OpenAI CEO Sam Altman Confident AI Industry Can Manage Safety Independently
OpenAI CEO Sam Altman has stated that the public should have greater trust in his company and other developers to act responsibly in the development of artificial intelligence (AI). This statement comes amidst growing public concern regarding the risks associated with the technology.
“The world must believe that we will do the right thing because it is the right thing, and we feel the magnitude of this responsibility,” Altman said on Tuesday while attending an annual conference organised by Salesforce in San Francisco.
Nevertheless, Altman acknowledged that public anxiety regarding AI is understandable given the extremely rapid pace of technological progress. “It does not take much imagination for us to imagine how this could go wrong. In my opinion, the world is right to be afraid of this,” he said.
Altman’s remarks follow a viral post by an AI researcher who resigned from Anthropic, claiming that AI could threaten humanity if left unsupervised. Anthropic CEO Dario Amodei subsequently called for a slowdown in the pace of AI development and urged governments to introduce regulations.
In the absence of official regulation in the United States, Altman believes AI companies are capable of implementing safe self-regulation. “We will do it correctly; I am very confident in the ability of our companies and our industry to do this safely,” Altman remarked. He added that they would “maintain alignment and safety far ahead of capability,” and are prepared to “slow down or stop” if they cannot guarantee such safety.
Nvidia CEO Jensen Huang expressed similar views, suggesting that AI companies themselves should determine the suitability of releasing new versions without outside interference. “We do not need new laws or regulations,” Huang said, adding that we should not create a “false choice” between the speed of innovation and product safety.
“Safety is paramount. However, safety is an engineering problem,” Huang added. He assessed that if an AI company leader doubts the safety of their product, the necessary step is to delay its release. “Run as fast as you can, but if at any time you feel the institution is out of control, take a break.”
However, the idea of allowing AI executives to regulate themselves has sparked sharp criticism from several industry figures. Anthropic co-founder Jack Clark argued that an industry without regulation is akin to rolling the dice with immense risk.
Logical Intelligence Chief Operating Officer Patrick Hillman also voiced public scepticism towards technology companies. “The only institution that might be less trusted by Americans than Washington right now is Silicon Valley. I have worked in and lived in both, and I assure you both deserve this sceptical attitude,” said Hillman.
He added, “If you believe what you are building is dangerous, show us what you are willing to stop.”
Currently, several AI company leaders claim to be moving towards industry-level agreements regarding safety standards. Dario Amodei confirmed that Anthropic is in dialogue with other industry players, while OpenAI executive Chris Lehane stated that they are working with other AI laboratories, such as Anthropic and Google DeepMind, to establish voluntary standards, with or without government support.