What AI Speaks for Your Community: Polling AI Agents for Public Opinion on Data Center Projects

TL;DR

This work introduces an AI agent polling framework that uses large language models to gauge nuanced public opinion on data center projects, enabling scalable, region-specific early engagement. By synthesizing county demographics with IPF, constructing representative virtual agents, and polling them across multiple LLMs, the approach reveals regional variation and model-specific biases, with water use, utility costs, and tax revenue identified as key factors. Calibration with real-world data is proposed to provide statistical guarantees, and comparisons with human polls show topical alignment despite methodological differences. The framework offers a practical, scalable screening tool to inform socially responsible AI infrastructure deployment and data center siting decisions.

Abstract

The intense computational demands of AI, especially large foundation models, are driving a global boom in data centers. These facilities bring both tangible benefits and potential environmental burdens to local communities. However, the planning processes for data centers often fail to proactively integrate local public opinion in advance, largely because traditional polling is expensive or is conducted too late to influence decisions. To address this gap, we introduce an AI agent polling framework, leveraging large language models to assess community opinion on data centers and guide responsible development of AI. Our experiments reveal water consumption and utility costs as primary concerns, while tax revenue is a key perceived benefit. Furthermore, our cross-model and cross-regional analyses show opinions vary significantly by LLM and regional context. Finally, agent responses show strong topical alignment with real-world survey data. Our framework can serve as a scalable screening tool, enabling developers to integrate community sentiment into early-stage planning for a more informed and socially responsible AI infrastructure deployment.

Figures (21)

• Figure 1: AI agent polling framework for data center public opinion assessment. The framework synthesizes county-level demographics with project specifications to generate representative virtual agents, validated through chi-square tests. Multi-model polling across GPT-5, Gemini-2.5-Pro, and Qwen-Max captures responses to 13 questions spanning 5 core domains, enabling cross-model, cross-regional, and human poll comparative analysis.
• Figure 2: Taylor County results using GPT-5 ($n$=1000). (a) Community attitude distribution showing neutral-leaning support attitudes, mixed economic impact perception, widespread environmental concerns, and neutral government trust. (b) LLM topic analysis of open-ended community feedback revealing top-three themes. (c) Full topic breakdown (1. Water Resource Protection, 2. Utility Costs, 3. Local Jobs & Employments, 4. Clean Energy, 5. Economic Benefits, 6. Transparency & Public Reporting, 7. Accountability & Enforcement, 8. Grid Impact & Reliability, 9. Housing Costs, 10. Taxes & Public Finance). Charts display only selected response categories, and complete survey options are in Appendix \ref{['appendix']}.
• Figure 3: Key differences in cross-model polling results ($n$=1000). (a) Overall economic attitudes; (b) Top economic benefits (1: Tax Revenue, 2: Infrastructure Upgrades, 3: Business Growth, 4: Economic Diversity, 5: Job Creation); (c) Government trust; (d) Top trustworthy information sources (1: Academic Research, 2: Federal/State Agencies, 3: Local Government, 4: Community Organizations, 5: Local Media). Note: Selected categories; see Appendix \ref{['appendix']} for complete data.
• Figure 4: Cross-regional polling results ($n$=1000). (a) Overall attitudes for proposal data center. (b) The top conditions that would increase AI agents' support for the project (1: Environmental Protections, 2: Lower Utility Bills, 3: Local Job Guarantees, 4: Stricter Oversight). (c) Overall economic attitudes towards the project. (d) The community's top economic concerns regarding the proposed data center (1: Higher Utility Bills, 2: Benefits to Outsiders, 3: Housing Cost Inflation, 4: Public Service Strain). Note: Selected categories; see Appendix \ref{['appendix']} for complete data.
• Figure 5: Taylor County results using GPT-5 on economic issues. (a) Community opinions about the most important economic benefits brought by the data center project. (b) Community economic concerns. Note: figures showing distribution for all response options in Appendix \ref{['appendix']}.