Lux Miranda

Talk: The absurd ease of making queer impact in tech
All technology encodes the values of its designers. With few exceptions, the tech in our everyday lives tends to be rigidly designed around heterosexual, cisgender, and monogamous norms. Even basic insights from queer perspectives can majorly revamp the inclusiveness and accessibility of many technologies, and this often even creates benefits for the non-queer along the way. This talk will explore the increasing need for queer epistemologies in tech, how Lux has approached this with her own research at Uppsala, and how you (yes, YOU) can make an impact, too.

Hi, I'm Lux (like the SI unit for lumens per square metre). My pronouns are she/they. I'm a PhD student at the Uppsala Social Robotics Laboratory. My ultimate research goal: If we can make AI and other emerging technologies one with our true needs, we might be able to wield them against social and existential threats and reshape our Earth—and anywhere else we choose to make home—to be more sustainable, fair, and kind for everyone.

Hannah Devinney

Talk: “This work only considers” : Nonbinary Disclaimers in NLP Bias Research Papers
Bias, particularly social and historical biases, has become a significant topic in Natural Language Processing (NLP) research in the last decade. Gender is by far the most popular type of bias to investigate (Gupta et al., 2024), meaning that the way “gender” and “bias” as concepts are (and are not) defined and discussed has serious implications for for the methodology of the field as a whole. Blodgett et al. (2020) find that “bias” is generally ill-defined, resulting in methods which are not well motivated. Devinney et al. (2022) similarly find that “gender” is rarely explicitly theorized or directly discussed, resulting in cisnormative methods and analyses which preclude understanding and mitigating harms directed to trans, gender-nonconforming, and/or nonbinary individuals.

One ‘symptom’ of this is the (increasing) prevalence of “nonbinary disclaimers,” i.e. short acknowledgements in texts which state that gender is ‘in reality’ not binary, even though it is measured, proxied, or otherwise treated as a binary variable in that text. These statements may or may not include a motivation for binary treatment, and are typically relegated to minor ‘commentary’ sections of the text such as limitations, ethical concerns, or future work sections.

Gesina Menz

Talk: Is mathematical modelling really just about maths?
Mathematical modelling, in many fields of application, is often perceived as a purely mathematical exercise, focused on equations, algorithms, and computations. However, this perception ignores ethical and societal implications both within modelling process itself as well as in its outcomes. Within the field of mathematical modelling there is often little to no reflection on the ‘what’, ‘why’ and ‘how' of modelling which can have wide-reaching effects when algorithms trained on biased data sets are implemented or model assumptions are not questioned when using a model in a different context, to name a few examples. Within my talk, I want to investigate these three questions ‘what’, ‘why’ and ‘how’ in more detail and talk about instances where ethical and societal aspects are influential to the models that are being designed.

Gesina (she/her) is a PhD student in Scientific Computing at Uppsala University. She studies how mathematical and computational models can be used to understand interactions within cell populations. Gesina is particularly interested in improving the impact of mathematical models by considering how to interpret the significance of their results and discussing the ethical implications of different modelling approaches.

Erik Lagerstedt

Talk: Care about pluralism and the relation between humans and technology
Designing technology in a human centred way requires an idea of what it means to be human, and my research has partly focused on questions regarding what de facto definition of human is used when studying and developing technology. Academically, my background is partly in cognitive science, but my personal experiences as a largely male coded genderqueer asexual person has meant that I have often not felt included in the models used to describe humans. I have usually been comfortable with the idea that I have simply been an outlier in otherwise good models; models are after all by necessity simplifications. Over time I have, however, come to realise how privileged I am in terms of the relative comfort of my closet, but also the larger consequences of relying on models that keep reinforcing a narrow view of what constitute human experience. With this shift in attitude, I have also found ways to complement reductionist simplifications with perspectives that preserve the mess of the meaningfully real, partly by appreciating struggle as something important and not to be thrown out for the sake of it.

Erik started his academic journey in theoretical physics and cognitive science at Lund University, and recently defended his doctoral thesis in informatics at the University of Skövde. Erik has recently focused his research on perception of agents (such as humans and robots), something that has been shaped by his experiences as a genderqueer asexual person.

Xengie Doan

Talk: Collective Consent for Genomic Data Sharing
Have you ever had a friend share a photo of you without your permission? Or someone in your family took a DNA test and put your information on a public family tree website? Data sharing in a highly connected world can lead to many benefits, but it can also have many unintended, unexplored collective risks. DNA data is a prime example of collective data, because it could impact genetic relatives, genetic minorities, and more. While DNA data sharing can improve health and wellness or connect families, how can all the people impacted have a formalized say in the decision to share data? One of the ways to inform people and safeguard their rights is consent. While collective consent is not new, there are many challenges for a digital, generalized model. This talk will cover the challenge of digital collective consent and solutions from a legal-ethical, data protection, and user-centered perspective to offer a framework for increased transparency, user-centered consent management, and methods with real-world applicability.

Xengie Doan (she/they) is a PhD Candidate at the University of Luxembourg’s Interdisciplinary Centre for Security, Reliability and Trust where they work with Prof. Gabriele Lenzini. Their work focuses on building collective consent for genomic data sharing, bringing together their interests in human-computer interaction, eHealth, and data privacy. Before this, they worked as a computational biologist and data engineer for an open science nonprofit in the USA.