While many celebrated efforts in Artificial Intelligence aim at exceeding human performance, the real promise of AI in real-world domains, such as healthcare and law, hinges on developing systems that can successfully support human experts. In this talk, I'll share several directions of research we are pursuing towards effective human-AI partnership in the open world, including combining the complementary strengths of human and machine reasoning, addressing concerns around trust, transparency and reliability, and using AI to improve human engagement.

Machine learning models are becoming increasingly opaque to human examination, even to their designers.  Yet these models are also increasingly used to make high-stakes decisions; who goes to jail, what neighborhoods police deploy to, and who should be hired for a job.  But how can we practically achieve accountability and transparency in the face of increasingly complex models?  And how do we know if the algorithmic decisions are fair or discriminatory - what does it mean for an algorithm to be fair?  In this talk, we’ll discuss recent work from the new and growing field of Fairness, Accou