AI and Feminist Worldmaking in Technofascist Times

Catherine D’Ignazio

Massachusetts Institute of Technology
Cambridge, USA

Abstract: Drawing on her forthcoming book, 10 Ways to Stop Technofascism (MIT Press, 2027), D’Ignazio will contextualize the rise of “technofascism” – the collusion of large technology firms with authoritarian, extractive, and anti-democratic political agendas. While we can see technofascism manifesting globally, it is not inevitable and we do not have to accept a future predicated on accumulation, violence, and exclusion. To that end, D’Ignazio will highlight the importance of feminist worldmaking through technology. Drawing from the large-scale participatory AI project described in Counting Feminicide (MIT Press, 2024), she will describe how the research team built technology with data activists and journalists to enact alternative epistemological approaches to information science that center care, memory and justice. These and other experiments in “solidarity technology” (Souza De Alvear et al. 2026) place knowledge and information in the service of informatic futures that foster life, living, and vitality.

Biography: Catherine D’Ignazio is a hacker mama, scholar, and artist/designer who focuses on feminist technology, data justice and civic engagement. She has run women’s health hackathons, designed global news recommendation systems, created talking and tweeting water quality sculptures, and led walking data visualizations to envision the future of sea level rise. Her 2020 book from MIT Press, Data Feminism, co-authored with Lauren Klein, charts a course for more ethical and empowering data science practices. Her second book, Counting Feminicide: Data Feminism in Action (MIT Press, 2024) is an extended case study about grassroots data activism to end gender-related violence. Her forthcoming book, 10 Ways to Stop Technofascism (MIT Press, 2027), is a resistance manual for the general public to abolish Big Tech, refuse AI-fueled authoritarian violence, and build the technology that we deserve. D’Ignazio is an Associate Professor of Urban Science and Planning in the Department of Urban Studies and Planning at MIT where she is the Director of the Data + Feminism Lab.

Learn to Compress or Compress to Learn? Shannon and Turing meet again in the LLM era

Paolo Ferragina

Scuola Superiore Sant’Anna
Pisa, Italy

Abstract: The year 1948 gave birth to two transformative pillars of modern computing: Claude Shannon’s information theory, which gave rise to data compression, and Alan Turing’s writings on intelligent machinery, which ignited the field of artificial intelligence. Today, nearly eight decades later, we are witnessing a profound hybridisation of these once-distinct fields. As we push to efficiently store, process, and retrieve massive datasets across diverse data types, users, devices, and systems — from vector and graph databases to cutting-edge agentic memories — AI and compression can no longer operate in silos.
In this talk, we investigate how Learning is fundamentally reshaping the fields of Indexing and Compression. We will focus on two challenging research directions: designing Learned Data Structures that fuse classical indexing with machine learning to achieve unprecedented efficiency, and developing next-generation Data Compressors driven by powerful autoregressive and diffusion neural models to reach new frontiers of efficacy. We will also examine the reverse direction — how Compression can reshape learning through model compaction and comprehension. By unpacking the distinct computational advantages and current practical pitfalls of these lines of investigation, we will outline a roadmap for these research frontiers and hopefully inspire new cross-disciplinary collaborations.

Biography: Paolo Ferragina is a professor of Computer Science at the Sant’Anna School of Advanced Studies, Pisa (Italy). He earned a PhD in Computer Science from the University of Pisa and completed a postdoc from the Max Planck Institute for Informatics. His research focuses on the design of algorithms and data structures for compression, indexing, and retrieval of information and knowledge from large collections of multimodal data. His work has led to several patents and 200+ articles published in leading peer-reviewed conferences and journals. He is also the co-author of five books written in Italian and English, the last one published by Cambridge University Press on Algorithm Engineering, and also translated into Chinese. He is an ACM Fellow, and the co-recipient of the ’2022 ACM Paris Kanellakis Theory and Practice Award’, and several other international awards. Previously, he was a Professor at the University of Pisa, where he served as Vice-Rector for Applied Research and Innovation (2010-16), as Director of the (regional) PhD Program in Computer Science (2017-20), and as Vice-Rector for ICT (2019-22).

Reliable by Construction: Query Processing over Unstructured Data with Guarantees

Nick Koudas

University of Toronto
Toronto, Canada

Abstract: Most of the evidence behind real decisions, contracts, clinical narratives, filings, reports, is text. Increasingly, people want to query it the way they query a database: with selections, joins, and aggregates whose semantics they can defend. Pipelines built on large language models make this feel within reach, but they are best-effort by design: a dropped clause or misread value produces no error, it silently changes the answer. For analytics, that is not good enough. This talk argues that reliability must be engineered into the query processor rather than hoped for from the model, and proposes four principles for doing so: treat the language model as a fallible, inspectable component whose internal states reveal more than its stated confidence; convert those signals into explicit, distribution-free error contracts per operator; make abstention and human review first-class operators whose cost is minimized, not ignored; and treat the user’s error allowance as a budget allocated across retrieval, extraction, filtering, and joining, much as an optimizer allocates cost. I will illustrate these principles with recent results, including our own, sketch an architecture in which they compose, and close with open problems: end-to-end guarantees, distribution shift, closed models, and evaluating systems that promise control rather than just accuracy.

Biography: Nick Koudas is a Professor of Computer Science at the University of Toronto. His research addresses fundamental challenges in large-scale data management, at the intersection of data systems, big data analysis, data science, and applied machine learning, with a focus on architecting systems and algorithms that process and extract insights from massive, heterogeneous datasets. A central theme of his group’s work is the deep integration of machine learning as a core primitive within scalable data platforms. He holds a PhD from the University of Toronto, an MSc from the University of Maryland at College Park, and a Bachelor’s degree from the University of Patras in Greece. His work has been recognized with several best paper awards at premier international forums, and the University of Toronto Inventor Award. He has served the data management community extensively, including as Associate Editor of ACM Transactions on Database Systems (TODS), IEEE Transactions on Knowledge and Data Engineering (TKDE), and the Proceedings of the VLDB Endowment, as Program Committee Chair of major conferences in the field, and in organizational and leadership roles within ACM SIGMOD. He is committed to translating research into practice through entrepreneurship: he co-founded Sysomos (now part of the Meltwater Group) and Aislelabs (acquired by Constellation Software), is currently a co-founder at Workorb, and advises ktau.ai, where the team is building a system that secures guaranteed ranking in AI search results.