Join on Zoom

This discussion introduces the concept of the silicon gaze to explain how large language models (LLMs) reproduce and amplify long-standing spatial inequalities.

Drawing on a 20.3-million-query audit of ChatGPT, we map systematic biases in the model’s representations of countries, states, cities and neighborhoods. From these empirics, Zook argues that bias is not a correctable anomaly but an intrinsic feature of generative AI, rooted in historically uneven data ecologies and design choices. 

Building on a power-aware, relational approach, Zook and his team develop a five-part typology of bias (availability, pattern, averaging, trope and proxy) that accounts for the complex ways in which LLMs privilege certain places while rendering others invisible.

Matthew Zook is a human geographer working within the sub-disciplines of Digital, Economic and Urban Geographies. His approach blends critical theories drawn from STS, data/software studies and heterodox economics and uses socially-grounded, technical methods (such as audits of data structures, algorithms and larger digital systems) combined with qualitative methodologies to analyze the evolving spatialities of urban practices, representation and economic/financial activity.  

There are three main threads within his work: (1) How are digital technologies changing the spatial economy, globalization and cities? (2) How do big data and digital technologies provide new ways to study economic networks and cities? (3) What do big data and digital technologies mean for governance and policy?