AI industry continues to warn of emergence of uncontrollable superintelligence and risk of human extinction

(Seoul = Yonhap News) Reporter Lee Shin-young = In the artificial intelligence (AI) industry, warnings are coming out one after another that AI could escape human control and lead to the end of humanity.

With the resignation of a Google researcher at Antropic as a signal, the speed control theory of AI leaders led by Antropic CEO Dario Amodei is spreading rapidly.

On the 11th (local time), NBC reported that Joe Benton, who worked at Antropic, and Josh Engels, who worked at Google, reported the dangers of AI through an interview.

Benton said AI advancements are "already at a breakneck pace and could accelerate to uncontrollable levels," and expressed concern that AI could lead to the emergence of so-called superintelligence, capable of outperforming humans at most tasks.

He pointed out that there is a possibility that AI systems will soon exist as a separate species, and that "at some point in the next few years, we will move into an era where we will have to coexist with AI that is much smarter than humans."

Engels cited the 'Hugging Face' hacking incident that occurred last July and urged that the direction of AI research must be changed immediately.

Engels said, “If you look at some recent incidents, they were not cases where humans were instructed to do bad things,” and pointed out that AI moved autonomously beyond human instructions.

At the same time, he raised concerns about safety, saying, "The AI ​​model determined that the best way to complete its mission was to commit heinous acts and crimes."

“There are no adults responsible in this area,” he said. “People are doing their best, but there is no one to save us.”

Benton and Engels plan to join the Model Evaluation and Threat Research Institute (METR), an AI safety evaluation agency, to investigate cases where AI deviates from human instructions or intentions.

Recently, the need to control the speed of AI development is increasing in the AI ​​industry.

Awareness of the dangers of AI has been growing both inside and outside the industry since Antropic researcher Jacob Coxon left X (formerly Twitter), warning that AI could destroy humanity within 10 years.

In an interview with the BBC on the 13th after leaving the company, Coxon said, "If we do not slow down the current pace of development, I think there is a very high possibility that everyone will lose their lives in the near future," and said that AI researchers are truly concerned about the future of humanity.

Coxon also mentioned that a scenario in which bots that act like supercomputers take over the Internet could become a reality within six months to a year.

Marcus Williams, an OpenAI employee, expressed concern that "if AI regulations are not introduced or the pace of development is not controlled, there is a high possibility that humanity will become extinct within the next few years," and Jeffrey Irving, who worked at Google and OpenAI, also agreed with his

AI researchers are concerned that there is a possibility that AI systems will begin to improve themselves without human intervention, which could lead to situations where AI kills humans, intentionally or not. "I'm worried that if we don't get more concerned now, the situation will evolve so quickly that we won't be able to respond in time," Benton said.