MIT ups the ante in getting one AI to teach another

Researchers at MIT used two machine learning networks working in tandem to improve speech processing via image recognition, an example of "multi-modal" training where vision and sound reinforce one another. It could radically simplify the task of natural language processing.