“Developing an data strategy should come before considering AI for your business
If you “don’t have top-down support and understanding for bottom-up initiatives, you’re going to fail miserably because there’s going to be … a longer timeline for ROI.”
Accelerating scale of compute use is “what keeps me up at night.” Advances in optimization and software will be essential to the future of AI, he said. OpenAI found that compute necessary for state-of the-art results has grown 10 times annually since 2012.
Google Cloud Platform AI head Andrew Moore urged developers to skip job offers from companies that have trouble articulating their purpose.”
“Nobody tells the baby that objects are supposed to fall,” said Yann LeCun, the chief AI scientist at Facebook and a professor at NYU…, “a lot of what they learn about the world is through observation.”
While algorithms based on supervised and reinforcement learning are taught to achieve an objective through human input, unsupervised ones extract patterns in data entirely on their own.
“Everything we learn as humans—almost everything—is learned through self-supervised learning.
Ultimately, unsupervised learning will help machines develop a model of the world that can then predict future states of the world, he said. It’s a lofty ambition that has eluded AI research but would open up an entirely new host of capabilities. LeCun is confident: “The next revolution of AI will not be supervised.”