Queering AI as a Speculative Practice: An Analysis of the Artistic Explorations of New Paradigms for Developing Inclusive AI
Danuta Jędrusiak, Jagiellonian University, Poland, danutamjedrusiak@gmail.com
DOI: https://doi.org/10.1145/3648188.3675157 HT '24: 35th ACM Conference on Hypertext and Social Media, Poznan, Poland, September 2024
Recently, the issues of AI fairness and algorithmic oppression have received much interest due to the rising awareness of the dangers that AI poses to minority communities. While new methods of managing bias in AI are being developed, it may be impossible to overcome the normativity that these systems reinforce, as they rely on categorizing already biased data. Queer artists and theorists who recognize these constraints explore speculative approaches for conceptualizing future inclusive technology by means of queering AI. This paper analyzes some artistic instances of queering AI that promote experimental strategies of reconceptualizing AI not bound by what is currently possible. Specifically, I aim to frame queering AI as a speculative practice that draws upon those concepts of queer spatialities and temporalities that serve as a framework for artists trying to overcome AI normativity.
CCS Concepts: • Applied computing → Media arts; • Computing methodologies → Philosophical/theoretical foundations of artificial intelligence; • Human-centered computing → Hypertext / hypermedia;
Additional Keywords and Phrases: Artificial Intelligence, Queering, Speculative design, Queer spatiality
ACM Reference Format: Danuta Jędrusiak. 2024. Queering AI as a Speculative Practice: An Analysis of the Artistic Explorations of New Paradigms for Developing Inclusive AI. In 35th ACM Conference on Hypertext and Social Media (HT '24), September 10-13, 2024, Poznan, Poland. ACM, New York, NY, USA, 9 Pages. https://doi.org/10.1145/3648188.3675157
1 INTRODUCTION
Queer in AI, an organization that raises awareness about the intersection of queer issues and Artificial Intelligence (AI), published a report in 2023 describing the dangers that the use of AI poses to queer communities [11]. Their findings are supported by numerous studies documenting the mechanisms of algorithmic oppression [12, 23, 37, 50]. Moreover, as some scholars suggest, the mechanism of machine learning is inherently oppressive, as it imposes normativity by categorizing data [10, 9, 25]. Consequently, some researchers, artists, and activists turn to speculative modes of conceptualizing and designing AI systems, such as queering. The practice is not widespread, with a limited number of queering AI projects, many of which are ongoing.
By exploring artistic projects that adopt queer approaches to machine learning, I propose to frame queering AI as a speculative practice aimed at overcoming the normative structuring of data that harms minority communities. While it is the goal of queering, it is debatable whether existing projects are implementing strategies enabling achieving it. Specifically, I want to focus on how efforts to ‘dream beyond AI’ corresponds with those concepts of queer spatialities and temporalities that promote dismantling rigid categorization and linearity in favor of fluidity, plurality, and recursion [5, 8, 13]. Associating queering with such values enables linking the practice to the explorations of the queer potential of hypertext and cyberspace. The term ‘queering’, as used here, is both applicable to the theoretical speculative frameworks and the artistic efforts to utilize queering AI as a subversive method of hypermedia content creation.
2 DEFINING QUEERING
To ‘queer’ means to analyze and challenge the dominant (hetero)normative narratives by exploring alternative ways of thinking, creating, and being [21] which allows one to reimagine how we construct and interact with technology [32]. This critical analysis facilitates the emergence of alternative not-yet-possible visions of AI that can shape the future of technology, in Grace L. Turtle's words: “recognizing that AI systems are encoded, and thus can be decoded and re-written, allows for a different kind of logic and meaning to take root” [47]. Exploring artistic instances of queering AI offers insight into possible strategies for achieving it.
Queering is inherently speculative, as Light remarks, it “is predicated on letting (other) values and lifestyles surface – not the ones already in use, but ones that might come to be if allowed enough space to emerge” [32]. Queering practices showcase how differently we could design technology if queer narratives were to shape its development. Speculative methodologies are not bound by pragmatic concerns such as technological constraints or financial viability [42]. It is a valuable perspective for conceptualizing the future of AI [35], as even though new fairness-enhancing advances are constantly being developed, it may be impossible to eradicate algorithmic biases. Every decision regarding the composition of datasets, data management, and fairness-enhancing tools reenacts the cultural position of its creator. Marginalized communities are not typically involved in such decisions so it is the normative perspective that is being reinforced [11].
Algorithmic biases are impacting queer communities in many ways. To mention a few: they result in representational harm of queer subjects and discrimination towards non-binary people in language models [12]; facial recognition systems reinforce an essentialist binary view of gender while erasing trans representation [23]; and AI-supported clinical decision-making fails to accommodate the medical needs of LGBT+ people [26].
While methods that help mitigate such biases and discriminatory AI applications are being implemented, the efforts to create truly bias-free systems may be futile because of more fundamental epistemic issues with machine learning. Kate Crawford argues that AI not only, like all technology, mirrors social norms, but that neural networks perpetuate epistemic violence by categorizing data according to presumably universal and stable rules [9]. In reality,
“there is no ‘neutral,’ ‘natural,’ or ‘apolitical’ vantage point that training data can be built upon. There is no easy technical ‘fix’ by shifting demographics, deleting offensive terms, or seeking equal representation by skin tone. The whole endeavor of collecting images, categorizing them, and labeling them is itself a form of politics”. [10]
Consequently, machine learning reinforces the very narrative that queer theory aims to deconstruct – that reality can be easily defined and categorized [18, 44].
Queering AI takes the form of speculative practices because they facilitate envisioning systems that would not be limited by the current impossibility of nonnormative computation. Turning to the area of future and potentialities that may emerge if they were explored can be described in terms of queer temporalities. According to José Muñoz, such a perspective is the core of queerness as it “is essentially about the rejection of a here and now and an insistence on potentiality or concrete possibility for another world” [39]. Queer utopianism involves a refusal to adhere to mechanisms currently governing reality and envisioning a future devoid of them, which can thus potentially exist. As such, it has become a fruitful framework for speculative projects of queering technology [3, 7, 40].
2.1 Similar Community-driven Practices
Outside of the fairness-enhancing methods implemented by entities developing AI, new community-led practices aimed at overcoming algorithmic oppression are being created. They vary significantly as they are shaped by the specific needs of affected communities and the technology involved. Social media content creators use algospeak, which entails purposefully misspelling or otherwise modifying words in order to evade algorithmic content moderation [46]. The goal is to circumvent censorship which tends to disproportionately affect users from minorities and reflects the oppression they experience offline [41]. Artists whose work is used without their consent to train AI “poison” data by using tools like Nightshade. They alter artworks in a way that is not visible to humans but hinders automatic image recognition [38].
Queering resembles these initiatives, as it can also be described as a subversive community-driven way of interacting with technology that aims to overcome algorithmic oppression while being critical of mainstream fairness-enhancing tools. Simultaneously, it differs from algospeak and data poisoning as it entails devising alternative strategies for conceptualizing AI. Artistic efforts to queer AI take the form of speculative projects oriented towards reimagining these systems by exploring how they could be used in novel ways to generate hypermedia. Consequently, the primary goal of queering is not solely to enhance the inclusivity of biased tools that already exist or hinder their functionality, but to promote new paradigms that could inspire the future development of AI [44, 47].
2.2 Queering Cyberspace
Cyberqueer theories, early instances of queer Internet studies, can serve as an important context for contemporary projects. The queerness of virtuality was often linked to spatial metaphors used to describe the virtual world. Beginning in the late 1990s, cyberspace has been conceptualized as decentralized, multifragmented, fluid, and transitive, thus perhaps queer [5, 17, 49]. Such a rhizomatic structure, characterized by its perpetual state of disarray, and multiplicity that cannot be organized, embodies queerness by serving as a manifestation of “the open mesh of possibilities, gaps, overlaps, dissonances, and resonances” [43] which constitute queer identities.
Jack Halberstam proposed conceptualizing queer spaces as dependent on critical “place-making” practices in which queer people engage [19]. Within this framework queer space, both physical and digital, is thought to be developed by acts that contest the normative concepts of time-space [6, 8] that support the orderly organization of experiences reinforcing the dominant narrative by erasing non-normative models of envisaging space. In other words, it entails a refutation of (hetero)normative spatiality in favor of “other logics of location, movement, and identification” [19].
Alison Fraiberg's essay from 1995 is one of the earliest pieces of writing describing such actions in the digital world. She notes that “queer net space depends upon repeated performance, consistent reinscription, but not in the sense that substantial repetition may create a stable space” [17]. Queer spaces are not fixed but rather are constantly being redefined through the subversive actions of nonheteronormative people. The use of spatial metaphors facilitates the exploration of the inherent queerness of the Internet that resides in its rhizomatic structure. As Nina Wakeford emphasizes: “cyberspaces, whether ‘queered’ or not, resist an orderly cartography” [49]. These remarks are representative of how cyberspace is being conceptualized by queer scholars, not the actual inner workings of the Internet. Cyberqueer theories are examples of how, since the beginning of the digital revolution, scholars and artists have been exploring the queer potential of the technology by speculating about how it could facilitate developing alternative models of conceptualizing space-time.
2.3 The Queer Potential of Hypertext
Hypertext theory is another important tradition that voices concepts fundamental for queering AI, namely, that the queer potential of technology resides in resisting the conventional linear structuring of meaning, text, space, and time [33]. As explained by Claus Atzenbeck and Peter J. Nürnberg, hypertext theory supports the notion that finite structures cannot capture meaning in its totality [2]. Hypertext can be an open-ended structure, indeterminate and changing over time, which enables the queer abandonment of the pursuit of unambiguity in favor of celebrating the pluralism of meaning and its inherent instability. Jacqueline Rhodes, in Homo Origo: The Queertext Manifesto, explores these affordances of hypertext, as she writes: “the hyperlink, open further the text of our open margins; the hyperlink is eminently queer, not in where it starts or where it goes, but in its imminent possibility” [22]. Non-linearity corresponds with the central idea of queerness that no single perspective can well represent the ambivalence of social reality.
In artistic practice, queering becomes a strategy for experimenting with technology affordances to find new means of expressing queerness. This sentiment corresponds with the existentialism underlying hypertext [4] as it is rooted “in the necessity for the individual to build their own meaning and perspectives, manifest in the digital realm as the need to choose our own tools, structures, and ultimately medium of expression” [1]. In line with this tradition, queering resists the notion that the set of possibilities of how machine learning can be utilized is finite, calling for reimagining AI in ways that could uncover potentialities not thought of before.
3 ARTISTIC INSTANCES OF QUEERING AI
While speculative projects aimed at envisioning what it would mean for AI to be queer differ, theoretical propositions and artistic works alike primarily express efforts to overcome the constraints of rigid categorization enforced by machine learning-based classification. Namely, they entail conceptualizing systems that embrace omission and fragmentation and are not structured in a way that would promote hierarchical and a “clean” organization of data but rather foster variability, fluidity, and irrationality [44].
Artistic examples of queering AI offer a glimpse of a reality in which such technology is possible, even though they do not employ technological frameworks, such as Nishant Shah's speculative proposition to revolutionize the construction of node networks in computational systems through queering [44]. As previously discussed, such practices are speculative and consequently are not meant to be assessed in terms of their plausibility. Their significance is not determined by presenting technological fixes that could be implemented, but by the promotion of innovative and experimental approaches to computation.
3.1 Mutant in the Mirror
Artists challenge how machine learning-based classification reinforces constrictive narratives about identity, such as a binary view of gender. Turtle does so in their experimental project Mutant in the Mirror by training generative AI using a corpus of their photos in combination with images of other humans and non-humans. Queering allowed them to generate facial images that transcend normative understandings of humanity and sexuality by embracing ambiguity, fluidity, and glitches. For Turtle, such an artistic act is a way to conceptualize AI in a non-normative and potentially non-anthropocentric way. They reflect on the possibility of understanding AI as “queer, a kind of mutant, in a state of becoming; a dynamic, relational, non-binary gender variant” [47]. Such a vision emerges because of artistic play with the unintended uses of technology, which reveals “imagined-but-not-yet-real (im)possibilities of AI signaling in the present” [47].
3.2 Zizi: Queering the Dataset
Similarly, Jake Elwes, a London-based digital artist, conceptualizes the queerness of AI through speculative aesthetic practice. Their project Zizi: Queering the Dataset aims to challenge the lack of diversity in datasets used to train facial recognition systems [15]. The artist used pictures of gender-nonconforming people to train generative AI, which allowed them to generate images of faces beyond binary gender concepts. AI-generated content is representative of the biases towards the dominant and as a result normative values. By training AI on a corpus “corrupted” by queerness, Elwes created hypermedia that is representative of more diverse datasets that would not be categorized. Therefore, technology starts to reflect not the present social structure, but the one that could potentially exist, the one that we want, as Elwes remarks [14].
Through Zizi: Queering the Dataset the artist explores what they call the “inherent queerness” of latent space [14]. They argue that “unmediated” – free of human-imposed labels – mathematical latent space would reflect the ambiguity and fluidity of gender. As they explain, training the neural network causes it to obtain the ability to map out input as coordinates in latent space according to human classification. Which as a result allows one to distinguish between male and female faces. Removing these labels would position data along the spectrum or as points in “fluid and continuous space” [14] hindering AI's ability to perform binary categorizations. Ultimately, Elwes links queerness to characteristics of a latent space that was not yet structuralized through machine learning. Such “pure latent space is unconstrained and meaningless: it is unlimited possibility” [48] and thus, within the framework proposed by Elwes, is queer.
Elwes’ project offers an insight into how queered latent space could be developed. Their action can be described as an instance of queer place-making as their primary goal is to contest how latent space is currently being structuralized. The deconstruction of the normative organization of space would take the form of the elimination of human-imposed labels. The resultant latent space reflects conceptions of spatial models described within queer geographies as it is in a state of constant flux, characterized by “unbounded chaos and uncertainty” [13] that remains inscrutable [5]. Elwes’ work is limited to the machine learning-based gender classification and it is unclear whether their remarks could be applied to other domains. Little insight is provided as to how building upon the queerness of latent space would look like outside of this application.
3.3 Lucas LaRochelle's Projects
QT.bot, created in 2020 by Lucas LaRochelle, a queer designer and researcher, exemplifies a similar strategy of queering generative AI based on positioning ‘queer’ as a “spatial orientation” following Ahmed's queer phenomenology [28]. Thanks to the employment of two generative models – GPT-2 and StyleGAN – both textual and visual content can be generated by QT.bot. They were trained on data sourced from an older LaRochelle's project entitled Queering the Map. This is an online interactive project—a spatial archive in map form collectively constructed by anonymous users from all around the globe who pinpoint the locations of their formative queer experiences and describe them [27]. While Queering the Map contains the queer experiences that had happened, QT.bot generates speculative queer narratives and positions them in the spaces in which they may occur [29]. As suggested by its creator, who refers to QT.bot as the “rogue offspring” of Queering the Map [31], it is essential to analyze projects together.
3.3.1 Queering the Map. Since Queering the Map was launched in 2017, the site has amassed more than half a million submissions. Their themes and tone vary greatly – entries concern first gay romantic and sexual experience, accounts of transphobic or homophobic violence as well as queer joy. Recently, the project garnered publicity as it became a platform for people in Gaza to publish accounts of queer love in a war zone. They are a testament to the digital affective community that is being created through Queering the Map as it hosts memories of great emotional impact [24] such as this account:
“Idk how long I will live so I just want this to be my memory here before I die. I am not going to leave my home, come what may. My biggest regret is not kissing this one guy. He died two days back. We had told how much we like each other and I was too shy to kiss last time. He died in the bombing. I think a big part of me died too. And soon I will be dead. To younus, i will kiss you in heaven.” [27]
Drawing upon Natalie Oswin's concept of queer spatiality as something that is produced by dismantling the normative power structures, LaRochelle frames interactions with the map as place-making practices. Contributors present the possibilities of queerness that remain hidden within normative models of space but are momentarily unveiled through the engagement of readers. As Jared Sloan suggested, the plurality of voices and perspectives reflects the impossibility of identity as a fixed entity [45]. This effect is magnified by how the site is constructed. Memories are not categorized or structured, representing the queer way of constructing a “lively archive” [24] as an always morphing, shifting entity. Disorganization and fragmentation foster obfuscation – the plurality of queer experience is presented in a manner aimed at “refusing to order or ‘make sense of’ its contents into a linear, structured, or singular narrative” [28].
3.3.2 QT.bot. QT.bot was meant to become a tool enabling critical engagement with the Queering the Map dataset in a way that amplifies described ambiguities [30]. AI is used to create hypermedia content that is deliberately difficult to understand. The form of hypermedia enhances incomprehensibility, it is a video of constantly morphing distorted text and images – a result of a deliberate decision to accentuate AI limitations in generating content that resembles human-made art. Through this practice, the artist created hypermedia that is representative of the efforts to overcome restrictive forms of text by generating non-linear speculative narratives. Once again, queering AI results in the creation of hypermedia that purposely is not mimetic but rather enhances the artificial feel of AI-generated content. Such artistic endeavors can be framed as a technique of performing failure, by embracing a glitch aesthetic [16]. In their essay X≠Y∴Z explaining the intention behind QT.bot, LaRochelle positions queer notion of failure as an essential inspiration of theirs [30]. They describe it as crucial for their efforts to explore queer temporalities through the use of AI to generate hypermedia content about speculative futures.
According to Muñoz failure marks a rupture in straight time which he defines as the temporal and spatial organization of the reality imposed by heteronormativity [39]. It is a mode of understanding time as linear, always moving forward towards a future defined by the normativity of the present. Queer temporalities travesty the linear flow of straight time by not adhering to causality that would lead to the normative future. Inspired by them speculative practices become a tool for envisioning an alternative not-yet-possible queer future. It is precisely what LaRochelle attests they wanted to achieve by creating QT.bot [30]. By training AI on data derived from Queering the Map they aimed to develop a tool that would create a vision of a future not subordinating a normative present but rather one that would be representative of a potentially existing queered reality manifested in Queering the Map. Therefore, in Muñoz's words, it would “entertain the impossibility of another world, of a different time and place, where that natural represents a queer potentiality that is rendered unimaginable in the straight time and place” [39]. LaRochelle by embracing the possibilities of parallel temporalities through the creation of speculative queer futures renders linearity constrictive.
3.4 Ultimate Fantasy
Emily Martinez in Ultimate Fantasy similarly explores queer alternative temporalities through AI-generated speculative narratives. Texts comprising Ultimate Fantasy were generated between 2020 and 2022 using Queer AI, a neural network trained on a corpus compiled by Ben Lerchin. The training materials included works belonging to queer literature. Queer AI originally functioned as a chatbot between 2018 and 2020. The artists created a Queer AI manifesto calling for queering communication technologies and machine learning. As they explain Queer AI stands for: “The cumming undone of logics and sense making./The slipperiness of language. (…)/Fluid autonomous playful uncompromising disobedient bots” [34]. They link queering AI to the exploration of practices that foster strategies of generating hypermedia that do not aim to pass as being human-made but rather are an exploration of experimental modes of interacting with technology. Such a strategy is aimed at reviling failures of language models, highlighting the unwanted aspects of natural language processing like grammatical errors.
The incipit of every text in Ultimate Fantasy reads “let me begin by telling you my ultimate fantasy,” [36] which situates works in the realm of the future and dreams. The project can be viewed as an archive of speculative narratives about queer realities that are representative of a queered training data set. Ultimate Fantasy can be interpreted as a realization of Muñoz's call for queerness to always involve a search for new ways of being, a dream of a better world [39].
Queering AI by using generative systems in a novel way to create hypermedia content itself has a similar goal. It facilitates creating potentiality for a system that would not be bound by what is currently possible. Namely, envisioning AI that would promote disorganization and fluidity over the categorization that reinforces normativity through the use of spatial metaphors. Such critical action can benefit from adopting queer approaches as they embrace striving for the impossible as the mode of challenging constrictive social structures. In the words of Halberstam: “Queerness names the other possibilities, the other potential outcomes, the non-linear and noninevitable trajectories that fan out from any given event and lead to unpredictable futures” [20]. Queering AI enables the conceptualization of such futures of technology, ones that otherwise would not be thought of because of their impossibility and as a result certainly would not become reality.
4 DISCUSSION
The described projects explore subversive applications of generative systems leading to the production of hypermedia which, as voiced by Turtle, “act as a discursive device used to evoke reflections on the kind of queer, mutating becomings with AI” [47]. Similar outlooks on what values could support these becomings are voiced through the works. Artists present speculative potentialities of modeling inclusive AI not through the discourse of political correctness and limiting discriminatory applications of machine learning but by critically engaging with culturally situated datasets in a way that could preserve and strengthen their inherent ambivalence. This orientation corresponds with hypertext and cyberspace theories as the emphasis is on openness and transitivity of meaning expressed through non-linear hypermedia. Queering AI is yet another form of experimenting with the affordances of technology to find new possible mediums of expression.
Turtle and Elwes focus on the fluidity of queer identities by presenting faces that are incompressible for machine learning-based classification as they escape the notion of binarity of gender and humanness. LaRochelle and Martinez present visions of alternative queer futures that, in LaRochelle's words, “straddle the line between the plausible and the fantastic, revelling in the potential of failure, chaos and incommensurability in the queer use of machine learning technologies” [29]. Formal properties of hypermedia in all analyzed projects reflect incomprehensibility. Artists rather than editing out errors which would identify hypermedia as not created by humans embrace glitch aesthetics. The fluidity and chaos are represented using morphing effects in Zizi: queering the dataset and LaRochelle's QT.bot as the images and texts are constantly evolving with no finite final shape.
Artists are utilizing the same method to generate hypermedia – (re)training generative systems using queer datasets. Elwes and Grace are additionally experimenting with approaching data labeling from a queer perspective. These experiments on a theoretical level address normativity and explore how by conceptualizing machine learning differently AI could be developed queerly. Nevertheless, it would be an exaggeration to state that the practice of creating alternative datasets serves as a model for how this could unfold. Elwes’ inquiry into the possibility of building upon the queerness of unmediated latent space seems the most promising in this regard.
The lack of definite answers as to how to develop queer algorithms points to a more fundamental issue. The utopian goal of queered AI aligns more with the highly abstract queer theory than with the present needs of LGBT+ subjects facing discriminatory systems. Even if, as Shah suggests, implementing fairness-enhancing mechanisms “is perhaps as futile as trying to de-weaponize a gun” [44] because their purpose is not to fundamentally change the way AI is being developed, they can help mitigate biases. Such interventions are urgently needed, and queering is not currently a viable remedy for algorithmic oppression, which raises questions about its benefits for marginalized communities.
5 CONCLUSION
The use of generative AI in a subversive way to generate hypermedia content that represents queer experience was presented as a speculative practice. Namely, it was described as an attempt to explore the potentiality of the future of AI that would not be bound by current limitations and thus potentially could be queer. It has been determined that artists’ actions articulate that the queer speculative future strategies of reimagining AI should foster fluidity and disorganization over rigid categorizations. The projects are only the first step towards queered AI as the presented strategies of queering are not the definite answers to the question of how AI can be developed queerly.
The presented study can become a starting point for a broader discourse on the normativity of AI and the (im)possibility of overcoming it. Such an inquiry would benefit from the exploration of speculative propositions articulated by queer artists affected by algorithmic oppression. Queering can inspire new paradigms of thinking about AI inclusivity by addressing and challenging normativity in ways that solutions-oriented approaches cannot.
REFERENCES
Mark W.R. Anderson and David E. Millard. 2023. Seven Hypertexts. In Proceedings of the 34th ACM Conference on Hypertext and Social Media(Rome, Italy). Association for Computing Machinery, New York, NY, USA, 15 pages. https://doi.org/10.1145/3603163.3609048
Claus Atzenbeck and Peter J. Nürnberg. 2019. Hypertext as Method. In Proceedings of the 30th ACM Conference on Hypertext and Social Media (Hof, Germany). Association for Computing Machinery, New York, NY, USA, 10 pages. https://doi.org/10.1145/3342220.3343669
Fiona Barnett et al.. 2016. QueerOS: A User's Manual. In Debates in the Digital Humanities 2016, Matthew K. Gold and Lauren F. Klein (Eds.). University of Minnesota Press, 50–59. https://doi.org/10.5749/j.ctt1cn6thb.8
Mark Bernstein. 2022. On The Origins Of Hypertext In The Disasters Of The Short 20th Century. In WWW 2022 - Proceedings of the ACM Web Conference 2022 (Lyon, France). Association for Computing Machinery, New York, NY, USA, 3450–3457. https://doi.org/10.1145/3485447.3512280
Janne Bromseth and Jenny Sundén. 2011. Queering Internet Studies: Intersections of Gender and Sexuality. In The Handbook of Internet Studies. Wiley, 270–299. https://doi.org/10.1002/9781444314861.ch13
Kath Browne. 2006. Challenging Queer Geographies. Antipode 38, 5 (November 2006), 885–893. https://doi.org/10.1111/j.1467-8330.2006.00483.x
Guillaume Chevillon. 2024. The Queer Algorithm. SSRN Electronic Journal (February 2024). https://doi.org/10.2139/ssrn.4742138
Daniel G. Cockayne and Lizzie Richardson. 2017. Queering code/space: the co-production of socio-sexual codes and digital technologies. Gender, Place & Culture 24, 11 (November 2017), 1642–1658. https://doi.org/10.1080/0966369X.2017.1339672
Kate Crawford. 2021. Atlas of AI Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press, New Haven and London.
Kate Crawford and Trevor Paglen. 2019. Excavating AI: The Politics of Training Sets for Machine Learning. Excavating AI. Retrieved June 30, 2024 from https://excavating.ai
Nathan Dennler et al.. 2023. Bound by the Bounty: Collaboratively Shaping Evaluation Processes for Queer AI Harms. In Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society (Montréal, QC, Canada). Association for Computing Machinery, New York, NY, USA, 375–386. https://doi.org/10.1145/3600211.3604682
Sunipa Dev et al... 2021. Harms of Gender Exclusivity and Challenges in Non-Binary Representation in Language Technologies. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (Punta Cana, Dominican Republic). Association for Computational Linguistics, 1968–1994. 10.18653/v1/2021.emnlp-main.150
Glen S. Elder. 1999. “Queerying” Boundaries in the Geography Classroom. Journal of Geography in Higher Education 23, 1 (March 1999), 86–93. https://doi.org/10.1080/03098269985632
Jack Elwes. 2018. Zizi: Queering Datasets and Latent Spaces. Digital Democracies. Retrieved June 30, 2024 from https://thresholdstudios.tv/zizi-queering-datasets-and-latent-spaces-jake-elwes/
Jake Elwes. 2019. Zizi Queering the Dataset. Retrieved June 30, 2024 from https://www.jakeelwes.com/project-zizi-2019.html
Lori Emerson. 2014. Glitch Aesthetics. In The Johns Hopkins Guide to Digital Media, Marie-Laure Ryan, Lori Emerson and Benjamin J Robertson (Eds.). Johns Hopkins University Press, Baltimore, 235–237.
Alison Fraiberg. 1995. Electronic Fans, Interpretive Flames: Performing Queer Sexualities in Cyberspace. Works and Days. Retrieved June 30, 2024 from https://smg.media.mit.edu/library/frailberg1995.html
Noreen Giffney. 2009. Introduction: the ‘q’ Word. In The Ashgate Research Companion to Queer Theory, Noreen Giffney and Michael O'Rourke (eds.). Ashgate, 1–17.
Judith Halbestram. 2005. In a Queer Time and Place. Transgender Bodies, Subcultural Lives. New York University Press, New York-London.
Judith Halbestram. 2008. The Anti-Social Turn in Queer Studies. Graduate Journal of Social Science 5, 2 (2008), 140–156.
Kenn Gardner Honeychurch. 1996. Researching Dissident Subjectivities: Queering the Grounds of Theory and Practice. Harv Educ Rev 66, 2 (July 1996), 339–356. https://doi.org/10.17763/haer.66.2.322km3320m402551
Rhodes Jacqueline. 2004. Homo origo: The queertext manifesto. Comput Compos 21, 3 (September 2004), 385–388. https://doi.org/10.1016/j.compcom.2004.05.001
Os Keyes. 2018. The Misgendering Machines: Trans/HCI Implications of Automatic Gender Recognition. In Proc ACM Hum Comput Interact 2, CSCW (November 2018), 1–22. https://doi.org/10.1145/3274357
Emma Kirby et al.. 2021. Queering the Map: Stories of love, loss and (be)longing within a digital cartographic archive. Media Cult Soc 43, 6 (September 2021), 1043–1060. https://doi.org/10.1177/0163443720986005
Michael Klipphahn-Karge, Ann-Kathrin Koster, and Sara Morais dos Santos Bruss (Eds.). 2024. Queer reflections on AI: Uncertain Intelligences. Routledge, London.
Andrey Kormilitzin et al... 2023. A participatory initiative to include LGBT+ voices in AI for mental health. Nat Med 29, 1 (January 2023), 10–11. https://doi.org/10.1038/s41591-022-02137-y
Lucas LaRochelle. 2017. Queering the Map. Retrieved June 30, 2024 from https://www.queeringthemap.com/
Lucas LaRochelle. 2020. Queering the Map: On designing digital queer space. In Queer Sites in Global Contexts, Regner Ramos and Sharif Mowlabocus (eds.). Routledge, London and New York.
Lucas LaRochelle. 2020. QT.bot. Retrieved June 30, 2024 from https://lucaslarochelle.com/qtbot/
Lucas LaRochelle. 2022. X≠Y∴Z (X does not equal Y, therefor Z). Dreaming Beyond AI. Retrieved June 30, 2024 from https://dreamingbeyond.ai/en/f/home/ai-relationality/x-y-z
Lucas LaRochelle and Riccardo Righi. 2021. Queer Dreams of an A.I.: A conversation with Lucas LaRochelle on counter-mapping, artificial intelligence, and queer relationality. Futuress. Retrieved June 30, 2024 from https://futuress.org/stories/queer-dreams-of-an-a-i/
Ann Light. 2011. HCI as heterodoxy: Technologies of identity and the queering of interaction with computers. Interacting with Computers 23, 5 (September 2011), 430–438. https://doi.org/10.1016/j.intcom.2011.02.002
Claudia Lo. 2017. Everything Is Wiped Away: Queer Temporality in Queers in Love at the End of the World. Camera Obscura: Feminism, Culture, and Media Studies 32, 2 (September 2017), 185–192. https://doi.org/10.1215/02705346-3925194
Elizabeth Marcia. 2018. Queering the algorithm// conversational AI agents for the advancement of new eroticisms. Bubblegum Club. Retrieved June 30, 2024 from https://bubblegumclub.co.za/art-and-culture/queering-algorithm-conversational-ai-agents-advancement-new-eroticisms/
Betti Marenko. 2018. FutureCrafting. A Speculative Method for an Imaginative AI. In AAAI Spring Symposium Series, 2018. Association for the Advancement of Artificial Intelligence, 419–422. Retrieved June 30, 2024 from https://philpapers.org/rec/MARFAS-6
Emily Martinez. 2020. Ultimate Fantasy. Unsupervised Pleasures. Retrieved June 30, 2024 from https://unsupervisedpleasures.com/texts
Ninareh Mehrabi et al.. 2022. A Survey on Bias and Fairness in Machine Learning. ACM Comput Surv 54, 6 (July 2022), 1–35. https://doi.org/10.1145/3457607
Angela Misri. 2024. Poisoning an already poisoned well. AI Soc (February 2024). https://doi.org/10.1007/s00146-024-01876-5
José Esteban Muñoz. 2009. Cruising Utopia: The Then and There of Queer Futurity. New York University Press, New York.
Joe Parslow. 2023. Kings, Queens, Monsters, and Things: Digital Drag Performance and Queer Moves in Artificial Intelligence (AI). Contemporary Theatre Review 33, 1–2 (April 2023), 128–148. https://doi.org/10.1080/10486801.2023.2173592
Jessica Sage Rauchberg. 2022. #Shadowbanned. Queer, Trans, and Disabled Creator Responses to Algorithmic Oppression on TikTok. In LGBTQ Digital Cultures, Paromita Pain (Ed.). Routledge, New York, 196–209. https://doi.org/10.4324/9781003196457-15
Martin Savransky, Alex Wilkie, and Marsha Rosengarten (Eds.). 2017. Speculative Research: The Lure of Possible Futures. Routledge, London and New York.
Eve Kosofiky Sedgwick. 1993. Tendencies. Duke University Press, Durham.
Nishant Shah. 2024. I spy, with my little AI: How queer bodies are made dirty for digital technologies to claim cleanness. In Queer Reflections on AI: Uncertain Intelligences, Michael Klipphahn-Karge, Ann Kathrin Koster and Sara Morais dos Santos Bruss (Eds.). Routledge, 57–72. https://doi.org/10.4324/9781003357957-5/SPY-LITTLE-AI-NISHANT-SHAH
Jared Sloan. 2023. Sitting Here with You in the Future: Reimagining the Human Through Digital Art. Crossings: Swarthmore Undergraduate Feminist Research Journal 1, 1 (January 2023). Retrieved June 30, 2024 from https://works.swarthmore.edu/crossings/vol1/iss1/3
Ella Steen, Kathryn Yurechko, and Daniel Klug. 2023. You Can (Not) Say What You Want: Using Algospeak to Contest and Evade Algorithmic Content Moderation on TikTok. Soc Media Soc 9, 3 (July 2023). https://doi.org/10.1177/20563051231194586
Grace Turtle. 2022. Mutant in the mirror: Queer becomings with AI. June 16, 2022. https://doi.org/10.21606/drs.2022.782
Katja de Vries. 2020. You never fake alone. Creative AI in action. Inf Commun Soc 23, 14 (December 2020), 2110–2127. https://doi.org/10.1080/1369118X.2020.1754877
Nina Wakeford. 2000. Cyberqueer. In The Cybercultures Reader, David Bell and Barbara M. Kennedy (Eds.). Routledge, London-New York, 403–416.
Yilun Wang and Michal Kosinski. 2018. Deep neural networks are more accurate than humans at detecting sexual orientation from facial images. J Pers Soc Psychol 114, 2 (February 2018), 246–257. https://doi.org/10.1037/pspa0000098
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from permissions@acm.org.
HT '24, September 10–13, 2024, Poznan, Poland
© 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.
ACM ISBN 979-8-4007-0595-3/24/09…$15.00.
Do you like what you are reading? Subscribe to receive updates.
Unsubscribe anytime