Terminator wird uns nicht retten

Neue Technologien wie KI eröffnen viele Chancen, bergen aber auch viele Risiken. Könnten wir nicht einfach den Umgang damit aus Science-Fiction-Filmen lernen, die bereits die großen Themen der KI durchgespielt haben? Leider nicht. Anstatt zu beraten, lenkt Science-Fiction von den aktuell relevanten ethischen und rechtlichen Herausforderungen der KI und den gesellschaftspolitischen Implikationen ab.

Aya Jaff is a successful entrepreneur and coder working with machine learning – which is currently considered the most promising method to build artificial intelligence (AI). When she talks about her job, people would think at first that she creates Hollywood-like robots which might take over the world, as she recently told in a panel discussion. But what she actually does is writing computer programs to optimize very specific tasks.

Fear of takeover by machines is not a German phenomenon. However, Germans seem to be particularly afraid and skeptical of artificial intelligence when compared to other countries. These fears are echoed in the media when illustrating articles on AI with robots from the science-fiction (SF) genre – like the iconic Terminator, the innocent looking Ava from Ex Machina, the humanoid Cylons from Battlestar Galactica – or with similar looking ones. Rightly so, a group of German scientists who research into the social impact of robotics and AI established the Twitter hashtag #notmyrobots inviting everyone to post such unrealistic or misleading visualizations.

Science-fiction-style robots are not about tech, but about us

But why are science-fiction-style robots unrealistic or misleading? Isn’t the genre about future possibilities of scientific discoveries and technological progress? Yes and no. On the one hand, science-fiction surely builds on technology. It is exactly the current context of digitalisation, big data and machine learning which has led to an explosion of SF-films and series dealing with dystopian surveillance systems, robots and AI. From this vantage point SF refers to real world discussions on the possibilities but mostly the dangers of AI^[1] and can be interpreted as technology impact assessment.

But on the other hand, this view misses one central point: SF also uses technology as a means for conveying a certain narrative. The films and series are not about a realistic view of technology, they are about us. First of all, science-fiction-films are “disaster porn”. As early as 1965 Susan Sontag found that SF-films are not about science, but about the extensive disaster as form of art. But even more, the films and series are projections for primeval human desires and fears: we long to create an artificial human-like being and are worried at the same time of being dominated and destroyed by the forces we created. These motifs can be found throughout all kinds of different stories from Mary Shelley’s Frankenstein over E.T.A. Hoffmann’s The Sandman to the Jewish folklore of the Golem. Also, science-fiction critically reflects current social-political issues. In this sense especially the humanoid robot is a “narrative canvas” for the “other”, for discriminated and marginalized (human) beings^[2]; it confronts us with a critical humanism, in which our humanity defines itself within our attitude towards strangers and aliens^[3].

We find all those motifs in the above mentioned examples: Terminator shows us in the light auf a disastrous apocalypse the human desire for a saviour; Ex Machina tells us a feminist inspired process of emancipation; Battlestar Galactica blurs the line between friend or foe where anyone could be an artificial cylon without even knowing it. Yet, in order to serve that kind of stories, AI in SF – be it a robot or not – must become almighty, magical or mystical.

Science-fiction distracts from the real challenges of AI

However, in real life AI is none of this. It is a technical tool already applied in many areas. We use AI in translation, in logistics, in detecting cancer, in recruitment and so on. AI has the potential to make our lives better. One of the downsides is a possible consolidation of discrimination through data bias, since machines are currently learning on the basis of data – which is nothing else than people’s digitalised experiences. This data carries the inequalities of history. If fed into a prediction system, this injustice will be transferred into the future. One of the examples is predictive policing in the US, where exactly those citizens come into the focus of the police who belong to a reference group that was supposedly criminal in the past – mostly Afro-Americans.

So, what we don’t need are anthropomorphized SF-robots with red eyes. What we need is a positive narrative for the future to use the opportunities of AI and to take measures to meet the challenges. This means diversity and a sense for the social context of data among the coders, an informed and attentive society and above all this means bold policy and smart regulation to make sure that our democratic values including the protection of minorities continue to count.

The many different commissions and strategies on AI on European and German level are not only indispensable to talk about the real – not the magical – challenges of AI. They are also crucial to start a broader discourse on AI and the economic, social as well as ethical implications. On this level, the German AI strategy – despite the fact that Germany could spend a lot more money – provides a promising starting point. People must not think of AI as almighty, mystical and uncontrollable force. This only distracts from the fact that technology is the product of our making with the aim to make our lives better.

^[1] Irsigler, Ingo/Orth, Dominik (2018): Zwischen Menschwerdung und Weltherrschaft: Künstliche Intelligenz im Film, in: bpb (ed.): Aus Politik und Zeitgeschichte: Künstliche Intelligenz, 68. Jahrgang, 6–8/2018, p. 39–46.
^[2] L. Meinecke and L. Voss (2018): ‘I Robot, You Unemployed’: Science-Fiction and Robotics in the Media. In J. Engelschalt et al. (Eds.), Schafft Wissen – Gemeinsames und geteiltes Wissen in Wissenschaft und Technik. Proceedings der 2. Tagung des Nachwuchsnetzwerks “INSIST”, München 2016, p. 208.
^[3] Jackson, Patrick Thaddeus (2013): Critical humanism: theory, methodology, and Battlestar Galactica, in: Kiersey, Nicholas J./Neumann, Iver B. (eds.): Battlestar Galactica and International Relations, Routledge, New York, p. 18-36.

Dr. Isabella Hermann ist wissenschaftliche Koordinatorin der interdisziplinären Arbeitsgruppe „Verantwortung: Maschinelles Lernen und Künstliche Intelligenz“ der Berlin-Brandenburgischen Akademie der Wissenschaften.

Titelgrafik: „Terminator – Arnold Schwarzenegger – anaglyph“ von cobravictor auf Flickr, Public Domain