Prof. Dr. Sebastian Löwe und Marc Engenhart © dai

10 September 2020

"The debate of the expandability of designers have to be pursued more forcefully."

One might think that machine-intelligent systems have already penetrated all areas of our life and have had a trend-setting influence on our working and living environment. And yet there are areas in which the practical applications of future technologies have receded into the background or are only treated shabbily in academic dis­courses. AI-based, application-oriented design is one of these areas.

This is where Prof. Dr. Sebastian Löwe, Professor of Design Management at the Mediadesign Hochschule Berlin, and Marc Engenhart, multi-disciplinary designer, musician and artist from Stuttgart, come in with ”Designing with Artificial Intelligence“. The three-day digital conference from September 17 to 19, 2020, will address and discuss questions and topics arising from the design of, for and with machine learning algorithms. #ki_berlin spoke with the two founders beforehand about the direction of the event, the status quo of design and AI and necessary cooperation between man and machine.

How did the motivation for the project come about, and what do you want to achieve with it?

Marc Engenhart: We met at the Rise of AI in Berlin, an AI conference. There I gave a presentation of SU, a musical AI man-machine installation. I remember that Sebastian and I had a meeting during the conference, and the key was probably that we were the only people with a design perspective in this field. That's where I think we found each other.

Prof. Dr. Sebastian Löwe: That’s right, we met and talked intensively for two hours. At one point it became clear that we were missing the design perspective in the large field of machine learning and that there are a lot of distorted ideas about autonomous design. Keyword: abolition of designers. We wanted to do something about this and later considered writing a monograph together. Marc already had something like that in his drawer as a rough table of contents. But we soon realized that we still had some big gaps in our understanding, that other designers certainly felt the same way and that we wanted to discuss this in an interdisciplinary format. And so the idea of the conference was born. It was also clear to us that we need a strong community of researchers to answer such questions together, and we want to strengthen this community.

Why is it that artificial intelligence and machine learning in the design field are still uncharted territory, especially in academic discussion? Are there barriers that need to be removed, or is it because of the lack of an AI design commu­nity?

Marc: It is not uncharted territory. In the last three years, I remember that at least one or two seminars or workshops per semester have been held at universities on the subject of machine learning in fields such as communication design, interaction design or product design. We have also conducted a workshop on the basis of dai for Sebastian's MD.H. Interest is increasing, but knowledge and application areas are only conveyed in manageable small bites. Of course, and this is how I see it too, the basis is that of design and the design education associated with it. But an under­standing of AI with ML is pressing and wants to develop, since we are already work­ing with assistant tools. The reason for dai was Sebastian's and my need to establish a scientifically high-quality conference that answers questions in a more precise and penetrating way, initiates discussions, has hands-on elements and is open to all designers. Academic discussion needs a place, a format, a culmination point. That is dai. It’s now even digital, with a really large number of front-row seats. Internationally available. For the design community, for young designers and professionals alike. Otherwise, everything goes down in a rush and in the end the machine is a bit faster than we are. dai invites. It is not an informatics congress, but the connection of a design congress with the focus on an understanding of ML. Without fear of contact.

Sebastian: There are still many gaps in scientific research on this topic. Most contri­butions to UX and machine learning come from computer science. Only a few dedi­cated design theorists publish on the topic. In Germany, therefore, there is still no really systematic scientific discussion of the topic of design and AI. I think this is partly due to the breadth and complexity of the topic, which is simply very technical. If you ask designers, it can happen that you get the answer that GANs are the core of machine learning. That is not true. Even simple statistical methods are used in machine learning. It doesn't always have to be deep learning, even though this area has made enormous leaps. The second part of the answer is certainly the current popular discourse on machine learning and design, which anticipates something like the abolition of designers. And yes, Salesforce has just unveiled a fully automated, intelligent, personalized design machine called "Einstein Designer," which really has made significant progress and certainly makes much repetitive design practice obsolete. But our vision is more in the direction of fusion skills and a co-creation of designer and machine that can produce new interesting and exciting results.

You both come from different fields - academic research meets artists and designers. Who is the conference specifically aimed at and how is the response?

Marc: We really complement each other very well. I clearly come from the field of applied design, from practice, with commissioned work for digital and real compa­nies, but I have been teaching at various universities for a number of years now, which has made the academic, scientific research perspective on applied design appealing to me. But my design studio has always offered me the opportunity to experiment, to speculate and this for and in all media. An applied research.

Sebastian: My background is more theoretical. I studied media art at Burg Giebichenstein, but then I did my doctorate and worked as a research assistant in design and am now Professor for Design Management at the MD.H. I was interested in organizing a conference that would bring together the theoretical and practical per­spectives on design with machine intelligence. Marc and I complement each other very well and try to bring this multi-perspective approach to the conference. We have speakers with artistic and design backgrounds, but also people who come from industry, computer science and data science.

Marc: The conference offers all those interested in design a look at the topic, a much closer look at how design work is already being expanded and modulated today with the help of artificially intelligent assistants. We have great speakers from a scientific perspective, as well as people who are very practical and playful with the topic. Researchers, artists, thinkers and designers. This is a unique mix.

Which questions occupy you when it comes to the connection between design and machine intelligent systems? What chances and possibilities do you see?

Marc: For me the machine has always been an assistant in design and artistic tasks. Before my studies I started to write small code elements on my former Commodore C64 and was fascinated that I could print out my texts on a needle plotter by myself as copies. As often as I wanted. It was so simple and so lasting. So the fear of not being able to learn the language of computers disappeared early. Today, machine learning simply offers the great function of developing usable, precise machines which help to train the creativity and intuition of the designer. And they are now beginning to become mechanically intelligent themselves. Designers are being trained to make variations so as to develop an eye for the most suitable, aesthetically appropriate and functional design. This exercise must be carried out throughout a design life. This could be improved and vitalised in the long term. I am fundamentally interested in the question of how we develop these intelligent systems, how they are designed, what they provide, take over and integrate into a man-machine workflow, so that empathy, fun and design quality as well as the responsibility of the effect of the designed systems and products in the design increases. And how we communi­cate this to young students who are the designers of tomorrow. For some time now, I have been exploring the question of transformative creative machine intelligence in more experimental projects which also research this with artistic means. Clearly, the essential opportunity in the wider spectrum is the step towards a new education and knowledge society.

Sebastian: The two questions themselves are very broad. Like Marc, I am just as interested in experimenting. At the moment I am also teaching creative coding to designers at MD.H and we use machine learning functions of the ML5.js library and P5.js. These tools can already do a lot. It's really fun to use them to their full potential and try out which intelligent design machines they can build. But a tool like ML5 also quickly shows its limits when you want to have functions that are not intended to be used in this way. On the other hand, as a theorist I am also interested not only in the abstract order of what we as practitioners try out every day, but also what large cor­porations are developing and universities are researching. I am thinking of a machine learning continuum for designers, which provides orientation and stimulates debate. I am also concerned with the question of how much technological expertise a theorist or designer needs to have in order to successfully complete projects. At the moment, I'm doing practical research with the consulting firm MHP on a human-centred machine learning framework that allows MHP employees to work on machine learn­ing projects in a truly interdisciplinary way. Often there is not such a deep under­standing between the disciplines. Then the data scientist does not know what the designer has done in terms of user research and vice versa. Especially with ML applications it is important to combine the user perspective and the technology per­spective in a meaningful way.

Becoming obsolete is a widespread fear among employees when the discus­sion turns to artificial intelligence. Without the cooperation of man and intelli­gent machine, will it not work in future in the creative field either?

Sebastian: I think so too. Simply automating design is certainly a bit boring in the long run. Many ML experts assume that only the fusion of user and machine skills will produce really new and interesting results. The debate of the expandability of designers have to be pursued more forcefully and it's imperative that we do not immediately give way to every dystopian scenario. Instead, I advocate a systematic look at which skills designers need in dealing with intelligent machines and which agency, in other words which co-determination, they want. I don't want to have to continue doing every task as a designer. I have identified these fusion skills together with Master students in the seminar and have received very interesting reactions. Designers are shown a possible way forward and understand that it's all about cooperation and not about repression.

Marc: Every revolutionary epoch which develops social and economic habits through fundamental change has been associated with the fear of unemployment, necessary reorientation and the question: "Will everything remain like this in the future?”. I like to show young designers a photograph by Robert Brecko Walker. It shows a Linotype typesetting machine from 1959, which accelerated typographic typesetting work via a keyboard interface for the industrial reproduction of books, magazines or newspa­pers. A single employee replaced an entire room of manual typesetters. The appear­ance of the machine could be frightening because of its metallic form and hissing, rattling noises. Combined with the fear of becoming obsolete, the machine was feared by the previous workers. So now I ask the question, what is the difference compared with today? And which of us designers still work as typesetters with lead letters today?

Let's move on to the content: What awaits us at dai 2020 and how much have your plans and your focus changed as a result of the global Covid-19 pan­demic?

Sebastian: Originally we planned to meet in May at MD.H. Not more than 100 experts for a concentrated exchange. We then had to cancel and postpone it. Now we're doing it digitally, like so many others, if only because our American speakers will probably not be able to enter Europe. But we have thought about a lot of new formats, made everything more discursive, and made sure that chance meetings and informal exchange, which are often missing in zoom conferences, are also preserved. After all, that's how Marc and I got to know each other.

Marc: Of course, like many other conferences we were forced to modulate the physi­cal presence into a digital location at short notice. But this has awakened our ambi­tion to structure the conference experience even better in terms of content, incorpo­rate a few surprises and make it more discursive. In addition, the digital location allows many more guests to participate than would have been possible before. I now see this as a great advantage for the first conference.

First and foremost is the digital conference, what are your plans for the next step?

Marc: The conference will be conserved. We are also considering making selected materials available in our dai digital hub and thus offer an extended benefit to all par­ticipants, as this hub can be used again if necessary. Of course this requires a con­tinuation and further internationalisation of the conference. After this year's confer­ence we will start translating the results and our previous findings into a usable medium for designers and theoreticians.

Sebastian: We have a publishing house and are in discussion with them about an introduction to the topic, which systematically presents the above questions of ML and design continuum, fusion skills and practical application not only in experiments but also in teams.

You are both extremely well networked: how do you rate the AI ecosystem in Berlin in general and through design glasses?

Marc: We have huge potential. But at the moment I can only think of five freelancers, small companies and also a few big ones in the Berlin ecosystem who are actively working on the topic. From the winter semester 2020 onwards, two new courses of study will be offered in Berlin, one in Computational Design and the other in the hybrid training programme Computer Science and Design. These are clear signs of a great need.

Sebastian: Some time ago there was an Innovation Camp in Berlin, hosted by the Federal Competence Centre for Cultural and Creative Industries, which had the beautiful name "Design Machine" and was dedicated to the topic of creative indus­tries and AI. This shows me that politics has discovered the topic and sees great po­tential in it. Berlin is an AI hotspot that allows such experiments. It was interesting for the designers to network with the AI companies in Berlin and use their models for a test drive and try out new unknown applications. Established start-ups such as EyeEm were amazed to find that their aesthetic AI can be used to simply build an image recommendation engine. My esteemed colleague Jan-Henning Raff, who will also be speaking digitally at dai, tried this out. So I think we will be seeing exciting things.