Confronting the Crises of AI Through Research – UCLA Livescu Initiative on Neuro, Narrrative, and AI

Image from SWIM (2024) by Eryk Salvaggio

We are collaborating with Shazeda Ahmed (Chancellor’s Postdoctoral Fellow, UCLA Center on Race and Digital Justice) to organize a series of seminars around the applications of AI to both studying AI as an object of inquiry and discerning the inherent trade-offs in uses of machine learning in research.

The goal of the course is to deepen students’ familiarity with a range of methods for producing scholarship about AI, which we will co-develop in response to students’ research interests. Participants are expected to attend as many sessions as possible to ensure a cohesive final project (described below). The seminar will be held fully in-person and requires no prerequisites. Graduate students from all disciplines are welcome to attend.

The seminars are scheduled on Mondays from 4 – 6 pm, March 31 – June 2 2025, at 3312 Murphy Hall.

For questions and inquires please contact shazeda@g.ucla.edu

Overview

As applications of artificial intelligence proliferate in public life, it is tempting to cast long-standing issues (e.g., cultural representation, the ethics of war, the politics of knowledge production) as emergencies with binary solutions – e.g., ‘innovate vs. regulate’, ‘AI safety vs. AI ethics.’ In this graduate-level reading seminar, we will slow down this manufactured urgency to understand approaches to both studying AI as an object of inquiry and discerning the inherent trade-offs in uses of machine learning in research.

Throughout the seminar we will consult AI’s histories and reinventions to gauge what has changed in the contemporary iteration of AI’s boom-and-bust cycle. What types of evidence and argumentation are marshalled in debates about whether ‘superintelligent’ AI is possible? How can we understand the bases of competing views on whether AI systems – and the natural resource-extracting infrastructure on which they depend – can destroy and protect the environment? From the use of synthetic data to the restructuring of search engines and literature review tools, how has machine learning come to change scientific research methods? We will weave these questions into discussions that contextualize AI’s appearances in recent headlines, from claims about how automation will revolutionize work to tracing AI’s role in the restructuring of the US government. We will critically assess how fields including science and technology studies, sociology, anthropology, computer science, and law and policy have devised methodologies to study AI’s effects on the world.

Each week, we will read and discuss approximately one book, 2-4 research papers and news articles, and at least one reading selected by the seminar’s participants. The seminar will require two assignments, one conducted as a group and one as individuals. For the first assignment, we will use a series of writing and discussion activities distributed throughout the quarter to collectively develop a set of written recommendations for conducting research on AI (modeled after similar examples). In the individual assignment, students will be responsible for writing a research design and annotated bibliography for a project contributing to their own independent scholarship. If there is interest, the instructor will edit a selection of these into a series of short essays accompanied by reading lists to be published through the Neuro, Narrative, and AI initiative.

Session I: March 31 2025

In Week 1, we will introduce the purpose behind this seminar: identifying the epistemics of how knowledge about AI is produced, and determining how this can inform our own sensibilities when developing our scholarship.

What does studying AI look like from the emerging field of “critical AI”? What is the view from a data science perspective? What features are each of these disciplinary backgrounds attuned to when reading grandiose proclamations from major AI companies?

Readings

Raley, R. and Rhee, J. (2023). Critical AI: A Field in Formation. American Literature, 95(2).
Donoho, D. (2024). Data Science at the Singularity. Harvard Data Science Review, 6(1).
Amodei, D. (2024). Machines of Loving Grace.

RSVP

Session II: April 7th 2025: “From Expert Systems to Large Language Models: What Was Lost, What Was Gained?”

Please note that Week 2 will only be from 4 – 5pm

In Week 2, we’ll start with an ethnographic account from Diana Forsythe of what the ‘expert systems’ approach to building AI looked like in the 1980s-90s. Which of Forsythe’s observations carry through to the present-day fixation with large language models (LLMs)? How do computer scientists perceive the limits of what can and should be done with contemporary machine learning methods? We’ll contrast technically-rooted skepticism with a provocative take from linguistic anthropology on how LLMs make meaning with language, a view that challenges emerging notions of ‘agency’ and ‘intelligence’ we will continue to interrogate throughout the quarter.