Tomorrow’s Blossoms: Hybrid Thinking, a new exhibition at Christine X Art Gallery in Sliema running until October 24, is a collaboration between professional artist Selina Scerri, artistic director at Malta’s Creative Science and Arts Institute, AI expert and entrepreneur Angelo Dalli, and cutting-edge AI technology.
It is the culmination of a five-year project that began while Scerri, studying for a master’s degree in digital illustration, met Dalli. Together they set to work at the intersection of human creativity and artificial intelligence. It’s been a fascinating journey.
The title of this exhibition was inspired by their first joint experimentation with AI in which they used innovative, bespoke software created by Dalli to combine images that Scerri had created with thousands of digital photographs of flowers. The result was a series of intriguing alien flowers suspended between reality and the imagination.
It was clear, however, that these AI-generated images lacked emotion and so they continued their collaboration further, exploring how to convey emotion in AI imagery.
“As a painter, emotions influence my work. The pieces an artist paints when in love are very different to the work they produce when they are sad or angry,” explains Scerri.
Runway 9“Their colour palette and their brush strokes change, for example, and this hasn’t been tackled in AI. We started our research before there were NFTs [Non-fungible Tokens], and when GANS [generative adversarial networks first developed for anomaly detection for security purpose] was the pioneering AI. Working from scratch, we had to find a creative way to bring the emotion of the creative process into AI-generated art: there were no tools to do this, no tutorials to learn from!”
Dalli and Scerri began researching imagery that embodies essential human concepts and emotions.
“Angelo created a bot to pull thousands of images with particular hashtags from Instagram and Flickr. These were #love, #anger, #sad, for example, and he created different datasets for each one. The majority were photos of people’s lives, and each collection looked markedly different.
“We then analysed each group, pinning down their key collective characteristics. For example, the images labelled #anger had powerful imagery that was predominantly black and red while #joy consisted of serene landscapes and pastel hues. #Sad was often blue and purple and there was lots of rain!
Nightbirds“#Love was the anomaly,” Scerri smiles, “because people love their partner and their family but they also #love Venice, Rome and their cappuccino!”
For each giant collection, Scerri identified both the core colour palette and the shapes and forms that emerged consistently. From the latter, she painted hundreds of abstracted images that simplified the predominant line directions and patterns she saw reoccurring. They were markedly different for each emotion: images with a sad tag were more likely to have vertical lines while zigzag shapes were often seen in the ‘angry’ set.
AI is simply a very powerful box of crayons and it all depends on how you use it
The resultant set of 5000 painted abstracts formed a carefully researched semiotic alphabet, which were then digitised. In addition, Scerri wrote 5000 nuanced prompts [a prompt is a description or task for an AI tool] for each emotion which reflected the imagery associated with each emotion, referencing her own art practice too.
These abstracts and prompts, Scerri then carefully combined in the painting app Procreate using AI tools – experimenting with many and favouring the generative AI model Stable Diffusion over Midjourney, for example.
“AI models are unpredictable. They often introduce surprise or inaccurate elements but with the careful choice of the tool you use, you can learn how to ‘aim’ them,” Scerri continues. “The testing and refining of your prompts actually gives an artist plenty of control although it is very time-consuming. It’s addictive too!”
This high-tech blending of internet-sourced data with analogue and digital paintings on a supercomputer generated thousands of new images through which Scerri sifted, a labour-intensive task, before honing the best to create art that, although AI-generated, clearly includes elements of her personal style.
An artwork from the exhibition Tomorrow’s Blossoms“It is the equivalent of producing sketches and studies before producing a final painting,” she explains. “Artists have aways looked at many other artists work to study their styles and techniques and to be inspired, combining all they have learnt with their own thoughts and research.”
“AI is simply a very powerful box of crayons,” smiles Selina, “and it all depends on how you use it. This collaboration has propelled me to produce better art.” The team used a variety of AI tools, including custom-coded AI systems as part of the Universal Machine Artist (UMA) and from third parties. OpenAI provided artist access to the DALLE models that underlie ChatGPT, Microsoft provided 3D city data and Stability provided access to their stable diffusion models.
This avant-garde exhibition includes a series of compelling art prints on canvas and on glass, and a large piece of wallpaper hung on a gallery wall – perhaps a wry reaction to the digital ‘wallpaper’ on our phones and screens. In addition, visitors can also enjoy a 10-minute film of the images which have been animated using AI: it’s powerfully ethereal, celestial, surreal and futuristic, transporting the viewer to a different dimension.
While this is a show with up-to-the-moment science at its core, running the whole gamut of emotions, it’s presented with human heart.
“The world is changing,” concludes Scerri. “AI is here; it is today’s ‘industrial revolution’. We have to embrace it and learn how to use it well. The human brain has already achieved so much: a hybrid of powerful AI tools and the best of human creativity can be incredible.”
Tomorrow’s Blossoms is showing at Christine X Art Gallery, Sliema, until October 24. For more information, visit christinexart.com/.