California (Web Desk) — There is a 10% possibility that artificial intelligence (AI) technology could cause the extinction of humanity within the next decade, a safety researcher at AI company Anthropic has warned.
Anthropic researcher Jackob Kosin announced his decision to leave the company and said on X (formerly Twitter) that AI companies are competing to develop self-improving superintelligent systems, putting human lives at risk in the process.
Self-improving systems refer to AI systems capable of continuously increasing their abilities without human intervention. Although such systems do not currently exist, various AI companies are working on developing them.
Jackob Kosin warned that the power of AI should not be underestimated, as it could develop into powerful systems capable of affecting large-scale systems, rapidly transforming sectors, and taking control of authority and resources.
He said AI systems are being developed on the assumption that by the end of this decade, they could acquire the ability to kill humans.
Evan Hubinger, head of Anthropic’s science team, supported Jackob Kosin’s concerns, saying that the possibility of AI causing human extinction does exist. In his personal estimate, there is a 10% chance of this happening within the next decade.
Evan Hubinger said that although Anthropic is making every possible effort to address the risks posed by superintelligence, the company currently has no plan to deal with this potential scenario.
He added that the threat from existing AI models is very low, while the real concern is the emergence of superintelligent systems, progress toward which is happening much faster than expected.