Mike Schroepfer, Facebook Chief Technology Officer, posted on his news feed this morning his thoughts on the future of artificial intelligence (AI).
From SIRI to self-driving cars, AI is developing fast and dominating the conversations in “cutting edge” computer science thought leadership circles. Many think AI is science fiction magically turning computers into intelligent bots/machines with human-like intelligence. That being said, AI is increasingly embedded in software algorithms for finding content in Google search, discovering cooking recipes using IBM’s Watson, and for autonomous systems such as military drones.
With Schroepfer’s approval via Facebook Messenger I’m reposting below his post, which he shared on his news feed. I love how Facebook is pushing computer science beyond their profit objectives and into the “wild” for all to use and advance. Props to Facebook.
AI Thoughts by Mike Schroeprer – CTO of Facebook
I was in New York this week attending NYU’s Symposium on the “Future of Artificial Intelligence,” which has gathered leading academic and industry researchers to discuss big questions around the field. Today, AI is in a state that is thrilling, yet humbling. Rapid progress has been made on many specific AI tasks such as image recognition, but at the same time it’s still trivial to stump the most advanced of these systems with tasks that a toddler could do with ease.
I spoke to the group about how these challenges led to the creation of the Facebook Artificial Intelligence Research (FAIR) Lab two years ago, and about the unique way that group operates. We recognized from the start that to solve the big challenges in AI research, we needed a separate group that would focus on collaborating in the open with other researchers. Last year FAIR published more than 40 papers and worked with more than 20 researchers from more than a dozen academic and research institutions. We published multiple data sets and lots of code that has been used by others in the field. Most recently we announced that we’re contributing designs of GPU-heavy computing platforms optimized for AI purposes to the Open Compute Project. This work has helped push AI technology forward.
For example, back in December I posted about the advances FAIR has made in basic methods for image recognition that improved accuracy by 60 percent in a few short months. At the conference I learned about the huge diversity of ways in which this fundamental AI technology can be applied for good. Image recognition is being used to build systems that are nearly as good as the worlds best dermatologists at recognizing melanoma, to interpret data from the Kepler satellite to detect planets in other solar systems that can support life, and running in more than two millions cars on the road today, giving them the ability to automatically brake to prevent an accident. This is just a sampling of the ways in which this work can save and improve lives.
At Facebook, these AI technologies are being applied at scale to help connect everyone on the planet. Billions of conversations a day are translated automatically to allow you to talk to friends and family in their native language. Tens of trillions of predictions are generated every day to make sure your News Feed shows you exactly what you want and ensures you never miss an important moment in your friends’ and families’ lives. Our AI systems are used to prevent spam and protect your account. Soon we will be using image recognition to help those of you with visual problems get descriptions of the images your friends are sharing.
It is doubly rewarding to know that because of the way FAIR publishes all of their work, the very same technology we develop for Facebook will be used to help improve science, health care, and make safer cars.
I’m thrilled we are part of this scientific community and I can’t wait to see whats next.
– Mike Schroepfer