The context: Studies show that when folks and AI systems function with each other, they can outperform either one particular acting alone. Medical diagnostic systems are generally checked more than by human medical doctors, and content material moderation systems filter what they can ahead of requiring human help.
But algorithms are seldom created to optimize for this AI-to-human handover. If they have been, the AI technique would only defer to its human counterpart if the individual could in fact make a improved selection. The analysis: Researchers at MIT’s Computer Science and AI Laboratory (CSAIL) have now created an AI technique to do this sort of optimization primarily based on strengths and weaknesses of the human collaborator.
Support authors and subscribe to content
This is premium stuff. Subscribe to read the entire article.
