Artificial Intelligence (AI) has the potential to revolutionize society, but realizing this potential requires more than technical effort. Developing effective AI systems involves balancing specialized knowledge within disciplines with the cross-disciplinary insights needed to address complex challenges. It also requires bridging fundamental research, which offers generalizable principles, with applied research, which ensures solutions are tailored to specific contexts. Crucially, it demands integrating expert perspectives with the lived experiences of communities, creating systems that are equitable and grounded in real-world needs.

Our research lab was established in 2020 as a collaboration between academia and public institutions to address these gaps. This article reflects on five years of the lab’s work, focusing on insights from studying the school choice algorithm in Amsterdam. School choice is a pressing issue in the city, and policymakers have adapted the well-known Deferred Acceptance algorithm to match students to schools. However, this adaptation led to inefficiencies, with students often placed in schools far down their preference list. This illustrated how a theoretically robust approach, even one that famously earned a Nobel Prize, can lose effectiveness when misaligned with local contexts.

We found that addressing this issue required integrating multiple perspectives: theoretical insights, practical considerations from community stakeholders, and interdisciplinary approaches combining quantitative and qualitative methods from AI, Economics, and Psychology. To articulate this, we propose a conceptual model that bridges three key dimensions in AI research: theory and application, science and society, and qualitative and quantitative inquiry. This project underscored a critical lesson: solutions rooted in a single perspective fail to address real-world complexities, and truly impactful research emerges when diverse approaches are synthesized.

We advocate for a shift in AI research that prioritizes flexibility and allows for fluidly navigating between our three proposed dimensions. Our experience suggests that such flexibility ensures AI research genuinely serves and uplifts society.

In this paper, I critically rethink how interdisciplinary approaches to artificial intelligence (AI) are framed, particularly in relation to technology-facilitated gender- and sexuality-based violence (TFGSBV). While collaboration across disciplines is often positioned as a solution to AI-related harms, I argue that these approaches can reproduce the very exclusions they aim to address. Drawing on my work with the Feminist Internet Research Network and dialogues with researchers and activists across the Majority Worlds, I propose liveability – not only as a principle, but as a practice – for interdisciplinary work.

Liveability refers to the conditions that allow individuals and communities to thrive – emotionally, politically and socially – even amid structural violence and technological harm. I treat interdisciplinarity not as a goal in itself, but as an ethical and relational process grounded in care, accountability and situated knowledge. Through feminist and queer perspectives, I examine how AI harms are often abstracted into technical problems, and I advocate for slower, more grounded practices that centre lived experience. Through lessons from Feminist Internet Research Network’s multistakeholder collaborations, I show how introducing liveability into interdisciplinary and multistakeholder work – particularly around TFGSBV – enables alternative, more inclusive forms of collaboration.

Liveability becomes a compass for working across difference, shifting interdisciplinary AI research towards justice, community and collective transformation.

The two cultures described in 1959 by C.P. Snow still haunt the University where he developed his reputation. While the policy legacy of that debate is often reduced to concerns about numeracy skills and the supposed objectivity of quantitative evidence, the class distinctions of sophisticated culture versus lowly trade are again being blurred by social media and generative AI. Working in the other Cambridge, at the MIT AI Lab, Philip Agre advocated critical technical practice. Amid widespread anxiety about the ethics of AI, this is often taken to be an appeal for engineers to be guided by the humanities, and for interdisciplinary centres to focus on the ends rather than the means of AI research. However, in the machine learning era, that agenda can easily become statistically reactive rather than analytically descriptive. In this paper, I argue from the perspective of craft and design, that the most significant distinction, and opportunity for collaborative endeavour, is not between the disciplinary differences between the sciences and humanities, but between interdisciplinary cultures of making and of observational critique.

This Reflection draws from an ongoing collaboration among the three authors, investigating mutual enlistment between the United States Department of Defense and research communities devoted to the advancement of the AI project. We are interested in the intersecting concerns and resonant sensibilities that draw us together – what we argue is the necessary starting point for interdisciplinary thinking – and in the differences that are the collaboration’s generative possibilities. Among the threads that join us are intersecting pathways between academic and commercial positionings within and against the AI project. Each of us has moved between locations in industry and the university, as researchers, practitioners, and students/faculty. Drawing on these experiences, we explore the possibilities that different positionings afford and what they preclude, how we have attempted to navigate these institutions within their frames of reference, and what has drawn us into relations beyond their putative boundaries. Based on Philip Agre’s call for a “critical technical practice” as a path towards more radical shifts in knowledge practices, we consider how we might weave together our biographical trajectories, disciplinary affiliations, political commitments, subjectivities, and skills into what Andrew Barry and Georgina Born name a more “auspicious interdisciplinarity”.

Monsters may frighten but also fascinate us in their weird and unfamiliar ways. As Gramsci once observed, periods of radical transformation are also times of monsters. AI fits the description. It is a bewildering entity, consisting of hard - and software, depending on infrastructures that need huge amounts of energy and water. It defies clear definition, yet seeps into every corner of our lives. Big Tech warns of existential risks while pursuing Artificial General Intelligence, AGI. However, real challenges today lie in how AI threatens to substitute rather than augment human capabilities.

This essay examines the deployment of an AI-based interdisciplinary approach. It has proven spectacularly successful, as exemplified by AlphaFold2's breakthrough in protein folding. This approach operates frictionlessly, combining knowledge domains with remarkable efficiency and speed. It seems to vindicate a technocratic dream of problem-solving without the messiness and time needed for human deliberation. Yet, when this artificial interdisciplinarity enters the social world, it encounters what it seeks to eliminate: friction.

Friction, however, is not an obstacle to overcome but an essential feature of human existence. The physical world requires friction for movement; the social world needs it for creativity, conflict resolution, and meaningful cooperation. Certainly, too much friction can bring havoc, and too little can lead to a standstill. But as AI continues its co-evolutionary trajectory with humanity, we must resist the seductive promise of a frictionless world run by automated efficiency.

Instead, we need to cultivate a humanistic culture of AI interdisciplinarity - one that bridges sciences and humanities while preserving human curiosity, deliberation, and epistemic diversity. Bringing friction back means taking the time to reconsider shared goals, acknowledging conflicts, and maintaining spaces for genuine human creativity. Only by embracing friction can we ensure that AI augments rather than diminishes what makes us human.

This paper investigates a specific culture of interdisciplinarity that has gained traction at the intersection of applied AI and ethics. To address social and ethical harms of AI applications, scholars have suggested importing norms, methodologies and governance frameworks from established disciplines such as the social sciences or medicine. I show how this importation presupposes and endorses a framing of applied AI as a domain separate from established disciplines. Yet, such separation is what initially allows AI practitioners to operate outside those disciplinary norms that have evolved to prevent harms now associated with AI applications. Conversely, if AI applications were understood as situated firmly within these disciplines, practitioners would already be accountable to their norms and standards. Paradoxically, this culture of interdisciplinarity might thus reinforce a problematic disciplinary isolation of applied AI underlying the very ethical issues it seeks to mitigate – fighting symptoms while playing into their cause. In response, I outline three paths forward.

Language AI has become a popular tool across the humanities and social sciences, but it has yet to gain traction in socio-cultural anthropology. Fieldnotes, the core data for anthropologists, present a unique opportunity and challenge for applying language AI to understand diverse human behavior and experience. Anthropological fieldnotes are communicative products in cultural contexts through immersive, extensive and idiosyncratic fieldwork. To read fieldnotes, anthropologists typically engage in qualitative, reflexive interpretations, attuned to local meaning systems and intersubjective encounters. This paper demonstrates a novel synergy, combining anthropological expertise and various AI technologies to analyze natural observation texts about children’s peer-interactions, especially their moral dramas, in the historical context of rural Taiwan during the Cold War. These fieldnotes were collected by the late anthropologists Arthur Wolf and Margery Wolf in the world’s first anthropological study focused on Han Chinese children. Engagement with AI in this project began as methodological cross-fertilization, transforming raw fieldnotes into a text-as-data pipeline and discovering how ethnographic close-reading, machine-learning techniques (e.g., unsupervised topic modeling), transformer models (e.g., S-BERT) and generative models (e.g., GPT) can complement and augment each other’s value. Capitalizing on the systematic nature of Arthur Wolf’s fieldnotes, as well as the special protagonists of these fieldnotes – playful children, the most voracious learners – this paper compares how children, the anthropologist and AI make sense of pretend-fight moral dramas. Such a human–AI hybrid experiment embodies layered-interdisciplinarity at methodological, epistemological and, to some extent, ontological levels, anchored at children’s social cognition. Situated at the intersection of anthropology, digital humanities, developmental science and data science, this work sheds light on the similarities and differences in how machines and humans learn and make sense of morality, and by doing so, critically reflect on the nature of socio-moral intelligence.