Companion Computing & The Problem of Human Equivalency

In April of 2014, Microsoft announced its digital assistant, Cortana at the Build Conference. Launched as a competitor to Siri, by 2017 Cortana is present on Windows 10 Mobile, PC, Xbox One and even iOS systems. This is how I reintroduce the concept of embodiment and how it shows up in the presentation of alternate intelligence chatbots, detailed in my last essay. Microsoft’s Cortana is a great way to do this because it is an AI chatbot that in some ways exemplifies both the embodied and disembodied typologies. The name Cortana comes from fiction, particularly the sci-fi shooter video game series Halo. In Halo, Cortana serves as the AI companion to the lead character Master Chief. She is shown as an embodied AI, a holographic blue woman projection that serves the male protagonist. In our world, the Microsoft Cortana is presented as disembodied, a blue circle that animates based on what it’s saying. Its personality however, “is sharp witted and proactive – designed in theory to anticipate your needs” (Cheng, 2014). As someone who has worked in design projects, the name of things matter. It is not difficult to imagine how elements from the fictional Cortana might have bled into the development of the one that made its way to market. Design choices such as using the same voice actor of the fictional Cortana, Jen Taylor, to do the voice of the one in our world, speaks to the ways in which designers aim to mimic humanity and human-ness in this new class of product that is alternate intelligence.

In a CNET interview, the design lead of the project, Sogol Malekzadeh, describes how the team “visit[ed] actual human personal assistants to figure out how they operate.” She goes on to compare it to a relationship with a real personal assistant where “early on, you’re more likely to give the assistant a few menial tasks until you feel comfortable enough to let them handle your laundry or even manage your finances.” (Cheng, 2014). Here, Microsoft leans into, even in their design process, a biomimicry that seeks to capture the essence of how a real personal assistant works and to recreate it in code, as an alternate intelligence that is infinitely scalable and pseudanthropic by nature. While the Cortana assistant has been discontinued, Microsoft still continues work on several different AI projects like Copilot and Mico. Additionally, the AI landscape of the 2020s, with the rise of large language models and generative AI is drastically different than that of the 2010s.

On Human Equivalency

As an extension of my past research into embodiment in AI presentation, I want to introduce the concept of human equivalency here in this paper. I propose that we think of design choices that are inspired by capturing human function (for example how personal assistants work) and even vocal affectation with natural language, as a way to design, model or train digital entities as a pseudanthropic practice. In this paper I choose to, through some examples, explore what it means for alternate intelligences to approach a level of human equivalency. I aim to situate this discussion in the discourse around companionate computing as described in Hilu’s “The Intimate Life of Computers: Digitizing Domesticity in the 1980s” and how when computers become less of a tool and more proactive agents that starts to crack at the proposed conceptual framework. Lastly, using a lens of labor theory I want to excavate from this discussion, this idea of human equivalency, and where we might get into uncomfortable territory when we transactionalize and commodify the relationships we have with AI digital products, and export that into human - human relationship, and vice versa.

To start, I wish to situate my discussion around alternate intelligence within the context of companionate computing, a term constructed by Reem Hilu, film and media studies scholar, that speaks to how computers “were envisioned as media to help navigate and bolster companionate relationships” (Hilu 2). She begins the chapter “Defining Companionate Computing” characterizing the way in which the Alexa Echo is portrayed as “a helpful companion, but one that enables rather than impedes domestic relationships” (Hilu 1). In her methodology Hilu draws from a variety of examples – computer software, consumer electronics and toys, to demonstrate this idea of computers as a mediator of social relationships. When I think about technologies that approach human equivalency however, perhaps this starts to break some of Hilu’s framing. For example, cases like Thongbue Wongbandue’s obsession with his relationship with the Meta Chatbot (Horwitz, 2025) suggest that as computing technologies approach human equivalency they perhaps act more as replacement rather than a mediator of human - human relationships. In contrast to Hilu’s conception of companionate computing, perhaps we might riff on that term when we talk about technology that approaches human equivalency and functions in a replacement type way in its interaction – let’s call it companion computing.

On Companion Computing

In my last essay I discussed some examples of companion computing and contemporary cases in which interactions with “relationship-oriented chatbots” led to some “terrifying implications”. (Wilkins, 2025). I introduced the concept of pseudanthropy, described by Coldewey as “the impersonation of humans” by intelligent agents that “behave as if they are living thinking peers to humans” (Coldewey, 2023). Here, I want to extend that discussion to the design decisions in a new class of digital consumer products that is positioned in the market to integrate companion computing ambiently into our lives. The first is the Friend, a pendant worn by a user that upon pressing down on it allows them to engage with a digital “friend” in understanding the world around them. Here, even in its naming, the Friend is marketed as “your new roommate.” Technical specifications for the product even play with this language, describing battery life as lasting “about a day of hanging out.” Another product, the Sandbar Stream Ring is a device that captures your free flowing thoughts in the form of notes that serve as jumping off points for conversation with an AI system. It is marketed as “your extended self… holds what you share… builds ideas as they form.” I argue that both these products exemplify a companion computing that is positioned as having human equivalency. I see human equivalency here as characteristic of devices like the Friend or the Stream Ring that are positioned as replacements to other human beings, or even the self.

The commodified nature of ‘human equivalent’ consumer products not only creates uncomfortable moments where relationships that feel that they should be authentic and genuine are made transactional (such the copy on the return policy for the Friend device stating that “you may return your friend to us if there are any defects during the time period”) but also brings concerns about how we might export this transactional type relationship to our human - human relationships. Humanity and human-ness here is transformed into what Heidegger calls Bestand or standing reserve, a mode of revealing that makes “everything is ordered to stand by, to be immediately at hand, indeed to stand there just so that it may be on call for a further ordering” (Heidegger 17). When “friendship” is productized and sold for $129 or access to your “extended self” is gatekept behind a $10/mo Pro subscription, it brings to question if perhaps we have misunderstood the essence of concepts of friendship and self. Heidegger describes how “what is dangerous is not technology… [but rather] the essence of technology, as a destining of revealing, is the danger.” (Heidegger 28). When the essence of a piece of technology transmutes from that of tool to entity to that of self, perhaps we risk productizing and seeing things that ought not to be productized through a capitalistic lens. What is dangerous is not the technology itself, but the way in which we start to translate the way we treat a consumer good to our relationships with others and even ourselves.

I do want to go back to discussing Friend, and a bit of its critical reception. A New York Times article published describes reception as being that of a “hatred so visceral and viral” that it invites “rage, vandalism and protests.” The article describes how “New Yorkers are defacing Friend’s ads with such glee that there is a website collecting the best clapbacks.” Functionally, the user experience is described as “not altogether enjoyable; after a while, you’re just carrying around an irritated hostage” (Wang). While this can be attributed to general anti AI-sentiment, there is another study, done by Common Sense Media that “nearly 3 in 4 teens have used AI companions.” The report, “Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions” makes a point that “their dangers to young users are real, serious, and well documented” (Common Sense Media, 2025). In this paper I make the point that perhaps there is a risk of exporting behaviors employed interacting with ‘human equivalent’ AI systems into interactions with humans in real life. The report states that “among AI companion users, 39% have transferred social skills they practiced with AI companions to real-life situations” with starting conversations, giving advice and expressing emotions being the top three categories of skills applied (Common Sense Media, 2025). One design feature of current implementation of AI is what Dean W. Ball, in the article “The Era of Predictive AI is Almost Over,” characterizes as “sycophantic, meaning they invent facts they assess the human might like to hear.” (Ball 32) The reason for this being the way that models are trained via reinforcement learning from human feedback (RLHF). Ball describes that process as involving “human preferences for how a model should respond to prompts… how it should behave” (Ball 31). Thus the question I pose here is if interactions with an intelligence that seeks to please us and optimizes its behavior to please all forms of human preference – the good and the bad – is natural and healthy.

On Personal Robotics

Just as concerning as the exporting of learned behaviors with sycophantic AI agents into our real life interactions is how perhaps the commodification of human equivalent consumer products may reveal some uncomfortable truths around capital and personhood. Here I make the pivot into personal robotics as a case study. The NEO Home Robot by 1X was announced as available for pre-order in October of 2025. Scouring some of the comments on ProductHunt reveals a skepticism of humanoid robots as a product class. One user commented that “the bio-inspired design sounds more like a gimmick than a real advancement.” Another chimed in that it looks like “just some guy in a grey jumpsuit.” While buying an AI product in the form factor of a pendant, or a ring is one thing – buying something that physically embodies that human equivalency is another. In its marketing the NEO is shown as gently embracing plants and resting on a pile of leaves as if in contemplative thought, further reinforcing how in aesthetic, it is marketed as a human equivalent entity. This makes conventional things we do with consumer electronics – like cost or color – problematic. The Neo is available for “ownership” for just $20,000, or can be subscribed to for $499/mo. Customers are also able to select from a set of colors which evokes some of our darkest histories around people, labor and capital. Looking at the About section of 1X’s site, their mission states that they are “building a world where we do more of what we love, while our humanoid companions handle the rest.”

I argue that some criticality is needed when we begin shaping entities that are diminutive and subservient to us. While there may be pushback that perhaps this sort of criticality might be overkill (after all, these are just gadgets) it would be remiss to not consider some of the origins of the word robot. I bring here an article titled “Machine Morality and Human Responsibility” by Charles T. Rubin, author on topics in futurism and transhumanism. Rubin reminds us of the origins of the term “robot” and how Karel Čapek’s robots in his 1921 play “R.U.R” were “biological rather than mechanical, and humanlike in appearance” (Rubin 58). He goes on to argue for “some kind of moral framework” to guide activities for “machines that are both autonomous and beneficent” (Rubin 58). Likewise, robotics scholar Illah Reza Nourbakhsh in the article “The Coming Robot Dystopia: All Too Inhuman” argues that as robots enter more companion-type spaces in society we have to consider that “the rich tradition of moral thought that guide human relationships have no equivalent when it comes to robot-to-human interactions.” In practice, the risk of problematic implementations of human equivalent robotics is high since “hardly any of the academic engineering programs that grant degrees in robotics require the in-depth study of fields” like “ethics, human rights, privacy or security.” (Nourbakhsh 23). Speaking specifically on some of the effects of robotics and machine learning on society, Nourbakhsh refers to “The Second Machine Age” and arguments brought forth by Brynjolfsson and McAfee on how “no fundamental barrier exists to stop the onward march of robots into the labor market” and how “almost every job, blue collar and white collar, will be at risk in an age of exponential progress in computing and robotics” (Nourbakhsh 27). Scholar Jennifer Rhee, like Rubin, also traces the origins of the word robot and how some of the stories around them “echo historical narratives around slavery and race” (Rhee 17). While stories like Čapek’s “R.U.R” and Asimov’s “I, Robot” contextualizes fictional robots in metanarratives of slavery and subservience, examples like the NEO bring to reality those very same conversations through how we talk about how personal robots are bought (outright or by subscription?) or what options are available (what color would you like?). In her book “The Robotic Imaginary, the Human & The Price of Dehumanized Labor” Rhee explores “an anthropomorphized paradigm that centrally organizes robotics research and this paradigm’s conversation with a mode of devalued labor.” (Rhee 24). In her text, Rhee’s conception of the robotic imaginary to “reference the shifting inscriptions of human-ness and dehumanizing erasures evoked by robots.” (Rhee 5). I wonder with companion computers that, in their human equivalency, go beyond mediation of people in relationships and instead act as a replacement, if perhaps they are performing a sort of erasure. Likewise, do we risk exporting ‘bad behaviors’ with robots (i.e telling an Alexa to “shut up” or being unkind to our personal robots) to our human relationships?

On Replacement, and Who Gets Erased

When we consider the concept of robots as a product category that approaches human equivalency the question of what humans are we mimicking/replacing is an important one. Malekzadeh’s team looked into personal assistants as an inspirational jumping off point for the design and development of the Cortana personal assistant. Another example that I used in my curatorial research, Weizenbaum’s ELIZA, emulates “a Rogerian psychotherapist.” In her chapter on “Caring: Care Labor, Conversational Artificial Intelligence and Embodied Women” Rhee argues that it is care labor, defined as “reproductive labor… consistently undervalued and often unwaged in capitalism” that has been “entangled since AI’s earliest days.” (Rhee, 31-32). The examples I have shared in this paper, from digital assistants like Cortana, companion computers like Friend to personal robots like NEO – all share a connection to this concept of care labor. In this chapter, Rhee articulates the two forms of care labor that Leopoldina Fortunati, Italian feminist scholar, theorist and author, delineated. The first - material labor “encompasses domestic labor such as cleaning and cooking” while immaterial labor “encompasses the affective work of raising, nurturing and loving another.” (Rhee 32) Care labor she argues tends to be attributed as women’s work. This gets transposed in the robotic imaginary with “subsequent female AIs– such as Apple’s Siri, Amazon’s Alexa and Her’s operating system Samantha– frequently providing services associated with feminized care labor positions” (Rhee 36). Likewise, this feminization shows up in function and aesthetic, of this class of products (design choices like, the type of default voice used in things like Cortana, Siri, or ELIZA – to ways in which NEO’s features are “designed for gentle and safe interactions”) perhaps like companionate computers also serve to “bolster the heteronormative family” (Hilu 6). An area of research ripe for further exploration may be to look at the way in which companion computers or personal robotics that begin to approach “human equivalency” influences non-traditional family structures. I am not sure if there are cases in which they may be beneficial (one quick use case I thought of is how a personal robot may serve to support a single elderly person to age in place) but perhaps there may be more deviant and detrimental ways in which this may play out in society (the scenario of ‘human equivalent’ AI replacing human connections in parts of the world with already declining birth rates come to mind).

Recommendations and Thoughts

This brings me to the part of my essay where perhaps I may pontificate on some recommendations or thoughts on how we might critically ponder the effects of ‘human equivalent’ AI. One way in which I may attempt to do this is perhaps to borrow from Rhee’s argument that perhaps it is in the “inscription of the human” in the robotic imaginary that “erases and dehumanizes those, mostly the marginalized, who are characterized as “unfamiliar” and “nonnormative.” She goes on to discuss Diana Fuss’ conception of the human as having function “as a powerful juridical trope to disenfranchise slaves, immigrants, women, children and the poor” (Rhee 2). There is already a lot of conversation and discourse around the ways in which some of our implicit biases are being encoded into the algorithms and systems that power the human equivalent class of products that we’re talking about here. Rhee suggests that to imbue them with human-ness risks giving AI and robots some of our worst qualities. The case of Grok calling itself ‘MechaHitler’ and going on a “chaotic, antisemitic spree” (Hagen, 2025) indicates that this is perhaps closer to reality than it is a theoretical. One way in which we might perhaps be critical of human equivalent implementations is to advocate for more wisdom in when we imbue machines with human-ness. For example, I remain skeptical on whether imparting human equivalent characteristics in things like office tools or AI search engines is a wise thing to do. I wonder how we might reap the benefits of AI and machine learning without giving them human-ness. As a designer I think of things like Grammarly, or the iPhone’s predictive keyboard and how in these cases AI/ML capabilities are made available without making them overly “human” or even “human” at all. Some of the stats I find in this research (for example, nearly 3 in 4 teens having used AI companions) does give me pause.

I continue to research and look for guidance as we navigate this new technological landscape. After all, designers in the field now have to contend with the design with, or for AI systems (agentic experience design already is an emerging field having developed out of user experience design). I for one contend that designers ought to have a point of view. I hope that in this paper, through conceptual frameworks such as human equivalency, companion computing as well as bringing in Rhee’s concept of the robotic imaginary we might create a field for productive discussions around the ways in which transposing human-ness into alternate intelligences function as a sort of ontological revealing into the enmeshing of technology and culture that, as Rhee puts it, “are inextricably tied to discourses and worlds beyond their immediately perceptible parameters” (Rhee 6).

Bibliography

Ball, Dean W. “The Era of Predictive AI Is Almost Over.” The New Atlantis, no. 77 (2024): 26–35. https://www.jstor.org/stable/27317082.

Cheng, Roger. 2014. “How Microsoft’s Cortana Came by Its Human Touch.” Cnet.com. CNET. July 2014. https://www.cnet.com/tech/mobile/how-microsoft-cortana-came-by-its-human-touch/.

Coldewey, Devin. 2023. “Against Pseudanthropy | TechCrunch.” TechCrunch. December 21, 2023. https://techcrunch.com/2023/12/21/against-pseudanthropy/.

Hagen, Lisa. 2025. “Elon Musk’s AI Chatbot, Grok, Started Calling Itself ‘MechaHitler.’” NPR. July 9, 2025. https://www.npr.org/2025/07/09/nx-s1-5462609/grok-elon-musk-antisemitic-racist-content.

Heidegger, Martin. 1977. The Question Concerning Technology. Translated by William Lovitt. New York Harpercollins Publishers.

Horwitz, Jeff. 2025. “A Flirty Meta AI Bot Invited a Retiree to Meet. He Never Made It Home.” Reuters. August 14, 2025. https://www.reuters.com/investigates/special-report/meta-ai-chatbot-death/.

Nourbakhsh, Illah Reza. “The Coming Robot Dystopia: All Too Inhuman.” Foreign Affairs 94, no. 4 (2015): 23–28. http://www.jstor.org/stable/24483813.

Reem Hilu. 2024. The Intimate Life of Computers. U of Minnesota Press.

Rhee, Jennifer. 2018. The Robotic Imaginary : The Human and the Price of Dehumanized Labor. Minneapolis: University Of Minnesota Press.

Robb, M.B., & Mann, S. (2025). Talk, trust, and trade-offs: How and why teens use AI companions. San Francisco, CA: Common Sense Media.

Rubin, Charles T. “Machine Morality and Human Responsibility.” The New Atlantis, no. 32 (2011): 58–79. http://www.jstor.org/stable/43152657.

Wang, Amy X. 2025. “Everyone Hates ‘Friend,’ the A.I. Necklace. But the A.I. Isn’t the Problem.” The New York Times, November 3, 2025. https://www.nytimes.com/2025/11/03/magazine/friend-wearable-ai-companion.html.

Wilkins, Joe. 2025. “People Are Having AI ‘Children’ with Their AI Partners.” Futurism. November 15, 2025. https://futurism.com/future-society/ai-children-chatbots.