Aus dem besten KI-Labor der MIT: Der geniale chinesische Forscher von OpenAI hat seinen Doktorgrad erworben.
[Introduction] He was exposed to deep learning in high school. During his undergraduate years, he started a robotics business, interned to contribute to the development of Gemini 2.0, and straddled the dual fields of AI and philosophy... Now, he completed his PhD at MIT in less than four years and passed his dissertation defense. At OpenAI, he will continue to advance the "world model" - a cutting-edge technology that could reshape the path of general artificial intelligence.
He completed his PhD at a top AI lab in less than four years, minored in philosophy along the way, was a member of the core five-person research team for GPT image generation, and is a member of the Sora video generation model team at OpenAI...
Just recently, Boyuan Chen, a Chinese research scientist at OpenAI, completed his PhD dissertation defense at MIT!
He said excitedly:
I'm very excited to continue advancing the development of the world model in the industry - now I'm part of the GPT image generation and Sora video teams.
There's nothing more exciting than seeing your own research change the paradigm of the field!
At such an important moment, he naturally thanked his mentors, relatives, and friends and received congratulations from everyone.
Finally, he emphasized: The visual world model will be crucial for embodied intelligence.
In addition, he promised to share knowledge with the community as always.
A Chinese Genius Aims at the World Model
Boyuan Chen, currently a research scientist at OpenAI, is one of the five researchers responsible for training GPT image generation technology and is also a member of the Sora video generation team.
He holds a PhD in Electrical Engineering and Computer Science (EECS) from the Massachusetts Institute of Technology (MIT) and minored in philosophy.
His research focuses on world models, embodied artificial intelligence, and reinforcement learning.
He believes that by combining these fields, AI can better understand and interact with the physical world.
From May to August 2023, during his internship at Google DeepMind, he studied under Dr. Fei Xia.
At DeepMind, he mainly participated in the training project of a multimodal large language model (MLLM) based on large-scale synthetic data; he built a complete data synthesis pipeline, and his instruction fine-tuning technology was later adopted by Gemini 2.0.
During his PhD defense, Boyuan Chen specifically thanked his mentor Fei Xia at DeepMind.
When Boyuan Chen was still a high school student, he participated in a summer camp.
This was the first time they met. Fei Xia introduced deep learning to Boyuan Chen - at that time, he didn't even understand Python and NumPy.
This was the starting point of his journey into the field of AI. Fei Xia was like his "Andrew Ng".
Fei Xia invited him to Google DeepMind for high-quality internships twice.
In the first year of his PhD, Boyuan Chen was in a slump because he hadn't produced any papers. This was the most difficult stage of his PhD, and Fei Xia helped him publish his first popular research, NLMap.
Project address: https://nlmap-saycan.github.io/
After that, they also collaborated on SpatialVLM.
Paper link: https://arxiv.org/abs/2401.12168
Many of his published papers have been recognized in both academia and industry, including "Diffusion Forcing", "SpatialVLM", and "History Guidance".
Committed to General Robots
In last year's blog, he made an optimistic judgment about embodied intelligence:
I can responsibly tell everyone that embodied intelligence will definitely be the most exciting technology in the next hundred years, and we have a good chance of witnessing the birth of general robots in our lifetime.
At the same time, he also hopes to see society make long - term and steady investments in the development of general robots -
To see researchers, as my mentor Russ said, "conduct research with results as the goal, not be guided by viral videos";
To see governments and investors be optimistic about embodied intelligence in the long term without blindly believing in large robot models just because of the financing needs of hardware companies;
To see entrepreneurs forge ahead and pave the way for real general robots with success in niche fields.
At the end of his essay, he said, "I'm also willing to spend my whole life bringing real general robots to the world."
It has been reported that OpenAI has increased its efforts in robotics technology in the race towards general artificial intelligence (AGI). It is forming a team capable of developing algorithms to control robots and seems to be hiring robotics experts specializing in humanoid robot research.
From a Prestigious School, Well - Versed in Arts and Sciences
From 2021 to 2025, he pursued his PhD at the MIT Computer Science and Artificial Intelligence Laboratory (MIT CSAIL), under the supervision of Professors Russ Tedrake and Vincent Sitzmann.
From 2017 to 2021, during his undergraduate years at the University of California, Berkeley, he studied under Professor Pieter Abbeel, a well - known figure in the field of robotics, and obtained dual degrees in the Honors Program of Computer Science (EECS Honors Class) and Applied Mathematics.
He graduated from the University of California, Berkeley, majoring in computer science and mathematics, and he also studied philosophy at Berkeley for one year.
From November 2017 to March 2020 during his undergraduate years, he founded a robotics education company for primary and secondary schools, leading the development of software and hardware for robotics kits for competitions, and the products were directly targeted at participating students.
Reference materials:
https://www.boyuan.space/
https://www.boyuan.space/blogs/jushenzhineng.html
https://www.linkedin.com/feed/update/urn:li:activity:7373764756475637760
https://www.linkedin.com/feed/update/urn:li:activity:7360784982740537344/
This article is from the WeChat official account "New Intelligence Yuan". Edited by KingHZ. Republished by 36Kr with permission.