LCAL Conference 2027

Call for Papers   

Latent spaces: probing, interpreting, situating   

The Leverhulme Centre for Algorithmic Life Inaugural Annual Conference

April 14-15, 2027  

Durham University, UK  (in person only)

Scientific Committee: Prof Louise Amoore (Chair), Dr Emilie K Sunde (Deputy Chair), Dr Alex Campolo, Dr Fenna Hoefsloot, Dr Jenny Moran, and Dr Ludovico Rella  

Organising committee: Lizzie Ranson (LCAL Senior Manager), Christian Drury (Centre Coordinator)   

ABOUT THE CONFERENCE  

Leverhulme Centre for Algorithmic Life will host Latent Spaces, an interdisciplinary conference on how AI latent spaces reconfigure knowledge, society, and culture. The conference invites technical, theoretical, artistic, qualitative, quantitative, and practice-based contributions.  

The concept of latent space has become a shorthand for talking about an AI model’s compressed, lower-dimensional, internal representation of its inputs. As is common with Zeitgeist phenomena, latent spaces are characterised by diverse, and to a certain degree, competing discourses. Computer Science’s attention is to the technical. Latent spaces can be interpreted as compact, highly compressed, but lossy numerical representations of information. The weights and parameters set during training determine the overall structure of a model’s latent space(s), which influences what the model outputs. What might the proxies through which latent spaces are made operable (visualizations, interpolations, probes, benchmarks, outputs) teach us about how to research their effects? From the perspective of the Humanities and Social Sciences, the production of latent spaces forces us to question our assumptions about what space can be, and how we might represent it. Latent spaces introduce new cultural, social, and technological ways of knowing and engaging with the world. As the etymology of the term suggests, a latent space is a hidden, non-observable state. This characterisation holds out the promise that compressing data yields concealed or unknown salient features. A latent space is simultaneously a powerful means of making the world of the model, and an opaque form of knowledge that is not directly amenable to human sensibilities and perception. How, then, should we go about researching latent spaces?  

During this conference, we want to examine these emergent phenomena across technical, political, and cultural perspectives. We welcome contributions that probe, interpret and situate how latent spaces are constructed, and their impact on knowledge, society and culture. The generous scope of this call is intended to encourage researchers to engage on the topic of latent spaces from their situated perspective. Our aim is to create space for interdisciplinary collaboration and knowledge-sharing.

There are many variations on how to refer to the ‘space’ of latent space and its related spatialities (e.g. vector space, embedding space, representational space, manifold), but most names refer to it as a ‘space’? What affordances does the use of ‘latent’ and ‘space’ insinuate, and why is this term preferred? What is the historical evolution of the term latent space? How has the term latent space evolved?? How has this evolution changed how we think about latent spaces?  What is the role of translation in understanding latent spaces, both between humans and machines, and between disciplines working on latent spaces?

Where is an AI’s latent space? Latent spaces do not have clear boundaries, from a computer science perspective. If it does not have boundaries or a clear material place, how do we locate it?  While it might be impossible to say where the physical boundaries of latent space are, are we imposing cultural, political or ethical boundaries – either through our imaginaries or through design, regularisation, ablation, architecture search?  What are the infrastructures surrounding latent spaces—from the materiality of their hardware to how these operational spaces are designed?  How do we govern latent spaces? What does it mean to have latent spaces as mediators of lived experience? 

If it is possible to identify or locate a latent space, what sort of ‘space’ are we talking about? And how might we make sense of the ‘space’ of latent spaces? Who decides what latent space is, and how do these narratives shape our perception of the technology? Can we ‘walk’ latent space, as several machine learning papers and artists have suggested? Can we assume that making it visible, or observable, would make it knowable?

What are the ethics and aesthetics of latent spaces? How does latent space operate as a reference of/for cultural output? What new modes of representation or generativity emerge with latent spaces? What new analytical frameworks do we need to analyse the harness and output generated by latent spaces? Can we conceptualise the distance and connections between concepts as data points in latent space? How can we resist latent spaces reproducing discriminatory knowledges and homogenising culture? Are the outputs of latent space mimetic, anticipatory, speculative, predictive, metonymic, metaphorical, normative? What are the ethics of generating representations based on the underlying statistical distributions of human data? How might this question help us think through harmful or violent outputs from latent spaces, such as deepfakes, nudification, CSAM?

We want to share, think, and creatively engage across disciplines on how we might probe, interrogate, and situate the novel paradigms of knowledge, perception and action that emerge with latent spaces.  We encourage interdisciplinarity and are open to all modes of academic and artistic enquiry.

Practical information 

Please send a short proposal (250 words) alongside a bibliography (100 words) by Monday 30 November 2026 to [email protected] with the subject line ‘Latent Spaces conference submission’.

All submissions will be anonymised before review. This aims to ensure that submissions are judged on their own merit by eliminating unconscious bias which may occur when gender, reputation, institution, or other factors are known to reviewers. The peer review will be conducted by academics in relevant fields. 

We aim to send out notification of acceptance by January 21 2027. Bursaries will be available by application at a later date to early career researchers, graduate students, and others on the basis of need.