The AI Safety Intern will develop original measurement and modeling frameworks for assessing the community-level risks and benefits of data center infrastructure as it pertains to artificial intelligence systems. This internship is grounded in an emerging and urgent body of scholarship demonstrating that AI systems may generate carbon emissions equivalent to a major global city and consume water on the scale of the entire global bottled water market annually and yet remain largely opaque to public scrutiny due to inadequate corporate disclosure. The Intern will work on developing quantitative frameworks that translate macro-level environmental footprint estimates into community-scale impact assessments, with particular attention to the disproportionate burdens borne by low-income communities and communities of color proximate to data center siting decisions.
Requirements
- Current enrollment in or recent completion of an undergraduate or graduate degree program in Environmental Science, Environmental Engineering, Data Science, Statistics, Geography, Public Policy, or a closely related field with demonstrated quantitative coursework
- Demonstrated familiarity with environmental footprinting methodologies, including life cycle assessment concepts, carbon accounting frameworks (such as the Greenhouse Gas Protocol), and water stress metrics
- Prior exposure to or coursework in environmental justice, civil rights policy, or community-based research is a significant asset
- Evidence of research or analytical capability appropriate to career stage, such as an undergraduate thesis, capstone project, research assistantship deliverable, or independent analysis demonstrating the ability to collect, process, and interpret quantitative environmental or sociotechnical data
- Proficiency in at least one quantitative programming environment — Python or R preferred — with demonstrated ability to collect and process structured and semi-structured data from corporate sustainability reports, regulatory filings, and open government datasets, and to construct and communicate quantitative models of environmental performance metrics
- Working familiarity with geospatial data tools and visualization platforms (such as QGIS, ArcGIS, or Python-based libraries including GeoPandas and Folium) sufficient to map data center infrastructure against community demographic and environmental vulnerability layers using publicly available census and environmental datasets
- Practical experience using AI-assisted research tools — including large language models for literature synthesis, data extraction, and document analysis — and the capacity to apply these tools responsibly in research contexts requiring careful verification of outputs and transparent documentation of methods
- Clear and adaptable written communication skills, including the developing ability to present quantitative research findings in formats appropriate for varied audiences — from technical research memoranda for academic and scientific reviewers to community-accessible fact sheets and policy briefs for advocates and local officials
