Virtual assistants have come a long way in recent years, but they’re still limited in the ability to interpret the spoken word and match objects with their corresponding descriptions with speed and accuracy. On today’s episode, Adria Recasens from the MIT Computer Science and Artificial Intelligence Lab discusses the research they’re doing to create a system that can outperform Siri and Alexa in terms of correctly and quickly identifying objects based on spoken descriptions, allowing for our interactions with computers to be more like interactions with other humans who speak our language.
Recasens discusses the use cases of this technology, which include higher and better-functioning robots or personal assistants, as well as image analysis in medicine. He also discusses the areas of research they plan to explore in the near future, which include teaching the system different languages and concepts.
Press play to hear the full conversation, and visit https://www.csail.mit.edu/ to learn more.