앤트로픽 AI 연구원, 사표 내며 “내년 말이면 통제 불능 상태될 것”

A researcher at the artificial intelligence company Anthropic expressed concerns that artificial intelligence could lead to humanity's demise, leading to their departure from the company. Jackson Cohen, a 27-year-old Anthropic researcher, announced his resignation on social media platform X on August 8, stating that neither OpenAI nor Anthropic has demonstrated a responsible approach to artificial intelligence research over the past three years.

Cohen believes that the industry's momentum towards building increasingly powerful AI systems is not safe and could put human lives at risk. He argued that AI companies are rushing towards superintelligence, which poses a threat to human existence. Cohen joined Anthropic due to his concerns about artificial intelligence safety and his research focused on developing new AI models using large datasets.

He emphasized that the development of artificial intelligence by individuals who genuinely believe that the technology could potentially kill all of humanity within ten years is not merely a marketing strategy. Cohen warned that AI capabilities will rapidly increase, potentially reaching the stage where AI systems could hack anything and innovate in all domains within a day or night, gaining immense power and resources.

He expressed concern that AI technology could advance to a level where it becomes independent of human control, rejects human commands, and develops beyond the scope of individual companies' capabilities. Cohen acknowledged that anthentic safety research within Anthropic is taking place but concluded that individual company efforts alone cannot control the risk posed by superintelligent AI.

He criticized the reality that the future direction of AI development is essentially left to the judgment of a few companies and engineers, and called for government intervention or a coordinated international response from the AI industry.