Tue. May 24th, 2022

Data2vec is part of a larger trend towards AI Models that can learn to understand the world in multiple ways. “It’s a clever idea,” said Annie Kemvavi of Seattle’s Allen Institute for AI, who works with vision and language. “This is a promising breakthrough in the field of generalized education.”

An important caveat is that although the same learning algorithm can be used for different skills, it can only learn one skill at a time. Once it learns to recognize images, it must learn to recognize speech starting from scratch. It’s hard to give an AI multiple skills at once, but that’s something the Meta AI team wants to see next.

The researchers were surprised that their approach actually performed better than existing techniques in image and speech identification and performed alongside leading language models in text comprehension.

Mark Zuckerberg already Dreaming of a potential Metaverse application. “It will eventually create AR glasses with an AI assistant,” he posted on Facebook today. “It can help you cook dinner, noticing if you’ve missed an ingredient, asking you to reduce the heat or do more complex tasks.”

For Auli, the key is that researchers should get out of their silos. “Hey, you don’t have to focus on one thing,” he says. “If you have a good idea, it can actually help across the board.”

Source link

By admin

Leave a Reply

Your email address will not be published.