Key figures in the artificial intelligence (AI) industry confronted each other directly over the pace of AI development at the same event. Dario Amodei, CEO of Anthropic, emphasized the need to slow down AI development due to its potential risks. In contrast, Jensen Huang, CEO of Nvidia, argued that both AI safety and rapid progress can be achieved without new regulations.

Reuters Yonhap News Agency

Reuters Yonhap News Agency

View original image

◆ Amodei: "A Safety Management System Like Automobiles Is Needed" = On September 15 (local time), at Salesforce's annual 'Dreamforce' conference held at the Moscone Center in San Francisco, Amodei stressed the need to moderate the speed of AI development—a stance he termed "speed moderation." He reiterated three proposals previously outlined in his blog post: to have independent external auditors within AI companies, to establish common safety standards among democratic countries, and to ultimately form a global cooperative system for AI safety.


Amodei illustrated how the automotive industry handles safety concerns as a model for the AI sector. Rather than using incidents like brake defects at rival carmakers as ammunition, he argued that the entire industry should pause to inspect its own safety management systems when such issues arise.

"Amodei: 'AI Development Should Slow Down' vs. Huang: 'No Need for New Regulations'" View original image

He said, "There is a strong temptation to criticize competitors, to say, 'That company is not safe,'" but continued, "A more responsible approach is to examine our own safety records first." He added, "We may not have experienced a major accident, but we are by no means perfect."


Additionally, Amodei highlighted the much faster-than-expected pace of AI progress. Based on a decade of AI development experience, he cited the speed of advancement as the most surprising aspect, noting that not only the technology itself but also its economic impact has grown much more rapidly than anticipated.


However, Amodei also emphasized the substantial room for the use of current technologies. He assessed that even if AI development stopped immediately, the value currently being realized is only 5–10% of the potential value AI can deliver, meaning there remains massive untapped potential in existing technologies alone.

Reuters Yonhap News

Reuters Yonhap News

View original image

◆ Jensen Huang: "Speed Control Possible Without New Laws or Regulations" = Taking the stage afterward, Jensen Huang countered the calls for regulation by asserting that market mechanisms and companies' autonomous decisions are sufficient for moderating AI development speed. "We don't need new laws. We don't need new regulations," he said. He also called the idea that one must choose between safety and speed in AI development a "false choice," and emphasized, "We can absolutely achieve both at the same time."


However, Huang did not deny the importance of AI safety itself. "Safety is the most important issue in many aspects. It is the top priority," he said, but added that "safety is an engineering problem." He argued that companies, rather than governments, should solve safety issues through technology.


Huang explained that companies should develop AI as rapidly as possible, but must moderate their pace if they feel unsure about safety. He said, "Run as fast as you can," but added, "At any point, if your company feels it is out of control or that your product may be unsafe, you need to pause and make sure things are right." He further asserted that companies should adjust their releases until they are confident the product can be accepted by the market, stressing that "the power of the market is already in action."



Huang also dismissed concerns over job destruction and the so-called "SaaSpocalypse," referring to predictions of the "end of software" due to AI, as "nonsense." He argued that the fears surrounding AI are exaggerated and reiterated that the use of AI will in fact increase both jobs and software usage.


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing