On the Concept of History (in Foundation Models)

Published on ● Video Link: https://www.youtube.com/watch?v=6WLv3zEfu6g



Duration: 46:00
724 views
18


Fabian Offert (UC Santa Barbara)
https://simons.berkeley.edu/talks/tbd-456
AI and Humanity

This talk discusses the concept of history inherent in so-called 'foundation models,' focusing on OpenAI’s CLIP model, a large-scale multimodal model that projects text and image data into a common embedding space. CLIP facilitates not only the automated labeling of images but also the automated production of images from labels (as one component of the DALL-E 2 composite model). Starting from Walter Benjamin’s concept of history, I argue that the spatialization of the past that occurs in models like CLIP invalidates the past’s potential to become history, to be productively reframed in a moment of crisis, to be both similar and dissimilar to the present at the same time




Other Videos By Simons Institute for the Theory of Computing


2022-07-22Leveraging Molecular Data for Drug Discovery
2022-07-22The Rewards and Challenges of Constructing Patient Registries in Mexico
2022-07-22Whole Genome Methylation Patterns as a Biomarkers for EHR Imputation
2022-07-22Biological Discovery and Consumer Genomics Databases Activate Latent Privacy Risk in...
2022-07-22How Do We Deliver Precision Health at Scale for All?
2022-07-22Longitudinal Phenotypes and Disease Trajectories at Population Scale
2022-07-22Towards Making Identification of Noncoding Causes of Human Disease Routine
2022-07-22Nanopore Basecalling for Directed Evolution
2022-07-22Breaking the Winner's Curse in Mendelian Randomization: Rerandomized Inverse Variance...
2022-07-22Computational Challenges in a Densely Sequenced Tree of Life
2022-07-16On the Concept of History (in Foundation Models)
2022-07-16Race Beyond Perception: Analysing Race in Post-visual Regimes
2022-07-15Designing Human-Aware Learning Agents: Understanding the Relationship between Interactions...
2022-07-15Reimagining the machine learning life cycle in education
2022-07-15Aligning Robot Representations with Humans
2022-07-15The Future of Good Decisions’: a research paradigm for quality in automated decision-making
2022-07-15The Flaws of Policies Requiring Human Oversight of Government Algorithms
2022-07-15Assistive Teaching of Motor Control Tasks to Humans
2022-07-15Where Does the Understanding Come From When Explaining Automated Decision-making Systems?
2022-07-14From Optimizing Engagement to Measuring Value
2022-07-14Intelligent Technology and the Attention Economy: A Buddhist Perspective on the Risks of...



Tags:
Simons Institute
theoretical computer science
UC Berkeley
Computer Science
Theory of Computation
Theory of Computing
AI and Humanity
Fabian Offert