A growing wave of departures from OpenAI, the leading artificial intelligence research lab, has cast a spotlight on concerns over the company's commitment to AI safety. Jan Leike, a former leader of OpenAI's "superalignment" team dedicated to aligning artificial intelligence with human values, resigned on Friday, citing disagreements with the company's priorities. In a series of posts on X, he accused OpenAI of prioritising product development over the critical issue of AI safety.
"Over the past years, safety culture and processes have taken a backseat to shiny products," Leike wrote. "We are long overdue in getting incredibly serious about the implications of AGI."
Leike's departure comes shortly after Ilya Sutskever, another key figure in the superalignment team, also resigned. Both executives had spearheaded efforts to address the potential risks of artificial general intelligence (AGI), a hypothetical future AI capable of surpassing human intelligence.
These departures have raised alarm bells among AI safety experts and have prompted concerns about a potential shift in OpenAI's focus. Wired reported that the company has disbanded the AI-risk team, absorbing researchers into other departments.
OpenAI CEO Sam Altman acknowledged Leike's concerns, expressing gratitude for his contributions while reiterating the company's commitment to safety. However, Altman's reassurance comes amidst a flurry of high-profile departures, including former Vice President of People Diane Yoon and head of nonprofit and strategic initiatives Chris Clark.
The recent shake-ups at OpenAI are raising questions about the company's priorities and its ability to effectively manage the ethical and societal implications of its powerful AI technology. As OpenAI continues to develop increasingly sophisticated AI systems like GPT-4, concerns about their potential impact on humanity are growing, leaving some experts questioning whether the company is truly prioritising safety.