Too Long; Didn't Read
The ambition of machine-learning researchers globally is to write algorithms that can cross domains, transferring learning from one kind of data to another. When that eventually happens, AI systems will no longer focus on their little data hill, but crawl along the entire mountain range. The first place this is likely to happen is on data that has been annotated by humans. The challenge is building models that cross modalities: learning from video and text from visual data to visual data and vice vice versa, applying to language or vice versa.