Two former employees of Anthropic, a company developing neural networks, have publicly raised alarm over the rapid development of artificial intelligence. The researchers fear humanity may lose control of the technology before it can prepare for the consequences. This was reported by Qazaqyia.kz citing Kursiv Media.
Former head of Anthropic’s safety research team Joe Benton said he left the developer of the Claude chatbot and joined the nonprofit organization METR.
“Last week I left Anthropic to join METR and work on embedded evaluation of AI risks,” Joe Benton said.
METR studies the capabilities of advanced models and tries to determine when artificial intelligence may begin to pose a serious threat to society. In an interview with NBC, Benton admitted that he is especially afraid of the speed at which the technology is developing, Reuters reports.
“I am concerned that artificial intelligence may start developing too quickly and we will not have time to take the necessary measures if we do not think about it now,” Benton said.
However, the alarming admissions from inside one of the largest AI companies did not end there. A few days later, Jacob Coxon, who had also previously worked at OpenAI, announced his departure from Anthropic.
The researcher accused both companies of a dangerous race to create self-improving superintelligence. According to him, developers continue to increase the capabilities of models despite the absence of a reliable way to keep them under control.
“Neither company is acting responsibly. They are racing straight toward self-improving superintelligence and playing with our lives,” Coxon wrote.
He claims that specialists directly involved in creating advanced AI seriously consider the scenario of the destruction of humanity in the coming years.
“The people building AI sincerely believe it could kill all of us by the end of the decade. This is not a marketing trick,” the researcher added.
Coxon has completely left the AI industry, while Benton will continue working on technology safety at an independent organization. At the same time, reports that both specialists left Anthropic within two days are inaccurate: Benton announced his decision on September 2, and Coxon on September 8.