What tracks do we leave behind with technology?

Robert Glaser

Christine Graf

Dieser Artikel ist auch auf Deutsch verfügbar

A gun is fired. Instead of the bullet leaving the barrel however, the whole cartridge whooshes back into the magazine. The clouds of dust vanish into thin air. Fragments hurtle towards the bullet hole in the wall, closing it up to an untouched surface. As if nothing had ever happened. The scene in Christopher Nolan’s film TENET shows the impossible: inverse entropy, in which disorder gives rise to order and the effect cancels out the cause.

Sadly, we all know that reality looks quite different. A bullet once shot will hit a target. An egg that has already been whisked into a scrambled egg can no longer be separated into yolks and whites. A technology once developed will also be employed. Our actions do have an effect. We leave behind marks and tracks, even when we make decisions that seem small and that solve things quickly. A fact that we are often unaware of in the hustle and bustle of day-to-day life, in the so-called “heat of the moment.”

It’s not easy changing the perspective from the inside out. We simply can’t distance ourselves.

How art encourages us “to start looking at the world differently”

The “view from the outside” was one of the main reasons why we kicked off our collaboration with new media art around three years ago. The idea behind our close cooperation with artists is that they let us see our work from a new light, whereas we throw our technical expertise and experience into the mix. As such, an exceptional art project that inspires and makes us reflect and learn tremendously from its realization is created each and every year. Teaching us not only about technology but also about ourselves. On top of all this, the work of art also provides our communication media with a fresh, distinctive identity every year, all without undermining its level of recognizability.

While the artwork “Vertigo in The Face of the Infinite” from the INNOQ Digital Art Edition 01 tackled the flood of information inundating our Internet culture, Edition 02 was completed together with the Studio Waltz Binaire and addresses another current topic: artificial intelligence.

The Berlin-based studio has dealt with the subject of design and artificial intelligence and how to visualize it for quite a while now. At INNOQ, we are also increasingly delving into AI and machine learning, concentrating however on how to put them to good use for our customers and in our projects. Excellent conditions for sparking a lively discussion.

The journey is more intriguing than the destination

We started by asking ourselves: What actually happens with the technology we bring into being? And how can we visualize that?

At first, this idea doesn’t sound original at all: Let’s try making an intelligent humanoid character learn to move. Or to put it more precisely, the intelligent agent should complete a specific goal in an artificial space by independently selecting various possibilities. What happened then brings us right back to the concept of entropy previously mentioned at the beginning of this article.

To clarify, the term “entropy” is originally derived from the second law of thermodynamics. It is frequently equated with – and sometimes also oversimplified as – “disorder.” Or to cite the words of astrophysicist and science journalist Harald Lesch:

Despite the intelligent agent having arms and legs and looking like a human being, the exciting thing was that it could and did also choose from countless options of movement. At times, this was strange and grotesque, and at other times, awkward, but always surprising.

We started a learning process using our technology. We gave this intelligent agent specific skills, but we had absolutely no control over how and in what manner it uses them. Did we create an entropy monster? Or more importantly, are we creating such monsters in the course of daily projects? And if so, is that a bad thing? Or are there lessons for us to learn from this?

In order to answer our question(s) and to visualize them, we soon realized that it was not the intelligent agent itself that was fascinating but instead its unusual patterns of movement. The artwork’s title was born:

NEUROEVOLUTIONARY TRAILS

The first part of the title refers to the principle behind how artificial intelligence learns on its own. The AI learns to navigate a space independently using neural networks or basic neuroevolutionary algorithms. The challenge for Waltz Binaire was now to take all the geometric data that had been gathered and to transform it into a sophisticated visualization. The decision was based on three forms of representation: SURFACES resembles a cloth-like, smooth-flowing surface; SPLINES has lines that flow through the space like waves to form silhouettes and contours; ELEMENTS looks like a three-dimensional topographical representation of tracks.

Don’t fear the entropy monster

The learning process of our intelligent agent, who we lovingly called the “entropy monster” in-house, is also referred to as “reinforcement learning.” In a nutshell: “Solve a problem with what’s available to you.” To be honest, it reminded us a lot of our daily work. Isn’t it much the same way in agile software development? Only as a project moves forward do you realize what different twists, turns, or detours are required to ultimately achieve an objective. You could almost say that we are that intelligent agent!

We too have specific tasks and a certain repertoire, or in other words, a number of ways in which to solve them. This includes, for example, existing technologies and our creativity (in order to use what is available as sensibly as possible or to develop something completely new). Other times, we simply get lucky. Just like our intelligent agent, a huge amount of entropy is generated once our projects are initiated. But are they also entropy monsters for that reason? On the contrary, it’s more like they harness or profit from entropy.

Now we’ve come full circle to the tracks that our work and the technologies we have developed unquestionably leave behind. We make decisions regarding technology on almost a daily basis. We should therefore also ask ourselves every day whether this technology creates value or whether it is just an end in itself. Are we only using machine learning because it sounds cool or because it really helps our customers? And as we all know, even minor decisions can have major consequences. It is therefore up to us to ensure that the drawbacks of our technology are minimized, starting with the selection of software: Is it future-proof? Does it save money and time? Can it be used on an international scale? Is it scalable, barrier-free, etc.?

Every decision needs to be looked at from different angles as well as in various contexts. After all, technology can be useful in one sphere, but then again even harmful in another. And then we won’t be able to retrieve the bullet back into the barrel. The hole will stay in the wall. That’s the responsibility we must bear as the creators of technology. So, let’s all work to ensure the tracks we leave behind with our technology are just as attractive as our NEUROEVOLUTIONARY TRAILS.