Marketplace Tech Is Google honest about the environmental impacts of Gemini?
Sep 16, 2025
Emma Strubell, an Assistant Professor at Carnegie Mellon’s Language Technologies Institute, dives into the environmental impacts of Google’s Gemini AI model. She challenges claims that the model has negligible resource consumption per query, urging a deeper examination of energy and water usage. The conversation covers best practices for making AI more resource-efficient and the need for standardized reporting in the tech industry. Strubell highlights the tension between AI's potential benefits and its environmental costs, pushing for greater transparency.
AI Snips
Chapters
Transcript
Episode notes
Per-Query Metrics Can Be Misleading
- Google's per-query numbers can hide the bigger environmental picture by ignoring scale.
- Emma Strubell warns that small per-query savings matter less if total queries grow rapidly.
Prioritize High-Value AI Uses
- Evaluate whether each AI use case justifies its energy and resource cost.
- Prioritize AI for high-impact problems like climate research rather than low-value tasks.
Efficiency Gains Need Usage Transparency
- Efficiency improvements per query are real but insufficient without transparency on total usage.
- Strubell stresses that Google reports per-query gains but not the rising total number of queries.