Updated: Sep 30, 2018
Berlin based media artist Marcel Schwittlick is constantly discovering new ways to bring tech into his artistic practice. His clear ability to conceptually capture the essence of an idea and communicate it to an audience using contemporary tech hacks, makes him a role model of how engineers can use their knowledge and skills to implement artistic ideas using technology.
Schwittlick recently moved in to his new studio in Berlin, and as the dust settled down after unpacking boxes, we got a chance to talk a bit about art and technology.
You work on different tech projects in Berlin, do you feel there is a connection between the art and tech industry?
Schwittlick: I guess there are connections. Have you heard of the joking expression that artists are lately the R&D department of design and advertisement agencies? Especially in the digital and technology rooted arts this is often the case, copying artists work for their clients commissions.
Often enough though, agencies are honest to ask the artists to create something for their clients, giving a chance to find a compromise and include the artist in the equation, providing an opportunity to earn some money, and big technology companies often offer residencies for artists to use their tools. For example Autodesk and Spotify. I've heard of many cases that a company sponsored a project because their technology was used in the process. For the artist it may not be a big deal to add some credits, for the company its surprising publicity and will gladly reward artists for using their products in interesting projects. Arduino and Google do that.
In your works you tend to use technologies and methods that relate to AI, how great of a real potential do you see in AI, and how much of it is just hype by the media?
Honestly, my understanding of most recent AI/Machine Learning (ML) research is very limited. It simply way over my expertise and area of interest as well. I know for a fact that there is more research being done for one person to have read even 10% of the papers that are released on these topics. I know how many people need to work hard to even tackle small domains of problems that they try to find a machine learning solution for.
Most interesting about all this for me is that the topics that researchers are working on are extremely inspiring. Literally just the topics and areas they are working on. For example simple and applicable tools like style transfer, image scale up and porn detection. Generative Adversarial Networks is an interesting topic.
I liked the image of artificial intelligence that has been painted by the media, but rather out of comedic and ironic reasons. The promises by companies to use AI usually sound a bit too promising in relation to the results that I am noticing. I kind of like the marketing video of the company that I am working with, Micropsi Industries.
For my artistic work, that tends to see the use of the technology in less positivist way, rather showing it as a bad example. Showing the failure of it to some degree close to it. Technically I am working with similar approaches like commercial companies do, though I am not interested in creating a usable and useful product, rather to experiment and play with possibilities.
Last the most important aspect, the technology is mainly not the problem in creating AI systems. It's finding good training data. That means A LOT of data, unbiased and thorough. Having data is the key. And the owners of the data know that it's worth pure money and usually don't make their data openly accessibly. This here is a good approach to open data.
What are a few example AI technologies you used recently in your art projects, and are they being used by the tech industry to develop products for the future?
I've used mainly style transfer for visual work and recurrent neural networks and Long sort-term memory systems for text generation. And self trained word2vec models for text classification and augmented writing.
Style transfer has been used by snapchat/facebook/instagram etc for filters on peoples photos, but they got out of trend pretty quickly. word2vec systems are being used by some text analysis businesses like Noduslabs. They offer people to get a grasp of all their text data, to model topics, see trends in their writing, have a birds-eye perspective on a lot of text. I am fascinated by this and have developed my own system for my own book library.
Many UX and design professionals in the tech industry tend to position themselves on the intersection of tech and art, do you see it also with engineers & developers?
I guess you mean whether developers and engineers position themselves on the art and tech spectrum. Many do indeed. Have you heard of artists&engineers, a London agency that connect artists with the necessary engineers that have the open state of mind it takes to make unconventional and big projects successful.
Job titles like the Creative Developer or Creative Technologist is something that I have come across very often. I like the interdisciplinary nature of these titles and professions.
This artist introduction and interview was curated by Eyal Zucker for Artqol