At the AI Impact Summit in New Delhi, OpenAI CEO Sam Altman dismissed claims about AI's water consumption as false, while acknowledging energy use as a critical issue requiring nuclear or renewable solutions, and sparked controversy by comparing AI's energy efficiency to that of humans.
Water Use Claims Denied
- Altman described online assertions that ChatGPT uses gallons of water per query as "completely untrue, totally insane," with "no connection to reality."
- Data centers traditionally use water for cooling, but newer facilities may eliminate this need.
- A report from Xylem and Global Water Intelligence predicts water draw for cooling could more than triple over 25 years due to increasing computing demand.
Energy Consumption Acknowledged
- Altman admitted that total energy consumption from AI is a legitimate concern.
- He urged a rapid transition to nuclear or wind and solar energy sources.
- Comparing AI training to human development, he noted that training a human takes 20 years of life and food consumption.
- For inference (using trained models), Altman claimed AI may already surpass humans in energy efficiency.
Criticism and Debate
- Sridhar Vembu, co-founder of Zoho Corporation, criticized equating technology with humans, stating on X: "I do not want to see a world where we equate a piece of technology to a human being."
- The debate unfolds as global data center electricity consumption nears levels of countries like Germany or France, per an IMF report.
- Governments are accelerating energy project approvals, but environmentalists warn of conflicts with net-zero targets.
- Local communities, such as in San Marcos, Texas, have opposed data centers over grid strain and cost concerns.
Industry and Global Context
- Tech leaders, including Altman, argue that data centers necessitate expanded energy production from diverse sources, including renewables and nuclear power.
- The discussion reflects broader anxieties about AI's environmental impact and resource demands.