AI's Environmental Footprint: Altman Addresses Concerns
OpenAI CEO Sam Altman recently weighed in on the environmental impact of artificial intelligence during an event hosted by The Indian Express. While in India for a prominent AI summit, Altman addressed growing concerns about AI's resource consumption.
Debunking Water Usage Myths
Altman categorically dismissed claims regarding AI's water consumption as "totally fake." He acknowledged that water usage was a concern in the past due to evaporative cooling methods in data centers. "Now that we don’t do that, you see these things on the internet where, ‘Don’t use ChatGPT, it’s 17 gallons of water for each query’ or whatever," Altman stated. "This is completely untrue, totally insane, no connection to reality."
Focus on Energy Consumption
However, Altman conceded that concerns about the overall energy consumption of AI are valid, especially given the widespread adoption of the technology. "It’s fair to be concerned about the energy consumption -- not per query, but in total, because the world is now using so much AI," he explained. He emphasized the need for a swift transition towards renewable energy sources like nuclear, wind, and solar power.
Independent Research and Industry Transparency
Currently, there are no legal mandates requiring tech companies to disclose their energy and water usage. Consequently, researchers are undertaking independent studies to assess these impacts. Concerns about data centers' energy demands have also been linked to rising electricity prices.
AI Query Efficiency vs. Human Battery Life
Addressing a question about the energy equivalent of a single ChatGPT query compared to charging an iPhone, Altman suggested such comparisons are inaccurate. "There’s no way it’s anything close to that much," he responded, refuting a previous discussion with Bill Gates. Altman also criticized the common practice of comparing the energy required to train an AI model with the energy a human uses for a single inference query.
The 'Training' of Humans
Altman offered a different perspective, drawing a parallel between AI training and human development. "But it also takes a lot of energy to train a human," he argued. "It takes like 20 years of life and all of the food you eat during that time before you get smart. And not only that, it took the very widespread evolution of the 100 billion people that have ever lived and learned not to get eaten by predators and learned how to figure out science and whatever, to produce you."
A Fairer Comparison
He proposed a more equitable comparison: evaluating the energy needed for an AI to answer a question after its model is trained, versus the energy a human expends to answer the same question. "If you ask ChatGPT a question, how much energy does it take once its model is trained to answer that question versus a human? And probably, AI has already caught up on an energy efficiency basis, measured that way," Altman concluded.
Stay Tuned to Devignitor Insights for More Updates
