A new mechanical species
In a world that increasingly champions efficiency, immediacy, compatibility, data, and algorithms – in short, a world that champions quantifiable optimisation – the goal to make robots that act like humans is made significantly easier by the fact that humans are learning to act like robots.
If robots are making vast leaps towards ‘human intelligence’ it’s worth questioning what exactly is meant by ‘human’ in this context? Which humans are the models for machines, and by what processes are models chosen? Which characteristics are assimilated, and which are omitted? With questions like this, it is easy to see how ‘human intelligence’ is a fictive construct, and that ‘human’ was already a cyborg to begin with, mutilated by the technologies of power, censorship and representation.
Robots, especially in recent years (assembly-line machines just don’t impress us any more), are discussed as if ‘human-ness’ is the hallmark by which robots are measured. The Turing Test is a series of questions asked by an examiner to a chatbot, and if the chatbot can successfully make the examiner believe they are having a conversation with a human, this supposedly demonstrates that robots have achieved the ability to ‘think’ like a human. The uncanny valley is a pseudo-scientific concept where non-human objects that approach a human likeness, but don’t precisely replicate it, often evoke fear or disgust in humans. Once a robot passes through the uncanny valley, robots supposedly ‘behave’ like a human.
Whilst such theories and tests are great for advancing the field of robotics, they position the robot as something that must live up to humanity, and is premature until the time comes when the robot can act as a mirror image of humanity, successfully blending in with a society known to demonise that which stands out. Human intelligence and behaviour are falsely posited as superior values, which potentially discredit the transformative ways in which robot subjecthood could alter this so-called human intelligence.
The character Dr. Haraway (a satirical nod to the acclaimed cyborg theorist Donna Haraway) in the anime film Ghost in the Shell 2: Innocence (2004) very acutely states that “raising children is the simplest and fastest way to achieving the ancient dream of creating artificial life.” Teaching someone (or something) how to negotiate with their surroundings is the precise definition of programming, and the technophobe is someone who wishes for sanctity in the myth of a self-authenticating ‘natural life’ that programs itself to be ostensibly superior to programming. Is nature, and by extension, the primacy of humanity, not then a technology of subjugation against the artificial, variegated, and, well, technological?
Professor Hiroshi Ishiguro created a robot replica of himself called a ‘Geminoid’ to answer the question of what is a human presence (the Japanese have a word for this – sonzaikan). As Dr. Haraway alludes to, the Geminoid, inasmuch as it is Ishiguro’s sonzaikan, is a form of reproduction. Reproduction is translation, translation of language, of code, of body, from one subject to another. The slight mistranslations and the holes of unconveyable nuances inscripted in the Geminoid are the sources of a derivative adaptation, a new robotic intelligibility. How might this intelligence corrode the language, gestures, etiquettes and habits we take for granted? What transformations will our ‘human’ systems of interaction undergo?
An example of this corrosion is the question of a robot’s gender. Documentaries about humanoid robots often show the presenter addressing an unfamiliar robot as ‘he, or it, or whatever’ (as James May does to one of Sony’s ASIMO robots). In the anime film Time of Eve (2010), the human characters often slip up and refer to their robotic companions as non-living (in Japanese, the verb ‘to be’ is differentiated for living and non-living objects). This at least demonstrates the arbitrary distinction between object (it) and subject (they/she/he) that the humanoid robot disrupts. This is all too familiar for trans* people, who, like humanoid robots, are often not afforded the synchronisation of gender and language, especially in the realm of non-binary identification. Still, this confusion surrounding a robot’s ‘gender’ is useful to highlight the arbitrary boundaries of gender. Furthermore, the gendered robot disrupts cisnormativity because they is essentially asexual in the first place; they has no ‘sex’ to correspond with a gender.
A pair of ‘Actroid’ robots developed by Kokoro in 2011, outwardly appear as a male and a female. However, except for their wigs and clothes, both are physically identical. That such a primitive trick can radically alter our perceptions of the Actroid is indicative of a fascinating period for robotics. In this period, where ‘artificial intelligence’ is in its infancy, crawling on all fours, we will learn how all intelligence is artificial, and how these intelligences hybridise with each other. We will be given the opportunity to break down the boundaries our human fathers built for us.