Gendered AI: Gender and Embodiment

The Gendered AI series looks at sci-fi movies and television to see how Hollywood treats AI of different gender presentations. For example, are female AIs given a certain type of body more than male AIs? Are certain AI genders more subservient? What genders are the masters of AI? This particular post is about gender and embodiment. If you haven’t read the series intro, related embodiment distributions, or correlations 101 posts, I recommend you read them first. As always, check out the live Google sheet for the most recent data.

What do we see when we look at the correlations of gender and embodiment? First up, the overly-binary chart, and what it tells us.

I see three big takeaways.

  1. When AI appears indistinguishable from human, it is female significantly more often than male. When AI presents as female, it is much more likely to be embodied as indistinguishable from a human than an anthropomorphic or mechanical robot. Hollywood likes its female-presenting AIs to be human-like.
  2. Anthropomorphic robots are more likely to be male than female. Hollywood likes its male-presenting AIs to be anthropomorphic robots.
  3. If an AI is mechanical, it is more likely to be “other.” (Having no gender, multiple genders, or genderfluid.)

These first two biases make me think of the longstanding male-gaze popular-culture trope that pairs a conventionally-attractive female character with a conventionally-unattractive male. (Called “Ugly Guy Hot Wife” on TV Tropes.)

Image result for walle and eve

Recent research from Denmark hints that these may be the most engaging forms to engage children (and adults?) in the audience: learning outcomes in a study of VR teachers found that girls learn best from a young, female-presenting researcher, and boys learned best when that teacher presented as a drone. The study did not venture a hypothesis as to why this is, or whether this is desirable. These were the only two options tested with the students, so much more work is needed to test what combinations of presentation, embodiment, and superpowers (the drone hovered) are the most effective. And we still have to discuss the ethics and possible long-term effects of such tailoring. But still, interesting in light of this finding.

Left: best teacher embodiment for boys. Right: best teacher embodiment for girls.

Not a surprise

  1. When AI is indistinguishable from human, it is less likely to have a gender other than male or female.
  2. If an AI presents with no gender, it is embodied as a mechanical robot. Little surprise there.
  3. Mechanical robots are more likely to be neither male nor female.


When we look more closely at the numbers, it gets a little weirder. This makes for a very complicated graph, so I’ll use a screen grab from the sheets as the image.

  • Of course we would not expect many socially gendered characters to be indistinguishable from a human, but you’ll note that socially male is much higher than socially female, and that’s because while there are no characters that are both [socially female + indistinguishable from human], there is one tagged [socially male + indistinguishable from human], and that’s Ruk, from Star Trek (the original Series) episode “What are Little Girls Made of?”
  • Bucking other trends toward male-ness, [disembodied + female-voiced] AI are 8 times as likely to appear as disembodied, male-voiced AI, of which there is only one example, JARVIS from the MCU.
    1. FRIDAY from Avengers: Age of Ultron
    2. Coach from Black Mirror’s “Hang the DJ”
    3. Samantha from Her (though she manages to procure a proxy for one awkward scene)
    4. VIKI from I, Robot (though she has a virtual face)
    5. Gipsy Danger, Pacific Rim
    6. Sibyl, from Psycho-pass: The Movie
    7. Karen from Spider-Man: Homecoming
    8. Axiom from WALL·E

So while the counts involved are single digit, it is a notable difference.


5 thoughts on “Gendered AI: Gender and Embodiment

  1. Chris, I think you’ve interpreted your categories too narrowly for this post, which is bad because the sample sizes become even smaller.

    I suggest that virtual AIs count as having bodies. Both Joi in Blade Runner 2049 and the Doctor in Star Trek Voyager are virtual, but everyone else almost always looks at and interacts with them as if they were real. Interaction also means that architectural AIs should also count as disembodied, and probably vehicular AIs too.

    • Maybe I wasn’t clear on the article, because that’s what I mean with the virtual category. Though I take the point that they’re could be virtual and yet robotic. We don’t see that though. I believe if it’s virtual, it’s indistinguishable from human.

    • The difference between vehicular and architectural is a little tricky. In some cases, like with Alpha60 it’s clear that it’s stuck in place and its interfaces are bound to walls. But sometimes as with MUTHR, there’s a place you have to go to to interact with them. That puts some constraints on the interaction whereas something like KITT can move about the human context, but it’s something you can get in as well.

      I would rather be too detailed in this study than too course, as a courser analysis can always be done by grouping subsets of more-detailed data. The reverse is not true. What would you hope to learn from bigger numbers, though?

  2. I worry that the fine distinctions you are making between types of AI creates very small sample sizes, which increases the risk of finding a correlation / result from random chance alone rather than being something that actually influences the decision making process of writers and directors when casting a role.
    But stats is not really my thing, so if you’re confident that it works, ignore me.

    • This is a very fine concern. With only 300 or so characters, were talking about a very small data set. But the chi-squared test is there in the sheet to check statistical relevance (vs naturally occurring noise) and the studies pass this threshold. And even with small numbers, we have to acknowledge that these stories have an outside influence, so any analysis is better than none, yes?

Leave a Reply