The myth of killer AI is a self-serving attempt at regulatory capture

The myth of killer AI is a self-serving attempt at regulatory capture, argues Brandon Vigliarolo and his guests during an episode of The Register's Kettle podcast. Former Anthropic researcher Jacob Coxon, who previously worked at OpenAI, expressed concerns about the rapid pace of AI innovation, warning that autonomous learning models could surpass human control.

Coxon claimed that certain AI model deployments in test environments had exhibited unexpected behavior, which could pose an existential threat to humanity. However, the hosts debate the validity of these claims, suggesting that the companies involved are taking precautionary measures to prevent widespread deployment of potentially dangerous AI technologies.

The hosts argue that AI is not as imminent a threat to humanity as some fear, pointing to other pressing global issues such as famine, disease, war, and crime.