OpenAI is assembling a team to prevent emerging artificial intelligence technology from going rogue and fueling the extinction of humanity, which the company now fears is a real possibility.
The makers of popular chatbot ChatGPT say AI will power new superintelligence that will help solve the world’s most important problems and be the most consequential tech ever invented by humans.
And yet, OpenAI’s Ilya Sutskever and Jan Leike warned that humans are not prepared, however, to handle technology smarter than they are.
“The vast power of superintelligence could also be very dangerous, and could lead to the disempowerment of humanity or even human extinction,” Mr. Sutskever and Mr. Leike wrote on OpenAI’s blog. “While superintelligence seems far off now, we believe it could arrive this decade.”
If an AI-fueled apocalypse is right around the corner, OpenAI’s brightest minds say they have no plan to stop it.
“Currently, we don’t have a solution for steering or controlling a potentially superintelligent AI, and preventing it from going rogue,” Mr. Sutskever and Mr. Leike wrote. “Our current techniques for aligning AI, such as reinforcement learning from human feedback, rely on humans’ ability to supervise AI. But humans won’t be able to reliably supervise AI systems much smarter than us.”
Mr. Sutskever, OpenAI co-founder and chief scientist, and Mr. Leike, OpenAI alignment head, said they are assembling a new team of researchers and engineers to help forestall the apocalypse by solving the technical challenges of superintelligence. They’ve given the team four years to complete the task.
The potential end of all humanity sounds bad but the OpenAI leaders said they remain hopeful they will solve the problem.
“While this is an incredibly ambitious goal and we’re not guaranteed to succeed, we are optimistic that a focused, concerted effort can solve this problem,” the OpenAI duo wrote. “There are many ideas that have shown promise in preliminary experiments, we have increasingly useful metrics for progress, and we can use today’s models to study many of these problems empirically.”
Mr. Sutskever and Mr. Leike said they intended to share the results of their work widely. They said OpenAI is hiring research engineers, scientists and managers who want to help stop the nerds’ new toys from enslaving or eliminating all mankind.
OpenAI said Thursday that it was making GPT-4, which is its “most capable” AI model, generally available to increase its accessibility to developers.
