What It Will Take to Make AI Sustainable

8 hours ago

Tech giants are expanding fossil fuel-powered data centers for AI, conflicting with earlier emissions reduction pledges.
Sasha Luccioni advocates for greater transparency in AI's environmental impact, highlighting growing demand from customers and companies.
Luccioni co-founded Sustainable AI Group to help companies mitigate AI's environmental effects and study under-researched areas like energy needs of specific AI tools.
Companies face internal pressure from employees and boards to quantify AI's impact on ESG goals and understand risks like data center locations and supply chain emissions.
Regulatory differences exist globally, with Europe integrating sustainability into the EU AI Act and other regions pushing for transparency in data center reporting.
Luccioni proposes features like energy usage meters in AI interfaces and believes sustainability could become a competitive advantage for model providers.
There is a need to educate users on alternatives to large models, such as simpler classifiers, to reduce unnecessary energy consumption.
The concentration of companies controlling both model development and compute resources limits diversity in AI options and transparency.
Accurate data on AI's energy and water use is essential for informed decision-making, despite challenges in communicating scale and nuance.

Hasty Briefsbeta