Why not announce the details of ChatGPT training? OpenAI co-founder: It's easy to cause huge harm

News on March 17, this week OpenAI released GPT-4. While the AI ​​chatbot continues to add new features, concerns are growing that the AI ​​tool could also be used for nefarious purposes. OpenAI co-founder and chief scientist Ilya Sutskever admitted in an interview that people with ulterior motives can easily use artificial intelligence to cause harm to others.

“These models are very strong and will get stronger,” he said. "In a way, it's very easy to do a lot of harm with these models if someone wanted to."

Sutskover was explaining why OpenAI no longer provides details on how to train these models. Said so.

He said in an interview: "As the performance of artificial intelligence continues to improve, there is a reason not to disclose them." "I am sure that within a few years, it will be very clear to everyone that artificial intelligence is open source It’s unwise.”

OpenAI CEO Sam Altman has expressed similar views in the past.

In an interview earlier this year, Altman said that when it comes to artificial intelligence, the best-case scenario is "so incredibly good that it's hard for me to even imagine." , but the worst-case scenario is "turn us all off."

Last month, Altman also said on Twitter that he believed artificial intelligence tools could help people become more efficient, healthier, and smarter, but also added that the world may be close to terrifying AI tools may not be that far away, so regulating them is "vital".

The above is the detailed content of Why not announce the details of ChatGPT training? OpenAI co-founder: It's easy to cause huge harm. For more information, please follow other related articles on the PHP Chinese website!