Artistic reflection

In 2017, we (DiPisaStasinski i.e., Alessandra Di Pisa & Robert Stasinski) were offered the opportunity to create an artwork in collaboration with KTH Royal Institute of Technology and IBM, utilizing their AI system Watson. The four-year project resulted in Being Unthinkable… (2017-2021), an interactive, robotic AI sculpture. 

One of the great successes of IBM Watson was its win in the quiz show Jeopardy! in 2011, marking a turning point for natural language processing (NLP), demonstrating the apparently intelligent potential of machines. Our project, Being Unthinkable…, draws on this legacy while it also investigates the prospects of machine intelligence. 

The first part of the title “being unthinkable” is a riff on the IBM slogan “Think” from the early 1900s, and an open-ended examination of what is conceivable and what is feasible – while the second part of the title gets AI-generated in real time, responding to audience interaction, as a testimony of its capacity to improvise.

Our creative process with Being Unthinkable… was shaped as much by the promises offered by corporate software as by its inability to fulfill them.1 Paradoxically, it was precisely these gaps between expectation and reality that fuelled our experimentation. Despite the limitations of early AI tools, we considered the project a great success and this work came to be our gateway into the world of artificial intelligence at a time when no such tools were yet available for personal use (if you weren’t a researcher or professional coder).

By the end of our collaboration with IBM, we were introduced to researchers at Linköping university interested in the intersection of art and artificial intelligence, and together with them – Jonas Unger, Jonas Löwgren and Gabriel Eilertsen – we initiated transdisciplinary seminars and workshops, meeting approximately every three weeks for almost a year, soon joined by Apostolia Tsirikoglou.

As researchers in computer graphics and image processing, Unger, Eilertsen and Tsirikoglou focused on machine learning for image manipulation and creation, which became the main topic of our dialogue. We worked with complex systems that could be trained to produce a desired output (remember, 2020-2021 was right before generative AI emerged as an accessible tool for the whole world). The critical issue in the process was our refutation to use human-made data as training material. While working with Being Unthinkable… we had critically rejected the humanoid as a visual representation of the AI, and now we aimed at further interrogating the otherness of AI.

To give AI a corpus was vital to us. Although AI may seem abstract, existing merely in the cloud, it is in fact highly physical, dependent on hardware and the extraction of material resources. Yet deciding on its physical attributes would, of course, constrain the possible outcomes of its generated images – a challenge we had to navigate. Consequently, we embarked on a process that evolved bodily forms while simultaneously producing traces used to develop the cognitive skills of the image-creating creature, in tandem with its physical growth. Evolutionary biologist Rike Stelkens became a vital member of our discursive collective, providing a solid foundation as we explored evolutionary processes, shaping both the corpus and the cognitive abilities of the artificial being. 

Thus far, the process had focused entirely on the organism and its becoming, explored speculatively within faux laboratory setups. Yet the time had come to move the organism beyond this insular space, to situate it within a different milieu. This transition marked a shift in focus from internal development to relational engagement: the creature was no longer solely an object of our experimental design, but an active participant in a broader ecosystem of perception and response. Its cognitive abilities, previously constrained within the laboratory’s logic, were now challenged, and it failed to perform as expected – much like a biological organism suddenly removed from the environment of its origin, struggling to adapt.

The whole evolutionary process at the core of our project seemed to fail us just as the work was about to be publicly presented, due to what could be conceived as a technological failure of the original objectives. In response to this failure, the organism had to shapeshift. This technological failure, in turn, gave us the opportunity to explore yet another evolutionary concept, known as punctuated equilibrium.

Having initiated the research project with an emphasis on the evolutionary integration of body and mind, we overlooked a fundamental condition: that any organism – biological or not – exists only in constitutive dependence upon its environment. In neglecting this relational dimension, we failed to fully apprehend the significance of the entanglement of body, mind, and environment. This oversight, however, became a productive point of departure, enabling us to pursue a deeper inquiry into the symbiotic relations that constitute a technoecology – an exploration made possible through the support of the artistic research grant awarded by the Swedish Research Council.2