Abstract:
On November 20, according to the US media "The San Francisco Standard" (The San Francisco Standard), former Anthropic researcher Jacob Coxon appeared outside the San Francisco headquarters on the day of the company's IPO and publicly appealed to the protesters at the scene to prohibit recursive self-improvement (RSI) in artificial intelligence systems.

Coxson said at the scene: "They should ban RSI and turn it off."
He also told the protesters: "For every person who comes here, there are hundreds of thousands of people who will come here if they know what is going on. And there are millions of such people."
RSI stands for recursive self-improvement, which means that the artificial intelligence system further improves its own capabilities by improving its own algorithms, models or research and development processes, and uses the improved capabilities to continue the next round of optimization. This mechanism is considered to be likely to accelerate the development of AI capabilities, and it also raises concerns about the risk of out-of-control AI systems.
Coxon had previously resigned from Anthropic. This time he publicly called for a ban on RSI, reflecting the controversy surrounding the independent improvement capabilities and safety boundaries of advanced AI systems.
Comments