David Edwards
Nvidia CEO Jensen Huang believes artificial intelligence itself may be the most effective tool to combat the darker applications of AI. Speaking at a Bipartisan Policy Center event in Washington, D.C. on Sept. 27, Huang emphasized that AI’s speed and capacity to generate false information will require equally advanced AI systems to counteract its misuse.
“It’s going to take AI to catch the darker side of AI,” Huang noted, highlighting the increasing sophistication of AI-driven misinformation. “AI is going to be producing fake data and false information at very high speeds. So it’ll take somebody with very high speeds to detect that and to shut it down.”
AI as a Defense Against AI
Huang likened the challenge of combating malicious AI to modern cybersecurity, where companies face constant threats of hacks and attacks. “Almost every single company is at risk of a hack or attack at almost all times,” Huang said, suggesting that better cybersecurity powered by AI will be necessary to stay ahead of the threat landscape.
The Nvidia chief’s comments come as concerns over AI-driven misinformation gain momentum, especially in the lead-up to the U.S. federal elections. A Pew Research Center survey, conducted among 9,720 adults and published on Sept. 19, revealed that nearly 60% of respondents — across political lines — are deeply worried about AI being used to fabricate information about presidential candidates.
In the same survey, 40% of respondents predicted that AI would be used “mostly for bad” in the election context, underscoring widespread fears of its misuse for political manipulation. This concern was further amplified when an anonymous U.S. intelligence official recently informed ABC News that Russia and Iran are using AI to alter videos of Vice President Kamala Harris in a bid to influence the upcoming elections.
U.S. Must Become an AI Leader, Not Just a Regulator
During his discussion, Huang urged the U.S. government to not only regulate AI but to actively engage with it. He stressed that every government department, particularly the Departments of Energy and Defense, should become “practitioners of AI.” Huang even proposed the construction of an AI supercomputer to advance the nation’s technological capabilities, noting that such infrastructure would drive innovation and allow scientists to develop cutting-edge AI algorithms.
The Future of AI and Energy Consumption
Huang also touched on the substantial energy requirements for future AI systems, predicting that AI data centers will eventually consume significantly more power than they do today. The International Energy Agency already estimates that data centers account for up to 1.5% of global electricity consumption, but Huang speculated this figure could increase tenfold as AI models evolve and rely on other AI systems for learning.
“Future AI models are going to rely on other AI models to learn, and you could use AI models to curate the data so that future AI uses an AI to teach another AI,” Huang explained.
To manage the growing energy demand, Huang suggested building AI data centers in regions with surplus energy resources that are difficult to transport. “We can transport the data center,” Huang said, proposing that facilities be located near these energy sources to capitalize on their availability.
