Artificial Intelligence Meets Natural Stupidity: Managing the Risks [pdf] (integrative-design-for-radical-energy-efficiency.stanford.edu)

Amory B. Lovins highlights serious risks involved in the rapidly growing demand for electricity driven by artificial intelligence (AI), emphasizing the speculative nature of current investments in data centers. With significant investments from major tech companies projected to exceed a trillion dollars, concerns arise over potential overbuilding of electricity infrastructure, reminiscent of the coal industry's past miscalculations. Despite forecasts suggesting AI could dramatically increase electricity demand, actual national usage remains stable, with data centers currently consuming only about 4.5% of U.S. electricity. This disconnect raises alarms about a possible electricity bubble that could financially burden utilities, leading to higher costs for consumers. The report argues for a more nuanced understanding of AI's electricity needs and advocates for pairing data centers with renewable energy sources to mitigate risks associated with grid reliability and financial burdens on existing users. Lovins proposes regulatory measures to ensure that developers bear the financial risks tied to overbuilding, advocating for smarter energy strategies that prioritize efficient management of demand and supply. He also warns that while AI's benefits are profound, its potential to drive fossil fuel extraction could exacerbate environmental issues, necessitating urgent scrutiny to ensure that investments in the AI era foster economic stability and sustainability rather than repeating past mistakes.