In an effort to continue exploring various civil rights and civil liberties concerns in our increasingly digital society, Fordham’s McGannon Center and the Civil Rights & Civil Liberties in Digital Age Colloquium invited USC Annenberg Research Professor Kate Crawford to speak about artificial intelligence (AI) on November 11.
Crawford, a leading international scholar of the social and political implications of AI, discussed key themes from her latest book, titled Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence, as well as her research that focuses on understanding large-scale data systems through the wider contexts of history, politics, labor, and the environment.
Adjunct Professor Ari Waldman moderated the virtual conversation alongside Professor Catherine Powell, founder of the Civil Rights & Civil Liberties in Digital Age Colloquium.
Students from Professors Powell and Waldman’s Civil Rights & Civil Liberties in a Digital Age seminar also had the opportunity to ask Crawford questions ranging from emotion recognition AI systems to the possibility and reality of democratizing AI.
The Inspiration Behind Atlas of AI
In 2018, Crawford collaborated with Vladan Joler, professor at the New Media department of the University of Novi Sad, to map out the technical and human infrastructure behind the Amazon Echo. Their research included studying the full material lifecycle of these AI devices, from where their components were mined to where these kinds of devices were disposed of at their end of life.
Crawford called her research of the environment and supply chain impacts “eye opening,” and it inspired her to conduct the same analysis for the entire AI industry and kickstart the shift from abstractly understanding AI into something more concrete.
“One of the problems, I think, that can often happen with artificial intelligence is that people are either off put by the type of very dense technical prose that tends to accompany it or cannot see the whole system because those systems are, in many ways, occluded from us and intentionally obscured,” Crawford said. “Creating an atlas, to me, was a way of both looking at ‘the great houses of AI’—the less than a dozen companies that can do AI at scale—and looking at those material imprints by going to the physical places where the mines are located, seeing who’s working there, going inside Amazon fulfillment warehouses, [and]going into the labs where training data sets are being made. It is a material geography, if you will, of artificial intelligence.”
The Intersection of AI and Policymaking
Draft regulation and laws already in place, such as the latest proposal draft for the EU AI Act and General Data Protection Regulation (Europe’s new data privacy and security law), focus on ideas surrounding data protection—including how data is acquired and if notice or consent should be given to users. However, Crawford argues that policy needs to examine a wider set of forces that go into AI production like forms of exploitation in labor, data, and natural resources.
“It’s time to use a wider set of parameters in terms of thinking about the impacts of these systems, both in the short term and the long term, and how we might think about it from a policy lens,” Crawford said, “rather than just focusing on the narrow technical capabilities.”
Understanding AI’s planetary scalability in technology, she also says, can open more doors for public governance reform. “One of my great hopes is that, within the next decade, we will have different ways of regulating AI and large-scale data systems that incorporate these broader elements,” Crawford said. “I think we have to have a moment of starting to demystify artificial intelligence—to look exactly at how these systems work and recognize that these systems are only as opaque as we allow them to be.”
Collaborating for a Better Future
Crawford expressed optimism for the future towards the end of the discussion, saying that issue-based coalition building is “essential.”
“We’ve got real substantive structural problems when it comes to AI, but there is hope. For me, that hope comes from looking at the places where people can join together around particular conversations,” Crawford explained. “Climate activists don’t often necessarily break bread and share the same room as labor rights organizers or people who are working on data protection and looking specifically at the legal regulation of AI. But you can bring those people together in a room and have extraordinary conversations because these systems are touching all of those issues simultaneously.”