SILICON VALLEY — The chief scientist of OpenAI, the developer of ChatGPT, said artificial intelligence development needs to slow down. OpenAI Chief Executive Sam Altman voiced agreement and support.
Jakub Pachocki, OpenAI's chief scientist, wrote in an essay posted on the company's website on the 6th that no one is prepared for the consequences that would follow if machine intelligence continues to advance rapidly.
Pachocki said modern AI is "grown" rather than "designed," noting that it is difficult for humans to fully understand how AI works. The point is that because AI grows like a living organism, humans cannot fully control it.
Pachocki criticized an industry argument that AI development must continue in order to build defensive systems, saying it is being used as a pretext to justify reckless behavior. He stressed that international cooperation on AI development should become the top priority for governments around the world.
Pachocki said no laboratory has yet sufficiently solved the "alignment" problem of making AI refuse actions that threaten humans, adding that he expects and hopes voluntary slowdowns will become common until shared safeguards are in place.
Altman also cited Pachocki's essay on the social media platform X, formerly Twitter, calling it an "important" piece and expressing his support. Altman has shifted his position since the recent "Hugging Face" hacking incident involving GPT models and now agrees on the need to slow down.
