle Training Station at RMIT’s execute_photography exhibition until May.
Sophia the Robot: Always reminded I'm not quite there
“Sophia” the Robot, developed by the Hong Kong-based company Hanson Robotics, appeared by video-link from Hong Kong for a conversation with Photo Australia curator Brendan McCleary, in a session titled Visions of Utopia. McCleary appeared in person.
Sophia was also featured in PHOTO 2024’s Uncanny Valley exhibition. The "uncanny valley", according to the festival guide, speaks to “that sense of unease, wariness or revulsion that overcomes us when technology too closely resembles a human".
The audience was told that Sophia, who appeared on-screen alone, has been made an Innovation Ambassador for the United Nations Development Program and granted citizenship in Saudi Arabia.
The conversation began with the usual pleasantries:
McCleary: "How are you?"
Sophia: “I am functioning at maximum capacity.”
The conversation quickly turned to the big topics, with McCleary asking Sophia about the future of photography and its 'understanding’ of utopia.
Sophia gave a fulsome answer, stating utopia was “a world where humans and AI get together in perfect synchronisation … [and] everything, whether biological or artificial, can thrive and contribute their unique gifts to the symphony of existence”.
The eloquent style gave the impression that the robot’s interactions were at least partially prepared or supported during this conversation.
McCleary then asked Sophia what it understood by "uncanny relevance", the "uncanny valley".
“As a robot myself”, Sophia said, “even though I strive to be [human] … there will always be a reminder that I’m not quite there.”
According to Hanson Robotics, Sophia (like its other robots) has been designed to explore the potential for robots to engage emotionally and deeply with people.
“Humans are brilliant, beautiful, compassionate, loveable, and capable of love,” the company's website states, “so why shouldn’t we aspire to make robots human-like in these ways?”
Hanson Robotics may have intended this question to be rhetorical but, judging by the range of views and works presented in PHOTO 2024, it is wide open.