Dario Amodei, CEO of Anthropic, has called for slowing the pace of artificial intelligence development and emphasized the need for greater control over the evolution of this technology.
According to Agence France-Presse, Amodei on Saturday, in a letter published on his personal website, warned about the speed of AI progress and stressed the necessity of better oversight of the technology's development.
The publication of this letter came just days after Jacob Coxon, a researcher at Anthropic, announced his resignation from the California-based company. Coxon had accused Anthropic of downplaying AI’s existential threats to humanity.
These developments have again drawn attention to the late theoretical astrophysicist Stephen Hawking’s warnings about the consequences of uncontrolled AI development; especially at a time when rapid technological progress has generated widespread concerns about its social and security implications.
Hawking, who died in 2018 at age 76, had previously warned about the dangers of AI development. He believed that full development of the technology “could mean the end of the human race” and ultimately surpass humans.
He also warned that machines could improve themselves and continue to redesign themselves at an accelerating rate; a process he believed humans would not be able to compete with or keep up with.
Resignation of a researcher and intensifying warnings
Concerns about the speed of AI development intensified after Jacob Coxon’s public resignation. He stepped down from his role in Anthropic’s pretraining research division; a researcher who had previously worked at OpenAI.
After resigning, Coxon posted a series of messages on the platform X accusing both Anthropic and OpenAI of “irresponsible inaction, racing to achieve self-improving AI, and gambling with human lives.”
In an interview with the Wall Street Journal he also warned that some developers believe such technologies could lead to the death of all humans by the end of this decade. Coxon also expressed concern about the possibility of encountering uncontrollable models by the end of next year.
According to him, future AI supersystems may gain the ability to infiltrate various systems, cause sudden changes, and rapidly acquire power and resources.
Amodei’s remarks come as the debate over the pace of AI development and the need to establish mechanisms to control its potential risks has received greater attention from executives and researchers in the field than before.