Talking AI, Art, and Entangled Realities with HeK Basel’s Director Sabine Himmelsbach
Interview by Marie Chatel
Ahead of Art Basel week, we’ve met with Sabine Himmelsbach, director of the House of Electronic Arts (HeK) in Basel, to talk about artificial intelligence and art, as the institution dedicates its latest exhibition Entangled Realities: Living with Artificial Intelligence to artists engaging with the revolutionizing medium.
Hello Sabine, nice to meet you. We’re going to talk about your exhibition Entangled Realities in a moment, but before that, I’d like to ask you about artificial intelligence in general. The technology is becoming more popular as an art medium, yet has been around for a while. Could you explain us its origins and why it sees such momentum right now?
AI research started in the 1950s with the famous Dartmouth Conference, and has since then gone through times of up and down. In the so-called “AI winters,” researchers and institutions lacked funding, and progress stopped for a while. But then the field saw another boom with the rise of expert systems, followed by another AI winter. The development of artificial intelligence has, for a long time, been based on what is called “Symbolic AI,” where researchers are controlling all the steps that machines make when processing information. In the mid-1980s, researchers turned their attention to neural networks based on biological models of the human brain and its processing of information. Deep learning, a specialized form of machine learning rooted in human brain function, and the availability of large amounts of structured and unstructured data have allowed the current success of AI. Today’s AI is largely dominated by machine learning, based on artificial neural networks which mimic the workings of the human brain with regards to assimilating and analyzing information. Generative adversarial networks or short GANs are trained for pattern recognition by using two competing systems, one for recognizing images and the other for generating them. In our exhibition, Entangled Realities, most of the works are reflecting machine learning. When we refer to artificial intelligence today, people mostly think of machine learning, but it is, in fact, a particular form of AI.
What is different is that, with these new training methods, machines start to learn independently, and researchers can’t actually tell how a system has reached a certain conclusion, which can scare people. The training of these algorithms would not be possible without the large data sets made available by all the material that everyone uploads online and on social media platforms. Overall, increased computer power and large data sets have revolutionized the development of artificial intelligence.
Could you tell us more about the curatorial line you chose for the exhibition and what challenges you faced in making the content accessible?
Our biggest challenge was to choose a specific focus for the exhibition. As you know, a lot is happening around art made with AI at the moment, and so we put a lot of thought into defining our specific approach. For instance, we didn’t want to have a historical overview — which would have matched what the Barbican is doing with its exhibition AI: More than Human. We decided to focus on the notion of entanglement and brought together works that are making use of AI as a creative tool, and others that reflect on the impact that AI has on society (without necessarily featuring the technology per se). We show works that focus on the training of AI systems and that reveal how differently these systems “see” the world. We also wanted to show how much these systems are already implemented in our daily lives.
We see a lot of techno-skepticism and dystopian discourses in the press. What would you say are the potentials and threats of training machines? Should we be so afraid of these new realities? I believe art has a role to play in questioning our fears and helping us perceive the true nature of this technology.
Yes, it’s an important point to be made, and one we address in various ways throughout the exhibition. We do not believe AI should be perceived as a threat and something we should avoid, not at all. Instead, through our selection of artworks, we advocate that the technology is needed as our world becomes increasingly complex, but we have to understand how it works, be clear about potential biases, and find solutions and strategies to avoid certain tendencies.
Mostly, it is vital to be mindful of how we train these algorithms and what we want them to learn about the world because training has consequence. That’s why we chose the title Entangled Realities — to emphasize these effects that are already manifested in our world. We are not talking about something distant, which might impact us in the future, and so we need to be sure of what we teach these systems. For instance, if the police use AI to check criminals and only feed the computer with data sets of black people, the machine will think all criminals are black — a partiality we have to avoid.
Artists make us aware of these facts. For example, in their installation i am here to learn, so :)))))), Zach Blas & Jemima Wyman reflect on the story of the Twitter-Bot Tay that turned into a racist personality within 24hours after being trolled by twitter users. But artists also use AI as a new tool. For example, Mario Klingemann describes AI as a new kind of brush for artistic expression and the electronic musician Holly Herndon uses it to explore and enlarge her possibilities for sound. While expanding their field of creation, artists remain very conscious of the way they train their algorithms and the input they provide.
From Lauren Mc Carthy’s Lauren to Mario Klingemann’s Uncanny Mirror, many artworks on the show rely on interactivity. Are we forgetting about the presence of interfaces and the inputs we give?
Interactive pieces were key for us to make people understand how algorithmic systems perceive the world. They are not “seeing” as we do, they perceive the world on an abstract and code-based level. The print series Activations of James Bridle illustrates well the different levels and depths of an artificial neural network. The prints are views from the road that reveal the different layers of the neural network turning recognizable street images into abstract patterns. Other works offer interaction so visitors can understand these levels of abstraction. For example, in Ursula Damm’s piece, people can turn nods on an interface with direct effects on the neural networks and the visual output. In Sebastian Schmieg’s Decisive Mirror, you experience real-time feedback in terms of written categorizations based on the face of the viewer, while in Mario Klingemann’s Uncanny Mirror, the machine is reflecting the visitors’ faces based on its learned data to represent a face.
You’ve also commissioned several works for the show, what role can a museum like the HeK play in supporting new media artists in their creative development?
Commissioning provides artists with a great opportunity to produce new works and I think it’s something institutions should do to support them. We commissioned several works for the show. Decisive Mirror by Sebastian Schmieg is one example. Another is Membrane by Ursula Damm. The video installation of electronic musicians Holly Herndon and Mat Dryhurst was also specifically made for the show — it documents the first training performance of their artificial neural network called “Spawn” as it interacts with other ensemble members and a general audience. The training and performance had already taken place, but the video installation was newly produced.
The most important commission is a work by Swiss artist group fabric | ch. We asked them to train an AI system for the scenography of the exhibition, or the placement of the selected works, so that AI would also impact the overall rendering of the show. The system registered all the data on the works to be showcased, such as the space needed, the amount of light or darkness required, but also practical needs such as walls, projectors, screens, etc., to constantly generate optimized curatorial solutions. Clearly, the machine sees everything as being a single, modular element. As a wall is made of several elements, the machine produced a layout that was fractured in a way that we would never have thought of. The piece by fabric | ch was therefore called Atomized Curatorial Functioning. This was very inspiring for us and in the end, we decided to build our own scenography, but the AI influenced our choices. I think collaborating with AI is fantastic because it opens up new directions we would have never dared to take. This collaboration made the scenography of the show very dynamic and we would not have done it the way we did it, if not for the system and its singular problem-solving approach.
It seems like the scene for media art is fast evolving. What changes have you observed in your audience and its concerns as new technologies gain leverage over our daily lives?
Part of our general mission at the HeK (House of Electronic Arts Basel) in Switzerland is to address the changes in society brought about by media technologies. Art always reflects on its own time and media artists use the tools of our time to reflect on these changes. At HeK we showcase outstanding artists in the field of media art and try to address relevant topics for society at large. For instance, in 2015 we had a show titled Poetic from Politics of Data that delved into the topics of big data and data mining — which had just started to become relevant in the media at the time. Art allows different perspectives on these topics, be it big data or data mining, or issues related to AI that we are addressing in our current exhibition Entangled Realities. Art allows us different views on these topics and makes them more accessible.
As Basel turns into the art world’s go-to destination, the HeK runs special events throughout the week. What would be your highlights?
We have several special events during the Art Basel week. On Tuesday, June 11 at 8 pm, we host a reception for the exhibition Entangled Realities followed by a concert of Norwegian artist Jana Winderen — a collaboration with the watchmaking company Audemars Piguet, who commissioned this new piece composed of recordings of natural sounds that would not be accessible to us without technology. Another essential event is the ceremony of the Pax Art Awards on Thursday, June 13 at 7 pm. This is only the second time we celebrate these new awards in collaboration with Art Foundation Pax. We have just announced this year’s winners — the artist group knowbotiq, who will receive the main award for their established artistic practice, and artists Alan Bogana and Félicien Goguey as featured upcoming artists. We will also showcase a selection of their works, although the main exhibition will take place at the HeK in spring 2020.