Module 2: AI Impacts & Responsible AI · BoK II.A
Environmental harms
Four quantified facts - the numbers are the exam bait. Training a large model can emit over 626,000 pounds of CO₂ (~five times the lifetime emissions of an American car); energy matches major cloud providers; each casual generative-AI use is like wasting a small bottle of water; and lithium extraction demands enormous water.
Four quantified facts. The numbers themselves are the exam bait.
- 🏭 Carbon emissions → training a large AI model can emit over 626,000 pounds of CO₂ → roughly five times the lifetime emissions of an American car.
- ⚡ Energy consumption → energy used to train top models matches the energy mix of major cloud service providers → a significant environmental footprint.
- 💧 Water usage → each casual use of generative AI is likened to wasting a small bottle of water → it accumulates fast at scale.
- 🔋 Lithium extraction → battery lithium demands enormous water usage; future availability is a worry → rapid demand growth, geopolitical concentration, slow mine development.