Making non-humans humanlike in literature: the Murderbot case
Magnus Söderlund, 2025
We have heard this many times: artificial intelligence (AI) will replace us in numerous sectors. Often, AI will also appear as humanlike in various ways, and it will allow us to interact directly with it, as in the cases of chatbots and service robots, so it is very likely that we will find ourselves spending more time in such interactions.
An area in which we may expect an increasing humanlike AI content is literary fiction – fictional stories that are character-driven rather than plot-driven and in which the human condition is examined in various ways. What will this be like for us who are readers? Would stories with non-human characters be enjoyable? Will they be meaningful?
It seems as if a captivating fictional story needs to have human characters that we can somehow relate to and perhaps identify with. One way to accomplish this, if the character is non-human, is to make it humanlike. This is challenging, because it demands a model of what it means to be human (and the complete model of this does not yet exist). Nevertheless, good writers of fiction have always had ideas about what it means to be human, and such ideas can be used to create humanness content for non-human characters. Who knows, when we create and consume such stories, particularly when the main protagonists are non-human, we may get additional insights about what it means to be human in a more salient way compared to stories with real humans. Indeed, several writers have begun to explore this: Klara and the Sun by Kazuo Ishiguro and Machines Like Me by Ian McEwan are two examples.
Another and particularly interesting case is The Murderbot Diaries by Martha Wells. It is a series of novellas in which the main character, Murderbot, is a non-human – yet a humanlike non-human – that many readers have found to be captivating. In brief, Murderbot is a Security Unit, a robot that has been developed by a company that provides security services. “I keep things from attacking the clients and try to gently discourage the clients from attacking each other” is how it describes its tasks in All Systems Red. Anyway, it is in this service role that Murderbot gets involved in various adventures comprising both good and bad humans in a multi-planetary future. So let me make an attempt to examine some of the ways in which Murderbot shares characteristics with real humans. These ways, I think, embody universal aspects of the human condition, and they make Murderbot not only relatable but an instrument to explore what it means to be human.
First and foremost, Murderbot has a sense of self. It talks to us with its own voice, in the first person, and it knows that it is a murderbot (“I am a talking weapon”). It is capable of self-reflection, which expresses itself as an inner dialogue with and about itself. This is very human; from time to time, all of us talk to ourselves about our place in the world and the dilemmas involved in what we do. Murderbot, like most of us humans, also understands its own limitations: it can get distracted and it can lose control and it understands that these things are not good. And Murderbot has humor. Its dry wit, sarcasm, and self-deprecating humor are not just coping mechanisms; they serve as a way to make sense of itself and its relationships with humans. Having a self means that you are able to view yourself as a distinct entity in relation to others, and Murderbot knows that it is different from humans. In Rogue Protocol, for example, it tells us that ”humans are so fucking unreliable when it comes to maintaining data”.
A sense of self is necessary for the development of mind, something that we humans typically think that we have more of than other living creatures. A simple but dominant way to conceptualize what it means to have a mind (particularly in studies of how we humans anthropomorphize non-humans such as robots) is to say that mind comprises two overall
capabilities, agency and emotionality. Agency is the ability to set goals and to develop a desire and plans to reach those goals; emotionality is the ability to experience emotions. Murderbot indeed has agency, because it has hacked its internal governor module so it can choose to disobey what it is told to do. Its struggle to make choices independent of external control is a very human endeavor. One can feel similar to Murderbot when it express the kind of anxiety that goes hand in hand with autonomy: are the choices we make really the best choices? And Murderbot experiences a range of emotions—frustration, anxiety, empathy, and surprise are some of them. It can miss people, it can get worried and nervous, and it can feel guilt. Murderbot is also capable of experiencing enjoyment; in fact, it takes every chance to consume entertainment media in ways that resemble how we humans turn to art, literature, music and television series.
Having a mind is a prerequisite for having theory of mind – the capability to attribute mental states to others. And this is another of Murderbot’s capabilities. When it is among humans, it knows who likes who, it knows who is frightened, frustrated and furious, and it knows that humans may identify it as a non-human (and as a rogue murderbot). Some think that the traditional body-mind dualism leads us wrong and that it is our embodied nature that allows us to have mind-related capabilities. In any event, Murderbot has a humanlike body and it is aware of it, too. It knows when its body gets damaged, it can feel cold and warm, and it can feel burning pain from its combat wounds. It is aware of its facial expressions and that they signal emotions, and it understands that such expressions may need to be controlled in the
presence of humans. But there is something bodily-related that Murderbot does not have – it does not have a gender and no “sex-related parts”. This is probably why it is not interested in sex. So, in All Systems Red, we get to know that when there are sex scenes in the entertainment media that Murderbot watches, it fast-forwards.
There is a social aspect, too: Murderbot has a sense of understanding its relations to others. It does understand its fundamental, subservient role as a service provider, which means that it has to travel as cargo on its missions with humans, that it should talk respectfully to humans, and that it will get its brain fried as a punishment if it does not follow orders. Murderbot is basically introverted and reluctant to engage with humans, but it still forms bonds with some humans and various non-humans. That is to say, despite its extraordinary capabilities, so that it can easily outperform humans in several domains, the socially anxious Murderbot struggles with interpersonal interactions, preferring to avoid humans when possible (“human-free vacation” is an expression that it uses). Yet, Murderbot often finds itself compelled to
protecting those under its care. This adds complexity to its character, as it often acts against its own preference for solitude to ensure the safety of others. And of course it adds humanness to its character, because we humans are inherently social beings, often seeking connection and a sense of belonging, but we are also often seeking avoidance in relation to certain others.
On top of that, Murderbot is an outsider in relation to the humans it encounters – a bit like an immigrant or a minority member – which evokes empathy from its clients and from the reader. Its role as an Other most likely enhances identification by those who are in a similar situation. This outsider aspect contributes to making Murderbot convincing as humanlike despite its non-human nature. It is indeed very human to feel pain when one understand that one is different from many others.