Stanford
University
  • Stanford Home
  • Maps & Directions
  • Search Stanford
  • Emergency Info
  • Terms of Use
  • Privacy
  • Copyright
  • Trademarks
  • Non-Discrimination
  • Accessibility
© Stanford University.  Stanford, California 94305.
Embedding Democratic Values into Social Media AIs via Societal Objective Functions | Stanford HAI
Navigate
  • About
  • Events
  • AI Glossary
  • Careers
  • Search
Participate
  • Get Involved
  • Support HAI
  • Contact Us

Stay Up To Date

Get the latest news, advances in research, policy work, and education program updates from HAI in your inbox weekly.

Sign Up For Latest News

Skip to content
  • About

    • About
    • People
    • Get Involved with HAI
    • Support HAI
    • Subscribe to Email
  • Research

    • Research
    • Research Programs
    • Grants
    • Marlowe (opens in new tab)
    • Student Affinity Groups
    • Centers & Labs
    • Research Publications
    • Research Partners
  • Education

    • Education
    • Executive and Professional Education
    • Government and Policymakers
    • K-12
    • Stanford Students
  • Policy

    • Policy
    • Policy Publications
    • Policymaker Education
    • Student Opportunities
  • AI Index

    • AI Index
    • AI Index Report
    • Global Vibrancy Tool
    • People
  • News
  • Events
  • Industry
  • Centers & Labs
research

Embedding Democratic Values into Social Media AIs via Societal Objective Functions

Date
April 26, 2024
Topics
Democracy
Your browser does not support the video tag.
Read Paper
abstract

Mounting evidence indicates that the artificial intelligence (AI) systems that rank our social media feeds bear nontrivial responsibility for amplifying partisan animosity: negative thoughts, feelings, and behaviors toward political out-groups. Can we design these AIs to consider democratic values such as mitigating partisan animosity as part of their objective functions? We introduce a method for translating established, vetted social scientific constructs into AI objective functions, which we term societal objective functions, and demonstrate the method with application to the political science construct of anti-democratic attitudes. Traditionally, we have lacked observable outcomes to use to train such models-however, the social sciences have developed survey instruments and qualitative codebooks for these constructs, and their precision facilitates translation into detailed prompts for large language models. We apply this method to create a democratic attitude model that estimates the extent to which a social media post promotes anti-democratic attitudes, and test this democratic attitude model across three studies. In Study 1, we first test the attitudinal and behavioral effectiveness of the intervention among US partisans (N=1,380) by manually annotating (alpha=.895) social media posts with anti-democratic attitude scores and testing several feed ranking conditions based on these scores. Removal (d=.20) and downranking feeds (d=.25) reduced participants' partisan animosity without compromising their experience and engagement. In Study 2, we scale up the manual labels by creating the democratic attitude model, finding strong agreement with manual labels (rho=.75). Finally, in Study 3, we replicate Study 1 using the democratic attitude model instead of manual labels to test its attitudinal and behavioral impact (N=558), and again find that the feed downranking using the societal objective function reduced partisan animosity (d=.25). This method presents a novel strategy to draw on social science theory and methods to mitigate societal harms in social media AIs.

Share
Link copied to clipboard!
Authors
  • Chenyan Jia
  • Michelle Lam
    Michelle Lam
  • Michael S. Bernstein
    Michael S. Bernstein
  • Minh Chau Mai

Related Publications

The Global AI Vibrancy Tool 2025
Loredana Fattorini, Nestor Maslej, Ray Perrault, Vanessa Parli, John Etchemendy, Yoav Shoham, Katrina Ligett
Deep DiveNov 24, 2025
Research
Your browser does not support the video tag.

This methodological paper presents the Global AI Vibrancy Tool, an interactive suite of visualizations designed to facilitate cross-country comparisons of AI vibrancy across countries, using indicators organized into pillars. The tool offers customizable features that enable users to conduct in-depth country-level comparisons and longitudinal analyses of AI-related metrics.

Research
Your browser does not support the video tag.

The Global AI Vibrancy Tool 2025

Loredana Fattorini, Nestor Maslej, Ray Perrault, Vanessa Parli, John Etchemendy, Yoav Shoham, Katrina Ligett
DemocracyIndustry, InnovationGovernment, Public AdministrationDeep DiveNov 24

This methodological paper presents the Global AI Vibrancy Tool, an interactive suite of visualizations designed to facilitate cross-country comparisons of AI vibrancy across countries, using indicators organized into pillars. The tool offers customizable features that enable users to conduct in-depth country-level comparisons and longitudinal analyses of AI-related metrics.

The Healthcare Policies of the November 2026 Candidates
David Scheinker, Kevin Schulman, Anton Backer, Maya Shah
Quick ReadOct 20, 2026
Research
Your browser does not support the video tag.

Recent polls show that healthcare costs are a top affordability concern for American adults and that a large majority of adults, 90% of Democrats and 75% of Republicans, support a fundamental redesign or major changes to our healthcare system. Across states, political parties, and different types of office, politicians and constituents hold different views on healthcare policies and support different proposals. Various stakeholders such as patients, physicians, nurses, veterans, employers, and industry groups could potentially benefit from access to data on how their representatives’ views align with their own. There is no systematic, openly available database and analysis of these views. We sought to create the first systematic, openly available database and analysis of the healthcare policies and proposals of the leading candidates in the November 2026 elections and beyond.

Research
Your browser does not support the video tag.

The Healthcare Policies of the November 2026 Candidates

David Scheinker, Kevin Schulman, Anton Backer, Maya Shah
HealthcareRegulation, Policy, GovernanceQuick ReadOct 20

Recent polls show that healthcare costs are a top affordability concern for American adults and that a large majority of adults, 90% of Democrats and 75% of Republicans, support a fundamental redesign or major changes to our healthcare system. Across states, political parties, and different types of office, politicians and constituents hold different views on healthcare policies and support different proposals. Various stakeholders such as patients, physicians, nurses, veterans, employers, and industry groups could potentially benefit from access to data on how their representatives’ views align with their own. There is no systematic, openly available database and analysis of these views. We sought to create the first systematic, openly available database and analysis of the healthcare policies and proposals of the leading candidates in the November 2026 elections and beyond.

Roles for AI in Implementation of Medicaid Work Requirements
Michelle Mello, Himaja Nagireddy
Quick ReadSep 03, 2026
Research
Your browser does not support the video tag.

By January 2027, 44 state Medicaid programs must redetermine the eligibility of millions of beneficiaries under criteria in the 2025 budget reconciliation law (HR 1). Among the changes are “community engagement” (work) requirements and increased redetermination cadence (from annual to at least semiannual). In preparation, state Medicaid agencies have been exploring technological solutions to automate redeterminations at scale, including artificial intelligence (AI). Six agencies have decided to use AI tools so far and 21 states are considering use of them. Strategic deployment of AI can help states tackle the redetermination juggernaut, but overly broad or hasty deployment could compound the harms of HR 1.

Research
Your browser does not support the video tag.

Roles for AI in Implementation of Medicaid Work Requirements

Michelle Mello, Himaja Nagireddy
HealthcareRegulation, Policy, GovernanceQuick ReadSep 03

By January 2027, 44 state Medicaid programs must redetermine the eligibility of millions of beneficiaries under criteria in the 2025 budget reconciliation law (HR 1). Among the changes are “community engagement” (work) requirements and increased redetermination cadence (from annual to at least semiannual). In preparation, state Medicaid agencies have been exploring technological solutions to automate redeterminations at scale, including artificial intelligence (AI). Six agencies have decided to use AI tools so far and 21 states are considering use of them. Strategic deployment of AI can help states tackle the redetermination juggernaut, but overly broad or hasty deployment could compound the harms of HR 1.

Addressing Medical Affordability Without Compromising Care
Kevin Schulman, Ezekiel J. Emanuel, MD, PhD
Quick ReadJul 09, 2026
Research
Your browser does not support the video tag.

The US spends 25 percent of all health care costs on administration, a big part of which involves billing for medical services. To realize savings requires standardizing and digitizing billing.

Research
Your browser does not support the video tag.

Addressing Medical Affordability Without Compromising Care

Kevin Schulman, Ezekiel J. Emanuel, MD, PhD
HealthcareRegulation, Policy, GovernanceQuick ReadJul 09

The US spends 25 percent of all health care costs on administration, a big part of which involves billing for medical services. To realize savings requires standardizing and digitizing billing.