OpenAI Team++ Update

OpenAI has expanded its research team with the addition of several full-time researchers and summer interns, signaling a concentrated effort in generative modeling, reinforcement learning, and unsupervised learning.

Full-Time Research Appointments

OpenAI has recruited several experts in deep learning and generative models to strengthen its core research capabilities:

• Ian Goodfellow: A prominent figure in machine learning, Goodfellow is known for creating the Generative Adversarial Network (GAN), the MaxOut Network, and serving as the lead author of the definitive book on deep learning.
• Alec Radford: The creator of DCGAN, a neural model capable of generating high-detail, coherent images and performing unsupervised image analogies.
• Yura Burda: A researcher specializing in generative models who discovered a fundamental improvement to the variational lower bound.
• Tim Salimans: An expert in variational methods and the author of the first paper on stochastic gradient variational inference, previously ranked number two overall on Kaggle.

Summer Internship Program

OpenAI has brought on a variety of PhD students, professors, and undergraduate researchers for the summer to contribute to diverse AI domains:

• Reinforcement Learning: Peter Chen and Rocky Duan (both former startup co-founders) and Rein Houthooft are focusing on deep reinforcement learning and incorporating uncertainty into RL algorithms.
• Deep Learning Applications: Linxi Fan has contributed to projects including Deep Speech 2, and Jon Gauthier has developed methods for the efficient training of large recursive neural networks on large datasets.
• Robotics and Imitation Learning: Jonathan Ho, a Stanford PhD student, is developing new approaches to imitation learning and previously taught robots to tie knots.
• Theoretical Computer Science: Eric Price, a professor of theoretical computer science at UT Austin and a former IOI perfect scorer, is known for developing a faster-than-n log n sparse Fourier transform.

Strategic Research Focus and Roadmap

OpenAI is currently prioritizing research in unsupervised learning and reinforcement learning. The organization stated that it is not intentionally mysterious about its work but has been focused on the operational launch of the organization and recruitment.

Looking forward, OpenAI expects to share results from these research areas over the coming months and planned engagements at ICLR and via a Quora Session in May or June to provide further transparency into their goals and methodologies.