Teachable reasoning systems can allow QA models to show their work, and allow researchers to more easily correct their outputs. That's why AI2's Aristo team is introducing TeachMe, a framework that could lead to better communication between models and users. With TeachMe, the team aimed to incorporate two attributes: understanding why the system chose an answer via Entailer, and correcting the system's behavior where appropriate by providing NL feedback.
"Common sense, in my view, is the dark matter of intelligence and language." AI2 senior director Yejin Choi was invited to speak with Emily Kwong on NPR's Short Wave about her career, artificial intelligence, and the future.
"The dreams are bigger — it's harder to stay a skeptic." Sergey Feldman, lead applied research scientist for Semantic Scholar, was a guest on Daliana Liu's podcast The Data Scientist Show to talk about machine learning for search, AI for mental health, and more.
Research scientist Kiana Ehsani is pushing the boundaries of indoor robotics on the PRIOR team. Learn about her career journey, her current work, and favorite Seattle past times on the AI2 Blog.
In Semantic Scholar, you can now set your pronouns once you've claimed your author page! We're proud to include this feature to allow authors to represent themselves how they choose.
