Jan Leike, the Co-Lead of the Superalignment initiative at OpenAI.
8.0 /10
Jan Leike, the Co-Lead of the Superalignment initiative at OpenAI.
As a prominent Research Scientist at Google DeepMind, Jan Leike stands out in the AI alignment community. His impressive credentials include connections with the Future of Humanity Institute and the Machine Intelligence Research Institute. At OpenAI, he plays a pivotal role in leading the Superalignment Team, which focuses on significant advancements in AI, including projects like InstructGPT and ChatGPT. With a PhD from the Australian National University, his research is dedicated to ensuring that AI systems align with human values.
Jan Leike is a co-leader of OpenAI’sSuperalignment Having played a part in developing key AI technologies, including InstructGPT and the alignment of GPT-4, Jan has authored the strategic research plan for the Superalignment Team and has pioneered innovative alignment methodologies at OpenAI. Previously, while at DeepMind, he was involved in prototyping reinforcement learning driven by human feedback. Jan's academic background is rooted in a PhD focused on theories of reinforcement learning.ChatGPT OpenAI is making monumental strides in ensuring the safety of superintelligent AI, with Jan Leike at the forefront of this vital mission.
2023
The Superalignment Team, which has been established by OpenAI under the leadership of Jan Leike, is tasked with the crucial challenge of steering superhuman AI towards beneficial outcomes instead of allowing it to operate independently. Their mission is urgent, as they race against the clock to ensure powerful AI systems are aligned with human intentions before any unaligned entities emerge.
Sam Altman faced dismissal from his position as CEO of OpenAI following a significant internal breakthrough in AI known as 'Q-Star,' which likely pertains to Q-learning.
The terminology surrounding 'precision' and 'recall' often seems counterintuitive to me; I must have revisited their Wikipedia entries countless times.
