➡️ I also experimentally assess the political implications of these views, demonstrating that employing algorithms in contexts where citizens perceive them as unfair significantly undermines the legitimacy of the policy decisions and interventions they inform. 5/6

➡️ I show that across a wide range of domains, people are more likely to accept algorithmic systems in decisions designed to assist rather than punish and in contexts requiring inferences about collectives rather than individuals. 4/6

I introduce a new theory, arguing that views on this matter vary as a function of:
1️⃣ the objective of the decision at stake;
2️⃣ the population directly affected by the decision.

I test the theory using original experiments embedded in a national U.S. survey. 3/6

🚨 Excited to share that my job market paper is now online at @thejop.bsky.social!

"When Do Citizens Resist the Use of AI Algorithms in Public Policy? Theory and Evidence"

shorturl.at/hHoxw

1/6

Honored to organize the mini-conference on AI & Democratic Governance at @ispsyale.bsky.social! This year, we explored the risks AI poses to democracy, approaches to governing it responsibly, and its potential to enhance democratic participation and representation.

It was an honor to host the conference on AI in Government at @ispsyale.bsky.social with stimulating conversations and reflections by the talented:
Savannah Thais, Kirk Bansak, Alan Gerber, Eddie Yang, Virginia Eubanks @geomblog.bsky.social @elizaoak.bsky.social @kaylynjschiff.bsky.social

Excited to organize the AI & Algorithms in Government conference at @ispsyale.bsky.social on April 19, 2024.
Join us to explore the promise, challenges, and politics of using AI in public policy.

Agenda + registration: bit.ly/4ayMzii

#Polisky

Results indicate that when people are asked about concrete redistributive measures, levels of support for government action change significantly and attitudes align more closely with respondents' self-interest. 5/

Using original survey data from both open-ended and fixed-choice questions, we show that most people have little grasp of what reducing inequality actually entails. 3/

Very happy to share my paper with Yotam Margalit, now online at PSRM: "Does support for redistribution mean what we think it means?" cup.org/3HthSOR

🧵Here’s a summary thread… 1/

Our results indicate that AI-related attitudes are not fully determined by prior dispositions or beliefs about the technology. 6/7

In contrast, exposure to information about the technology did lead to significant attitudinal change over time. 5/7

Gov agencies increasingly turn to AI for policy implementation decisions (e.g. policing, welfare). While leaders acknowledge public input is needed to ensure AI regulation aligns with citizens’ values, it's unclear what such input entails.

We explore this question.. 2/7

Happy to share a new working paper with Yotam Margalit, "The Politics of Using AI in Public Policy: Experimental Evidence"
papers.ssrn.com/sol3/papers....
1/7