Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Beware of fictional AI narratives

To the Editor — Science fiction (SF) has become a reference point in science communication as well as in the public and media discourse about the ethics, opportunities and risks around artificial intelligence (AI). As a result, AI in SF — science-fictional AI — is considered as part of a larger corpus of ‘AI narratives’ that shape the development of AI, including the research agenda, public acceptance, and political decision making1. From this perspective, films showing AI and robots can be understood as a reflection of our hopes and fears regarding the development and application of these technologies. However, I am concerned that science-fictional AI may be taken too literally as a factual representation of the technology and its possible consequences in the real world, for example that machines will sooner or later attain agency — like HAL 9000 from 2001: A Space Odyssey (1968) — or claim personal rights — like Data from Star Trek. Rather than to technology, films show us our social relationships to each other in the form of a drama. To make the drama work, computer systems and robots must possess human or even super-human capabilities regardless of the technical limitations — after all, HAL 9000 and Data are voiced and acted by real people. Taking the SF representation of conscious and autonomous machines seriously as a critical technology assessment gives a distorted impression of the capabilities of AI in reality.

Credit: Gremlin / E+ / Getty Images

Nevertheless, if science-fictional AI is not realistic, what exactly is it? It is an artistic and narrative means: Over the years, AI has become a common trope in SF, establishing intertextual and cinematographic references to other representations of AI in the genre, showing the latest visual effects and allowing for dramatic turns in the story. Let’s take a look at some modern SF films after 2000: The famous female robot Ava in the 2014 film Ex Machina, for example, was supposed to look artificial, but intentionally not metallic like Maschinenmaria from Metropolis, nor white-plastic like Björk in the music video ‘All is full of Love’2. Rather, Ava needed to look human and attractive enough that ‘she’ could credibly become the love interest of one of the film’s characters and manipulate him. And again, Ava was played by a real human — just like the hologram Joi in the 2017 Blade Runner 2049 or the robot boy David in A.I. from 2001, who all experience human dramas of love or family relationships. In this vein, the humanoid robot Sonny in I, Robot from 2003 was the state of the art in computer animation — not in robotics. All these anthropomorphized artificial beings are a projection canvas for human issues, not robot issues3.

So, first and foremost, SF films should be analysed metaphorically, as a critical commentary on socio-political aspects beyond technology4. In this sense, Ex Machina is about suppression and deception in a “male-dominated libertarian world where women are still seen as window dressing for sales booths”5 and intelligent women like Ava — not robots — are perceived as a threat. Similarly, the sexy hologram Joi in Blade Runner 2049, both as gigantic living billboard and as personal virtual girlfriend, is part of a city “entirely geared towards pornified male sexual desire”6 and thus a normalization of sexism that exists today. Apart from this, Bladerunner’s bioengineered replicants — although not AI in the strict sense — and also the robot labour force in I, Robot are treated de facto as slaves, which demonstrates not only the human primeval fear of enslavement and oppression in a profit-oriented economic system, but also colonial/postcolonial perspectives and the problematic of racial discrimination and Whiteness7.

These depictions of AI serve the purpose of problematizing human issues through AI as a metaphor. If we understand science-fictional AI too literally as a representation of real AI, we not only miss the richness of the SF genre when it comes to robots and machines, we also get distracted from questions concerning AI that are relevant right now. These questions have nothing to do with humanoid robots or conscious machines, but with the implementations of ethical values such as fairness, accountability, privacy and transparency.

References

  1. 1.

    Cave, S. & Dihal, K. Nat. Mach. Intell. 1, 74 (2019).

    Article  Google Scholar 

  2. 2.

    Murphy, M. ‘Ex Machina’ features a new robot for the screen. The New York Times (2 April 2015).

  3. 3.

    Telotte, J. P. Robot Ecology and the Science Fiction Film (Routledge, 2018).

  4. 4.

    Hermann, I. Text Matters 8, 212–228 (2018).

    Article  Google Scholar 

  5. 5.

    Robbins, M. Artificial intelligence: gods, egos and ex machina. The Guardian (25 February 2016).

  6. 6.

    Rose, S. ‘I’ve seen things you people wouldn’t believe’: what Blade Runner 2049’s dystopia tells us about 2017. The Guardian (6 October 2017).

  7. 7.

    Cave, S. & Dihal, K. The whiteness of AI. Philos. Technol. https://doi.org/ghc8r3 (2020).

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to Isabella Hermann.

Ethics declarations

Competing interests

The author declares no competing interests.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Hermann, I. Beware of fictional AI narratives. Nat Mach Intell 2, 654 (2020). https://doi.org/10.1038/s42256-020-00256-0

Download citation

Search

Quick links