Jan Leike, the former head of OpenAI’s alignment and “superalignment” initiatives, took to Twitter (aka X) on Friday to explain why he left the AI developer on Tuesday. In the tweet thread, Leike cited a lack of resources and attention to security as reasons for his decision to resign from the maker of ChatGPT.
OpenAI’s alignment or superalignment team is responsible for security and creating more human-centric AI models.
Leike’s departure marks the third high-profile member of the OpenAI team since February. On Tuesday, OpenAI co-founder and former chief scientist Ilya Sutskever also announced that he will be leaving the company.
“Stepping away from this job was one of the hardest things I have ever done,” Leike wrote. “Because we urgently need to understand how to guide and control AI systems much smarter than we are.”
Yesterday was my last day as Head of Alignment, Head of Superalignment, and @OpenAI Executive.
— Jan Leike (@janleike) May 17, 2024
Leike noted that while he thought OpenAI would be the best place to do AI research, he didn’t always agree with the company’s leadership.
“Building machines that are smarter than humans is an inherently dangerous undertaking,” Leike warned. “But in recent years, safety culture and processes have taken a back seat to brilliant products.”
Noting the dangers of artificial general intelligence (AGI), Leike said OpenAI has a “huge responsibility,” but said the company is more focused on achieving AGI and not security, noting that his team “has sailed against the wind” and fought for computing resources.
Also known as singularity, artificial general intelligence refers to an artificial intelligence model that can solve problems in various areas as a human would, as well as having the ability to self-learn and solve problems for which the program is not been trained.
On Monday, OpenAI revealed several new updates to its flagship generative AI product, ChatGPT, including the faster and smarter GPT-4O model. According to Leike, his former team at OpenAI is working on several projects related to smarter AI models.
Before working at OpenAI, Leike worked as an alignment researcher at Google DeepMind.
“It has been a truly wild journey these past 3 years,” Leike wrote. “My team launched the first ever [Reinforcement Learning from Human Feedback] LLM with InstructGPT, the first scalable aid on LLMs has been published, [and] pioneered automated interpretability and weak-to-strong generalization. More interesting things will come out soon.
According to Leike, a serious debate about the implications of achieving AGI is long overdue.
“We need to prioritize preparing for them as best we can,” Leike continued. “Only then can we ensure that AGI brings benefits to all humanity.”
While Leike didn’t mention any plans in the thread, he encouraged OpenAI to prepare for when AGI becomes a reality.
“Learn to feel the AGI,” he said. “Act with the appropriate gravity for what you are building. I believe you can “transmit” the necessary cultural change.”
“I’m counting on you,” he concluded. “The world is counting on you.”
Leike did not respond immediately Decipher them request for comment.
Edited by Andrew Hayward
