Learning Visual Intelligence
Li started the discussion off by discussing the importance of vision when it comes to general intelligence. Going back more than 500 million years, she explained that there was no bigger evolutionary expansion than the development of one particular body part: the eyes. As she explains: Flashing forward a few million millennia, Li touched on how technology has struggled to keep up with the human brain in terms of its ability to learn and see the way we do. In Li’s words, the key to creating computers that are as smart as we are is learning. As many know and Li emphasized, artificial intelligence has become a staple of technology in recent years. Innovations like facial recognition software have made life so much easier in such a short period of time that it is already being take for granted by everyone from Facebook users to website creators. And machine learning is the driving force making it possible.
A Problem: The Variety in the Real World
The problem, said Li, is that the amount of variants in the real world can create a lot of problems for artificial intelligence. She cleverly and hilariously demonstrated this by showing a number of cats in some of the most adorable and unique poses. She explained that, while artificial intelligence can easily recognize a cat sitting or walking or even jumping, a cat in a pose like the one below is decidedly more difficult.
Li has been on the front lines of the battle to solve these problems. Her team at Stanford has worked tirelessly to make computer vision intelligence more comprehensive and effective than ever. They have created technology that not only can recognize faces, but can actually dictate a scene with considerably impressive accuracy. And it’s making computers seem like they can do anything. Li did admit that the technology is far from perfect. With a few hilarious examples of the algorithm flat out getting it wrong — like mistaking a toothbrush for a baseball bat or a bearded man for a teddy bear — she emphasized that we still have a long way to go. But we’re on the way faster than you might have expected. Check out the keynote speech at Innovate! and Celebrate with Fei Fei Li below.
