Identifying, Deceiving, Protecting and Hunting: What Fictional Machines and Humans Do with Machine Vision Technologies

This presentation explores the cultural imaginaries of machine vision as it is portrayed in contemporary science fiction, digital art and videogames. How are the relationships between humans and machines imagined in fictional situations and aesthetic contexts where machine vision technologies occur?

We define machine vision as the registration, analysis and generation of visual data by machines, and include technologies such as facial recognition, optical implants, drone surveillance cameras and holograms in this. The project team has selected 335 creative works, primarily games, novels, movies, TV shows and artworks. We have entered structured interpretations of each work in a database (http://machine-vision.no/knowledgebase). We have identified situations in each work where machine vision technologies are used or represented. For each situation, we identify the main actors involved, and specify which actions each actor takes. For instance, the scene in Minority Report where eyedentiscan spider-bots scan Anderton's newly-replaced retina to identify him involves the character John Anderton, who is evading and deceiving the machine vision technologies. The machine vision technologies biometrics and unmanned ground vehicles (the "spyders" or spider-like bots that crawl through the apartment building to find Anderton) are searching, identifying and deceived.

Many contemporary games and narratives have key characters who are machines, cyborgs, robots or AIs, ranging from the Terminator to contemporary figures like the emotionally awkward SecUnit in Martha Wells' Murderbot novels, or the android player-characters in games like Detroit and Nier: Automata. Our analysis of 36 such characters finds that their actions in relation to machine vision can be grouped around three key action verbs: analysing, searching and watching. Interestingly, the watching cluster has two distinct sides, where one set of related actions seems to cluster around communication and social activities, with verbs like hiding, impersonating, confused and feeling, while the other side shows the passive and uncomfortable ways these machine characters engage with machine vision, as they are disabled, overwhelmed and disoriented. Of course, all these machine characters are imagined by humans, and their very positioning as focalisers, narrators and protagonists in narratives and games tends to lend them human qualities.

The 235 human characters we analysed use machine vision and are affected by machine vision in many different ways. Humans are watched, identified and scanned, and they are scared. The most frequent action taken by humans in relation to machine vision is evading it, but the next more frequent action is to attack using machine vision technologies. There is of course far more nuance in the material than this might suggest, and human characters also use machine vision technologies for activities such as deceiving, embellishing and killing. Our quantitative analysis will be qualified using close readings of excerpts from the works we have analysed.