AI and Social Normativity: Rethinking Error, Bias, and Truth

Date/Time
Tuesday
28 Jan 2025
1:00 pm - 5:00 pm

Location
Dwinelle 7415

Event Type
Non-CSTMS Event

Armen Khatchatouro
Associate Professor, Université Gustave Eiffel

David Bates
Professor, UC Berkeley

The proliferation of AI-based systems has led to new ways of understanding what normativity is. On the one hand, following Foucault, we have to reconsider how social normativity is at work in dynamic AI systems rather than in a slow-evolving set of social rules. Here, the questions are what is (provisionally) considered as error or bias, and with respect to which referent – and how deviation and alignment are addressed.

On the other hand, social normativity itself — not only in its content but in the very way we relate to it — is affected by the opacity and adjustability of machine-learning models, which nevertheless produce new expectations, behaviors and a form of algorithmic governmentality.

This Symposium aims to hold both questions together, in order to address their intertwined nature; it will reunite interdisciplinary scholars in philosophy and humanities from Berkeley, Paris and Stanford. Position talks (1 p.m. to 4 p.m. including a break) will be followed by an hour discussion (4 p.m. to 5 p.m.).

Thomas Gilbert (Hortus AI): A Concept of Life for AI: Public Purposiveness as the Basis of Norms

Warren Sack (UC Santa Cruz): AI as automation of institutions: a design perspective

Julia Irwin (Stanford): Alignment and the Problem of Value Pluralism in Contemporary AI

David Bates (UC Berkeley): Error, decision, machine learning

Anne Alombert (Université Paris 8, Paris): Norms and errors in machines and humans : a Simondonian approach

Johan Fredrikzon (Stanford): Training the Deceased: Deadbots and Technological Spiritualism

Armen Khatchatourov (Univ. Gustave Eiffel, Paris): Truths and rewards of algorithmic governmentality

 

Armen Khatchatourov is Associate Professor of Information and Communication Sciences at DICEN-IdF Lab, University Gustave Eiffel, Paris, France. He has a double background (engineering degree and PhD in philosophy of technology) and has previously worked as Senior Researcher at Institut Mines-Télécom; as HMI-AI Associate Researcher at Sony Computer Science Lab Paris; as HMI Researcher at Institut National Polytechnique de Grenoble as well as at the Institute of Research and Innovation/Centre Pompidou. He has also taught at University of Technology of Compiègne, Ecole des Ponts, Paris 1 – Sorbonne as well as in art and design schools (EESI, Strate, ENSAM). His research interests include digital identities, privacy and data protection, smart-cities and more generally the effects of Big Data an AI on the society and governance. He published Digital Identities in Tension: Between Autonomy and Control (ISTE/Wiley, 2019) and directed Corps Connectés. Figures, fragments, discours (Presses des Mines, 2022) and he is Editor-in-Chief of the Journal Etudes Digitales.

This event received financial support under the European Union’s MSCA-RISE programme, grant agreement no. 10100915 NEST – Networking Ecologically Smart Territories.

Speakers: 

  1. Armen KhatchatourovAssociate Professor, Université Gustave Eiffel
  2. Thomas Gilbert
  3. David BatesProfessor, UC Berkeley
  4. Anne AlombertLecturer, Université Paris 8, Paris
  5. Johan FredrikzonVisiting Scholar, Stanford University
  6. Warren SackAffiliate Faculty, Digital Arts & New Media (DANM) Program, University of California, Santa Cruz
  7. Julia IrwinHAI Postdoctoral Fellow, Stanford

This lecture will be held both online & in person. You are welcome to join in South Hall or via Zoom. For more information about this event, please visit the UC Berkeley Rhetoric Department website. If you have any questions about this event, please contact Eve Letendre, Access Coordinator, at rfa@berkeley.edu.

European Union’s MSCA-RISE Programme
This event is sponsored by: Department of Rhetoric • European Union’s MSCA-RISE Programme • Networking Ecologically Smart Territories

Other Events of Interest