Comparing AI to nuclear facilities, OpenAI calls for global strict control of AI development

According to news on May 24, three senior executives of OpenAI recently jointly issued a document stating that AI needs to be strictly supervised and the development of AI should be managed like nuclear facilities.

(OpenAI official website has uploaded this article)

OpenAI's three top executives believe that super-intelligent AI is a double-edged sword. In the next ten years, the output of AI will catch up with many large companies and must be regulated. They believe that the world needs to quickly establish an international artificial intelligence supervision. mechanism.

They also said that artificial intelligence is innovating so quickly that we cannot expect existing authorities to fully control the technology.

Currently, this article has attracted a lot of attention on Twitter, and even Musk himself commented below saying "Control matters". Many people believe that OpenAI's approach is selfish because it hinders people's vision of the future openness and application of artificial intelligence.

In fact, long ago, Altman, one of the three senior executives of OpenAI, put forward many proposals on AI security when he participated in the U.S. Congressional hearing, including the establishment of a new agency responsible for issuing licenses for large AI models, and Have the power to revoke permission; create a set of safety standards for AI models to evaluate the dangers, including whether they can "self-replicate" and "escape from the laboratory and act on their own"; require independent experts to independently evaluate various indicators of the model audit. This time they once again called the public’s attention to this issue through their corporate identity.

Everyone should be familiar with the "Terminator" movie series. Although the "Skynet" artificial intelligence caused major global destruction in the movie, we cannot underestimate artificial intelligence just because it is a fictional movie. The future is here, and now that AI models such as ChatGPT and Midjourney have shown such powerful capabilities, it may be time to start paying attention to the future development direction of this technology.

The above is the detailed content of Comparing AI to nuclear facilities, OpenAI calls for global strict control of AI development. For more information, please follow other related articles on the PHP Chinese website!