OpenAI isn’t being shy about its ambitions. The research outfit just announced a wave of new hires that reads like a who’s-who of machine learning’s most inventive corners. We’re talking Ian Goodfellow, the mind behind Generative Adversarial Networks (GANs), and Alec Radford, whose DCGAN work made those eerily coherent AI-generated images possible. They’re joined by Tim Salimans, an authority on variational methods who once held the number two global ranking on Kaggle, and Yura Burda, a math PhD who found a fundamental improvement to a decades-old statistical technique only a year and a half after switching into the field.
The summer intern class is equally stacked. It includes Peter Chen and Rocky Duan, a duo that previously co-founded a startup together before diving into deep reinforcement learning, and Jonathan Ho, a Stanford PhD student who literally taught robots to tie knots. Eric Price, a theoretical computer science professor with a perfect IOI score in his past, is also coming aboard. The concentration of talent here is a clear signal that OpenAI is consolidating expertise to push the boundaries of what’s possible.
So what will they all actually be doing? The organization is staying tight-lipped for now, though they deny any intentional mystery, chalking the silence up to the chaos of launching. They’ve confirmed a dual focus on unsupervised learning and reinforcement learning, with promises of “interesting results” in the next month or two. For now, we know they’re building the team. The output will have to speak for itself.
It’s a gathering that feels reminiscent of early Bell Labs or Xerox PARC—a deliberate clustering of high-variance individuals around a shared, unspecified goal. Whether that goal is as grand as artificial general intelligence or something more immediately applicable remains the multi-billion-dollar question. The only immediate promise is a Quora session with co-chair Sam Altman in May or June, where we might finally get some specifics beyond the impressive resumes.
