Beyond the Unknown - An interview with Valencia James and Alexander Berman, the creators of AI_am
An interview with Valencia James and Alexander Berman, the creators of AI_am
AI_am is a unique international collaboration that investigates the potential of the creative use of artificial intelligence (AI) through contemporary dance. Their experimental performance AI am here will debut at 10 October at Trafó House of Contemporary Arts, in the frame of smART!-series at CAFe Budapest Contemporary Arts Festival. We talked with the creators, Valencia James and Alexander Berman about their project, the creative potential of AI and the increasing role of machines.
How did AI_am start and what did compel you to work with artificial intelligence (AI)?
V: I have been quite curious about all these technological changes that transform our lives –– disruptive technologies they call them. I saw a documentary with Marvin Minsky, who is known as the father of AI, and my husband, Botond Bognár who works in Silicon Valley also told me a lot about this field. What grabbed me is the idea of how the performing arts, and especially contemporary dance, a field where the human body is at the centre will change through these new technologies. I got the opportunity very shortly after to research this topic through a residency called Research to the Unknown which was offered by Workshop Foundation in 2013 at Trafó Budapest. I started the research with Botond and then we reached out to Gábor Papp from XORXOR who invited his colleague Gaspar Hajdú and then Alex to join. We started to research how I could have an improvised duet with AI. This became the basis of our project.
A: I worked as a software designer and artist and I was enthusiastic to participate in this project as I was already very interested in AI. I started researching soon after Gábor introduced me to Valencia as the presentation of the residency was quite near in time. The technology we used back then was not really an AI, but it could react to Valencia’s movements. At the beginning of 2014 I managed to develop a system that was capable of improvising autonomously based on the recorded dataset of movements. We started to write about our findings and present them at conferences, such as the MOCO Symposium, one of the most recognized international workshop for movement and computing and TEDx Danubia where a larger audience could learn about our project. We also participated at the IJCAI, the most prestigious conference on AI, where we presented the software in an exhibition alongside renowned artists like Olafur Eliasson.
You have a very international team; the members of your project are from five different countries. How did you select these people and how can you manage working together?
V: This is really unique about our project. Our team developed more in an organic way, many of our members were recommended to us or we knew them before through other projects. It took a while to find all these people because AI is still quite rejected in the performing arts and the dance world is more conservative. We wanted to work with people who are open for technology based art projects. Most of the time we work remotely and online, and we only met at the residencies or at the rehearsals.
Your performance shows an improvised dance between human dancers and an avatar. How does the virtual dancer work? Is this system also capable of generating movements autonomously?
A: At the beginning we recorded Valencia’s movements with a Kinect camera, and then I created a map of possible poses out of them. I applied a machine learning process called dimensionality reduction and added a method which works like a cursor that moves along the map. At each point there is a pose and by moving the cursor along it a sequence of movements is generated. This movement itself is autonomous because there is no human who pulls the sliders. However the system is not fully autonomous it is more like a partnership between the algorithm and the users. Right now we are switching to a motion capture suit called Perception Neuron which is able to capture full-body complex movements in real time. We are also improving our system so that it can learn things on the fly, during the actual interaction.
What was the most interesting outcome during your project so far?
A: For me it was to see the results of the pose mapping method and hearing about how Valencia was using it in the studio.
V: This would be in 2014 when Alex had a breakthrough with the research and presented the software with an avatar that could generate novel movements autonomously. It was an improvising stick figure, and doing the most graceful movement sequences I had ever seen before. I could never have imagined such movements on my own. It was surprising and inspired me to want to move with the avatar so much, so that I experimented with using it as my choreographer for a while. Whatever it did I let it influence my movements. Today my movement style has been partly shaped by interacting with the avatar.
Another outcome that surprised me is how much I developed a connection with the avatar, in a similar way than I would connect with another human dancer. Throughout years of working together as a creative and performance partner, I have developed a sense of empathy for the avatar. I even named him, Merce!
There is a lot a fear towards the increasing role of machines in society. What do you think of that? Do you think these fears are realistic?
V: It depends a lot on us. A more diverse group of people need to work and think in the field of AI. We need folks from different social, cultural, economic backgrounds. More women, more people of colour, and especially more artists. The creation always reflects the creator so we have to be very careful not to replicate the social injustices and biases we should be fighting against in the technology we develop.
A: I think it is very interesting to see forms of intelligence that are very different from us.
It is very difficult to predict the future in this area, but I would prefer if AI development goes less to humanoid forms and more towards systems that are different from us because then we can work better with them. If we do not try to mimic human behaviour, psychology and appearance then I think we can find more useful forms of computer intelligence that can complement us. The human brain is amazing but also very limited, we cannot deal with large amount of data, whereas computers can accurately deal with all the information we give them. There is potential for forms of intelligence that go way beyond of what we are capable of.
What do you think about machines gaining consciousness? Do you think it is possible in the future?
A: In theory yes. These are very deep philosophical questions, for example I believe that humans are also machines, biological machines that follow physical laws just like computers and still have consciousness. So in theory machines can also have consciousness. However it is very difficult to say anything about machine consciousness when we do not know much about human consciousness either.
V: One side of me would be really interested in that becoming a reality. Until very recently I thought it is possible that we will crack the code, and as technology is evolving it is possible that we get there. However it is also very doubtful, and again we do not know much about what human consciousness or intelligence really is.
What is the concept of the performance AI am here, that debuts in October at Trafó?
V: It is now evolving, but in its core lies the idea that human and machines can be equal creative partners and we can be inspired by technology. Instead of a dystopian and pessimistic view on the machines that are taking over, we wanted to create a different narrative which shows AI as a creative partner that can help us to make our lives better. What we are looking at and researching is our relationship to technology and to the AI. We use these machines to reflect to ourselves, so the more we talk about AI the more we are questioning what is human. We tend to think creativity is something only humans are capable of and now there are machines that at least can simulate this.
What are your plans after the show?
V: We have six more performance dates in Gothenburg at the end of October. Then we would like to tour the show further and continue developing our project.
AI am here - premiere
10/ October/ 2017
TRAFÓ HOUSE OF CONTEMPORARY ARTS
41 Liliom Street, Budapest 1094, Hungary