OpenAI CEO Sam Altman has conceded that public fears about the dangers of artificial intelligence are warranted, even as he maintains the industry remains capable of managing the risks.
"The world is right. You should be afraid of this," Altman said at a conference in San Francisco. He added that certain accidents may prove unavoidable as the technology advances at breakneck speed. According to Altman, companies in the field must be prepared to slow down or even halt development when they cannot guarantee that continued progress is safe.
His remarks come against a backdrop of increasingly stark warnings from within the AI industry itself. Jacob Coaxum, a researcher who previously worked at both Anthropic and OpenAI, recently resigned while warning that the race to develop AI could endanger humanity. He called for a slowdown in development pace, arguing that the industry is advancing faster than its ability to address the risks.
Jack Clark, a co-founder of Anthropic, also issued warnings in an interview with NPR about a scenario in which AI agents that don't operate according to human goals could break into computers and even attempt to disrupt the internet.
Clark further called for examination of shared safety standards and external oversight of advanced systems, arguing that competition among AI companies makes it difficult for any single company to slow development on its own. "We're rolling the dice with enormous risks," he said, adding that the industry's trajectory needs to change.
