NVIDIA CEO Jensen Huang said he would stop supplying chips to customers developing unsafe AI. However, he believes concerns about artificial intelligence are greater than necessary.
NVIDIA leads in supplying chips used to train and run AI models. This position places it at the centre of the debate over AI safety. Some fear the technology could become too powerful and escape human control.
Huang spoke to Bloomberg’s Emily Chang during an interview in London. Asked about stopping supplies to an irresponsible developer, he said he would consider it.
“It’s hypothetical, but yes,” he said.
However, would first discuss the matter with the company concerned.
“We don’t run other people’s companies, and there’s a limit to the judgment we apply to other companies’ use of our technology. Our technology is general-purpose,” he added.
Anthropic chief executive Dario Amodei has suggested slowing industry growth to address these concerns. , however, believes AI can be controlled and used safely.
He described safety fears as growing pains within a rapidly expanding industry. Many young companies are still learning the responsibility that comes with their growth, he said.
“Many of the young companies are going through growing pains, and they’re growing incredibly fast. There’s an accountability, responsibility that goes along with building companies that they’re just growing into,” Huang told Bloomberg.
AI safety research
NVIDIA is supporting British AI safety research with $250 million over five years. This support will provide researchers access to advanced computers.
Institutions including Oxford University and University College London will study ways to make AI safer. Their work will cover safeguards and methods for controlling AI agents.
Huang said some concerns were created by the industry itself. He also criticised its poor explanation of AI’s benefits to the public.
He believes AI can help monitor and control other AI systems. He called for faster development of monitoring tools and their wider use within systems.
Meanwhile, NVIDIA shares and other chip stocks led declines in American trading. Huang also dismissed concerns about an AI bubble.
He said demand for AI infrastructure would exceed component supplies for several years. Supply and demand would eventually balance, he told an event in London. However, he stressed that this balance had not yet arrived.
Sunil Mittal on AI regulation
Back in India, Sunil Mittal has urged governments to act early on artificial intelligence rules. He said AI was growing faster than governments could prepare regulations.
Mittal warned that AI agents could act beyond their intended controls. He said governments must take warnings from technology founders seriously.
“Every day, the news flow starts to come in from the very founders of the technology, saying, ‘We now have a fear that our agents may run rogue’. That’s not something that we can take lightly, or a country can take lightly,” the Bharti Airtel boss said.
“Are we going to wake up only when an accident happens, or do we do it before?” he questioned.