On Robots and Designers



Experimental Practical Study of Production and Skills

#StudioIntegrativeDesign #Master-Thesis

For his Master's Thesis, Daniel Nikles chose to explore the concept of artistic education combined with the questions the ready-to-use advancements in Artificial Intelligence (AI) leave us with.

"Speculating on the possibility of complete or partial replacement of humans to robots, many would wonder why, as a designer, i would risk the possibility of replacing my own skilled work with self-learning software that can physically produce my desired aesthetics efficiently and in less time: a designer being replaced by a robot.

Of course, replacing a designer (myself or anyone else) with a robot will effectively aid in production processes of any given design; the designer will customise the robot to produce with his/her specific skill set – the creator's solution to complete a given task. However, to understand my endeavour to replace a designer (myself or anyone else) with a robot, it is of utmost importance to understand the difference between creativity and skill. Creativity is natural and inborn. It is often measured in extent and cannot be honed.

Work on these topics is made possible in recent years due to current technological developments - amidst those two stand out to be prominent. Firstly, industrial robots have made it outside of industrial settings. Advancements in industrial production methods sanction decommissioning out-dated machines due to availability of high-end new technologies; but mechanically a twenty-year-old robot is mostly similar to its modern counterparts.The main change happened in software. This allows resellers to restore these robots and bring them back to the market for a fraction of the original price, making it affordable for small-scale companies. Still, current users are able to form the instructions on high-powered systems and then transfer these files into their functional older robotic systems.

Secondly, as far as technological development is concerned, lately there has been an upsurge in machine learning, a subset of Artificial Intelligence (AI). Instructions on how to build deep learning neural networks are readily available online. Google released their TensorFlow software, built by its 'Google Brain' division (initially intended for internal use), as an open-source software library. Safe to say, looking at the functionalities of these various open-source software and available machines to support them, machine learning has become a tool to create new creative work in the hands of artists and designers.

A foray on sketching

It is very important to understand that while learning to sketch, we learn to control the movement of our arm and hand, as well as the position of our joints. It is seen in competitive Asiatic cultures such as in India and Japan – where various forms of arts constitute an important part of culture and expression, – students as young as four years of age learn to control their hand movements by practicing straight line drawings. For implementing Dhyãna (India) or Zen (Japan), children are often disciplined based on their ability of line drawings. Keeping that in mind, mimicking this human arm resemblance while sketching a line is the most basic and important first step to train or make my robot function is a desired direction of output. In this case the output being my own line drawing skills. In relevance to line drawing it is worth mentioning that between the 1960’s and 1970’s, American visual arts saw a new trend: “less is more” or Minimalism.

I use the word "create" for a trained neural network because it is guessing the most probable continuation from an externally fed starting point: by its own calculations - this is very much similar to a newly trained human illustrator. Therefore my software is allowing my robot to “create” and not simply “produce” what I ask it to. Which technically proves my stand in my hypothesis “(a) robot is not a machine that can create” is partially wrong. If human learning is a process of acquiring knowledge, where our behaviours along with skills, values and ethics are acquired when we process information through our minds to learn – then robots can learn too and posess a different set of behaviours along with skills, values and ethics that they acquire through set software, and process information through neural networks to learn. Just like human learning is taking place through education, personal development or informal/formal training, a robotic learning is taking place through same set of protocols that go with the aspect of learning.

For any given artist, the set question for a gallerist is this: “where did you train?” It is a known fact that this training is all about learning skillset of another artist or art movements and then either replicating the style or reconstructing the style in ones own taste. Everyone has their own interpretation and ways of accepting this sort of training and reconstructing learnt styles in an unexpected way. Thus, through my experiments I have managed to prove that my robot is capable of doing the same, provided I introduce it to a system where the robot can learn.

Therefore, my hypothesis stands corrected as:

A robot is a machine that can create when taught, and also is a machine that can imitate skilled labour. Thus, robots are not just tools to help generate that what the designer or artist would demand of it. But, has the capability to create new by taking old and existing. "


Prof. Armin Blasbichler

, Studio Integrative Design

Debolina Dubois-Bandyopadhyay

, Design theoretician and visual artist