AI Company Safety Resignation and Risk Debate
An Anthropic researcher stating that the probability of AI causing the end of humanity is more than 10 percent has heightened safety concerns in the industry.
A researcher working at Anthropic stated that the probability of AI killing all humans within the next decade is more than 10 percent. This statement came after an employee resigned on the grounds that companies are endangering human life.
Statements from the Resigning Researcher
Announcing his resignation by stating that neither Anthropic nor OpenAI are acting responsibly, an Anthropic researcher named Jacob Coxon stated that these laboratories are rushing toward self-improving superintelligence and gambling with human life.
Superintelligence and the Self-Improvement Process
Self-improvement is based on the idea that artificial intelligence systems can improve themselves without human intervention. Although recursive self-improvement is not yet possible, AI laboratories continue their work to achieve this goal.
Admission by AI Safety Expert
Following Coxon's claims, Anthropic alignment science lead Evan Hubinger shared on social media that the assessment made was correct and that the company does not yet have a plan to solve superintelligence alignment.
Control Concerns in the Industry
Concerns that artificial intelligence could spin out of control have been voiced for a long time. Previous vulnerabilities in the OpenAI model and warnings from figures like Elon Musk highlight the magnitude of the risks in the industry.