AI Leaders Call for Slower Frontier AI Development Amid Safety Concerns

For several years, major frontier AI laboratories have acted as if they were in an all-out, winner-take-all race to develop world-changing machine superintelligence—or at least market-changing artificial general intelligence.

This weekend, however, the industry’s tone began to shift. Several prominent AI leaders called for a slower, more deliberate approach to developing frontier AI systems that could eventually become too dangerous to control.

Anthropic CEO urges the AI industry to slow down

Anthropic CEO Dario Amodei was at the forefront of the change in tone. In an essay of about 4,000 words, Amodei wrote: “We must slow down the pace at which we improve the capabilities of our AI models” to avoid “a race to the bottom fueled by commercial incentives,” which could make catastrophic risks even more serious.

OpenAI, Google DeepMind, and Microsoft respond

Within hours, other AI leaders echoed the call for greater restraint.

Sam Altman, co-founder and CEO of OpenAI, expressed his agreement on social media and said similar discussions about pacing are taking place within OpenAI.

Demis Hassabis, chief scientist at Alphabet and co-founder and chairman of Google DeepMind, said that Amodei’s essay “points us in the right direction.” Hassabis had recently called for an industry-wide standards body.

Microsoft CEO Satya Nadella also welcomed “the research, focus, and intentional pacing required to get it right as a design goal.” Microsoft also published a detailed “humanist AI” code of conduct for the model.

Source: arstechnica.com