Google has lost Ian Goodfellow, a foremost expert in AI research and former member of the Google Brain team, to rival company Apple. Specifically, Cupertino poached Goodfellow to be the director of machine learning in the company’s AI-focused Special Projects group. Apple has a number of AI ventures going, such as improvements to Siri and its own self-driving car unit, just to name a couple. Exactly what Goodfellow is up to isn’t being disclosed right now, it would seem, but his background can give some hints.
For starters, Goodfellow started his long and auspicious career as an intern at a now-shuttered robotics startup. This led him to his initial period at Google, during which he eventually ended up as part of the elusive Google Brain team. After that, he had a stint at OpenAI, an AI startup funded by a number of big names in tech. The nonprofit was founded in 2015 with the goal of using AI to benefit humanity and is still open today, but Goodfellow decided there were greener pastures back at Google. At that point, he became a research scientist with the company, moved into a senior role, and finally, just recently, went over to Apple.
Speaking on his achievements, the crown jewel is definitely his implementation of the concept of generative adversarial networks, an AI concept that essentially revolves around using neural networks in contest to compare data and look for favorable outcomes and how to reach them. GANs, as they’re referred to in the industry, are used in a wide variety of machine learning applications due to their scalability and the relatively small amount of human-guided training that they need to begin learning and growing independently.
Considering all of that, one can only assume that Apple is looking to develop and implement AI products that employ different methods of machine learning. After all, if you’re going to use GANs in an AI product, it’s quite helpful to have the guy who designed the concept on board. The company may well be looking at doing something else entirely, though, given Goodfellow’s qualifications as an AI researcher. Apple may even be looking to develop its very own AI convention with his help, though that’s a bit of a stretch.
Since there’s no telling just what Apple’s up to with Goodfellow on board, there’s no way to predict its impact. One thing that is quite possible and quite obvious, though, is the potential for lawsuits. It’s not unheard of by any means for companies to poach top talent from rivals, only to be slapped with trade secret theft lawsuits when those former rival employees put their skills and knowledge to work in their new role. The fact that he was once part of the secretive Google Brain project adds more fuel to that particular fire.
The one aspect of Goodfellow’s knowledge base that’s likely safe is GANs; the centerpiece of Goodfellow’s achievements actually has a basis in research dating well into the 1990s, though his interpretation and implementation, along with the name “generative adversarial network”, all happened during his time with Google. That means while that work has a high chance of being safe due to its derivative nature, some aspects of it, or at least some methods of implementing it into commercial products, are probably tied to Google in some way.