What does AI look like? Google DeepMind’s Visualising AI demystifies through original and imaginative artworks

In the realms of public consciousness, AI is a multi-faceted and mysterious topic, dominating global headlines with its promise to accelerate and revolutionise many industries. When it comes to the creative industry specifically, the question of ‘for better or worse’ is still up for grabs. Some creators advocate for the efficiency of generative AI and ease of predictive technology, while others remain dubious due to issues of copyright infringement and bias – not to mention pressing climate concerns and the threat to human-skilled labour. Cold, dystopian visual connotations further propel scepticism. This is why education surrounding AI as well as open, accessible conversation with its creators is crucial. With this in mind, Google DeepMind’s ongoing project Visualising AI aims to engage and foster public understanding of AI through approachable communication.

Visualising AI invites pioneering digital artists to creatively reimagine the transformative technology that’s changing culture as we speak. By laying bare its tangible qualities, the series of artworks highlights the most relevant topics currently in AI, empowering individuals to better participate in contemporary discourse. The project explores complex concepts within AI, shedding light on the lesser known aspects of the hotly disputed innovation, from groundbreaking medical breakthroughs to sustainable energy solutions that could power our cities in the future. Now in its fourth iteration, this season’s artists include Aurora Mititelu, Bakken + Baek, Dada Projects, Jesper Lindborg, Twistedpoly, Winston Duke, and Zünc Studio. The work is now freely available to download on free stock image repositories Unsplash and Pexels.

Above The latest round of Visualising AI from Google DeepMind: A showcase of diverse artistic interpretations challenging conventional visual representations of artificial intelligence.

The chosen artists are curated for their rich diversity of perspective across a range of demographics, from geography to gender and culture. In collaboration with the industry’s leading experts at Google DeepMind, the artists are tasked with visually exploring chosen domains with the help of consultations from Google DeepMind’s researchers, ethicists, engineers and project managers. Taking into consideration the social benefits of AI, scientific excellence and integrity, not to mention responsibility, the artists are encouraged to freely interpret concepts with unconventional originality and fresh perspective.

For the London-based, female-led 3D motion studio Dada Projects, the challenge lay in visually interpreting the science-heavy topics of sustainability and weather forecasting within AI. Describing the brief as “a challenge we couldn’t resist”, the studio leaned into the portrayal of AI not as a machine, but something more organic. “Something that coexists with nature rather than dominates it,” they explain.

Dada Projects

As a future-facing studio concerned with issues of climate change, Dada Projects quickly learnt how AI is becoming increasingly integral in predicting climate-related disasters and in turn, helping reduce their impact. It does this by analysing real-time data against historical data in order to forecast potential risks. In turn, for artwork around the topic of sustainability, interlinking themes of technology, nature and art arose during the ideation process. “We were fighting against the idea that AI is ‘other’, a separate entity, and instead wanted to show it as an extension of our natural world, a tool that can harmonise with it.” Creating soft, organic textures amidst fluid motions to evoke the natural environment, Dada Studios utilised real-time geographical data to add an additional layer of storytelling to the otherwise quantitative information. Additionally, they used a plugin called DEM Earth to extract territorial data from Google Maps which was then implemented into Cinema 4D. Such techy tools allowed the designers to experiment with differing visual styles corresponding to the data.

For Dada Project’s visualisation of weather forecasting, the studio poetically portrays the various stages of changing weather through slow motion movement. The viewer can observe temperature changes through flowing shifts in colour. In contrast to the sterile, mechanical preconceptions of AI, Dada Project’s works emphasise how AI is not purely futuristic. In fact, AI is deeply rooted in the present. “The motion is deliberate,” says the studio, “echoing how AI processes information. Calm on the surface but with immense power and complexity underneath.” Allowing the beauty of nature to guide the creative process, the studio drew inspiration from scientific imagery – microscopic snapshots of coral reefs, satellite imagery of changing coastlines, the way light moves through water, how water systems develop – to inform their exciting new collaboration bridging humans and technology.

Above XK Studio: Artificial General Intelligence

Jesper Lindborg

London-based artist Jesper Lindborg similarly looked to nature to inform his visualisation of root node problems – fundamental challenges that, when solved, have a cascading effect, unlocking solutions for numerous interconnected problems. Commonly used in AI algorithms, root nodes are seen at the beginning of decision trees – directional graphs made up of a series of nodes representing classifiable categories. They are a type of flowchart used as a decision-support tool and importantly, they are easy to read. Representing entire sets of information, root nodes can be split into vast decision trees according to various features and sub-decisions. While root nodes are popular for their wide-reaching accessibility, their disadvantages lie in a tendency to simplify complex sets of information, thus predicting inaccurate outcomes or generalising test sets. If the wrong decision is made at the root node, it has a huge downstream impact on the outstanding decision trees. In Jesper’s interpretation, he imagines a tree – an effective metaphor for communicating branching systems, interconnectivity and growth. Contrasting artificial translucent textures meet concrete. Jesper explains: “The concrete serves as the ‘problem’ and breaking through it symbolises opening up new possibilities. Or in the context of a node tree problem, new nodes.”

In his second film exploring the same topic, a colourful pathway nods to the process of self-learning. The film’s viewpoint treads a path of repetitive problem solving, navigating obstacles along the way. “I aim to show the breakthroughs in science and research within AI while facing adversities. I also wanted to keep the visuals detailed to reflect the complexities inherent in root node problems.” Purposely moving away from the stock image-led connotations found a quick image search away, Jesper hopes the project can move perceptions of AI away from stereotypical, simplistic portrayals. “The way we view AI shapes our understanding of it,” adds Jesper. “Moving away from cold, lifeless imagery allows us to engage with AI more thoughtfully and see its potential in a more relatable way.”

Winston Duke

For motion designer Winston Duke, it’s important for creatives to engage in conversations around AI as a way to “return creative power back to artists.” Tasked with visualising the hotly contested generative AI, Winston’s approach lay in explaining how the technology actually works in order to make the topic more digestible. “Generative AI allows us to create new content by learning patterns through existing datasets,” explains Winston on his process. “These supervised or unsupervised models can generate text, code, imagery, video and so on.” After unpicking the details of this complex process, he started drawing webs of interconnectivity. Tracing a straightforward framework of ‘input, predict, output’, he created a traditional cell animation to demonstrate the complexities of the technology. Using a mixture of 3D software and cell, Winston analogises how agents and models are triggered once input data is programmed. Tying everything together with an analogue grain to inject a sense of imperfect humanity, the opposite of robotic sterility, Winston combats the oversaturated depictions of clinical AIs.

Generative AI’s boom has shaken up the creative industry, with its concerns as fiercely debated as its benefits. For Winston, it was essential to incorporate this dialogue into the nuance of the work. “The fundamental aim is to create a better perception of what may currently be seen as negative,” says Winston. “I feel like the creative industry would be more receptive to AI knowing artists are at the forefront of change through their own volition.”

As practising artists in the creative industry, creators including Winston, Jesper and Dada Projects have witnessed firsthand how AI has dramatically changed the creative process and the industry at large. All three creators have incorporated the technology into their process in some way, either to cut corners, ideate or perform technical tasks. In this way, they are open-minded as to how AI can transform and contribute to our existing sociocultural scape. “If we continue to view AI as potentially negative,” evaluates Dada Projects, “we risk alienating ourselves from the tools that could help us solve some of our most pressing global challenges. By visualising AI in a beautiful and accessible way, we open up new possibilities for collaboration.” AI and the narrative around it becomes richer with more informed, diverse perspectives and interpretations. The technology is already impacting culture as we know it, and whether we like it or not, we are already learning to live and adapt to it.

Above Bakken and Baeck: Quantum Computing