By Abbas Nazil
Artificial intelligence may be revolutionizing industries and daily life, but experts warn that its environmental footprint is becoming a silent crisis.
According to Eliot Jacobson, a retired professor of mathematics and computer science who appeared on Santa Barbara Talks with Josh Molina, the massive data centers powering AI systems consume vast amounts of electricity and water, straining natural resources already under threat from climate change.
“The big problem is the power they consume and the water they consume,” Jacobson said, stressing that AI could undermine progress made through renewable energy sources like wind and solar.
As droughts intensify and rivers run dry, AI’s dependence on water for cooling servers is adding new pressure on ecosystems.
Jacobson estimated that global AI power usage now rivals the entire electricity consumption of England, raising alarms about sustainability.
Even everyday interactions with AI carry hidden costs.
A simple query could use the equivalent of a tablespoon of water, while generating an AI image might require as much as two cups.
Beyond immediate energy and water demands, Jacobson linked AI’s growth to broader environmental disruptions.
He highlighted how melting ice in Greenland is destabilizing jet streams through a phenomenon known as Atlantic Meridional Overturning Circulation.
He also pointed to Jevon’s paradox, where gains in efficiency often result in increased consumption, further intensifying ecological strain.
A self-described “Doomer,” Jacobson believes the damage may already be irreversible for humanity, arguing that focus should shift toward preserving Earth for future species beyond the looming sixth extinction.
Despite his grim outlook, he continues to use his platform to warn about overconsumption and unchecked technological expansion.
With tens of thousands of followers on social media and appearances on major networks like CNN, Jacobson has become a prominent voice urging society to confront the environmental price of AI before it is too late.
