Sustainable Digital
summary
Provides measured per-request energy estimates for web-scale AI and search (e.g., ~0.2 W per Google search; ChatGPT ~3 W, Bloom ~4 W, next‑gen models ~9 W per request) and argues data centers and AI cause quantifiable environmental costs rather than vague guilt. Presents a hardware efficiency ratio (1 W useful → ~27 W total) to illustrate systemic inefficiency, explains how large‑scale inference fanout and water use for cooling multiply the impact, and points to EU regulations and the ‘do‑no‑significant‑harm’ framework as concrete policy levers.
AI-written summary · about
chapters (10)
- 0:00Digital sustainability intro
- 3:56Digital butterfly effect
- 8:14Data center energy use
- 14:50Device energy efficiency
- 20:10Generative AI: how it works
- 24:01AI vs search energy use
- 28:43AI energy optimization
- 30:27EU data center rules
- 32:33Digital ecology: data identity
- 37:11Q&A
related talks
45mGreen AI: Making Machine Learning Environmentally Sustainable
44mThe Invisible Monster: The Carbon Footprint of AI
23mSustainable IT and API Performance - How to Bring Them Together.
07mKeynote: Finding Your Power to Accelerate to a Sustainable Future
14mGreener Intelligence: Reducing AI’s Carbon Footprint
25mBuilding a Sustainable Future
01h 02mJoint Keynote Address: Scaling AI Sustainably: An Uncharted Territory
20mBuilding GenAI That Doesn't Cost the Earth