Vergangene Veranstaltungen

Talk by Eric Horvitz at ETH


12.12.2016 16:15 - 12.12.2016 17:15
ETH Zürich, CAB Building, Room G 61
Universitätstrasse 6 - 8092 Zürich

Eric Horvitz, Microsoft Research: Data, Predictions and Decisions in Support of People and Society


ETH Computer Science Department Colloquium


The Computer Science Department of ETZ Zurich hosts Eric Horvitz from Microsoft Research with a talk on Data, Predictions and Decisions in Support of People and Society. More details can be found here



Eric Horvitz

Technical Fellow and Managing Director, Microsoft Research—Redmond Lab


Title: Data, Predictions and Decisions in Support of People and Society

Abstract: I will share directions and results enabled by the confluence of large-scale data resources, jumps in computational power, and advances in machine intelligence. I will focus on efforts that leverage learning and inference to help people with decisions, covering work in transportation, healthcare, and interactive systems. I will present projects that draw from traditional sources of data, such as GPS signals and electronic health records, as well as efforts that harness more exotic streams of information, including the use of planes in flight and anonymized behavioral data drawn from web services. I will conclude by discussing the promise of leveraging data, learning, and reasoning to enable new kinds of collaboration between people and machines to address challenges in the sciences, society, and daily life.

Bio: Eric Horvitz is a technical fellow and director of the Microsoft Research lab at Redmond. He has pursued principles and applications of artificial intelligence, with contributions in decisions under uncertainty, machine learning, bounded rationality, information retrieval, and human computation. His research and collaborations have led to fielded systems in healthcare, transportation, human-computer interaction, operating systems, ecommerce, and robotics. He has been elected fellow of the National Academy of Engineering, the American Academy of Arts and Sciences, the Association for the Advancement of AI (AAAI), the American Association for the Advancement of Science (AAAS), and the Association for Computing Machinery. He received the Feigenbaum Prize and the ACM-AAAI Allen Newell Award for his research contributions. He has served as president of the AAAI, chair of the AAAS Section on Information, Computing, and Communications, and on advisory committees for the National Institutes of Health, the National Science Foundation, the Computer Science and Telecommunications Board, DARPA, and the President's Council of Advisors on Science and Technology. More information can be found at


Two other related talks on AI also take place in the same week:


David C. Parkes

Paulson School of Engineering and Applied Sciences, Harvard University

Title: How to elicit information when it is not possible to verify the answer?

Abstract: We frequently want to elicit information from users; e.g., consider product reviews, peer assessment in online courses, panel feedback about movie trailers, reports on places around a city, or user reports of fake vs real news. But how to score and thus motivate contributions when it is not possible to verify the answer, either because this would be too costly or because there is no objective ground truth? Peer prediction mechanisms address this problem by scoring users based on correlations between reports. But the Achilles heel of these mechanisms has been that they also encourage other, coordinated behaviors that score more than truthful reports. In this talk, I describe the correlated agreement mechanism, which aligns incentives with effort and without introducing new, bad equilibria. I demonstrate its robustness through replicator dynamics and also through quantitative analysis on models derived from edX and Google Local Guides.

Joint work with Arpit Agarwal (U Penn), Rafael Frongillo (CU Boulder), Matthew Leifer (Harvard), Debmalya Mandal (Harvard), Galen Pickard (Google), and Victor Shnayder (Harvard).

Time and place: 11:00-12:00, 13.12.2016, CAB G51, Universitätstrasse 6, Zürich


Eric Horvitz

Technical Fellow and Managing Director, Microsoft Research—Redmond Lab

Title: “The One Hundred Year Study on Artificial Intelligence: An Enduring Study on AI and its Influence on People and Society”

Abstract: I will present an update on the One Hundred Year Study on Artificial Intelligence (AI100). I will describe the background and status of the project, including the roots of the effort in earlier experiences with the 2008?09 AAAI Panel on Long?Term AI Futures that culminated in the AAAI Asilomar meeting.  I will discuss highlights of the report by the 2016 study panel and reflect about several directions for investigation, highlighting opportunities for reflection and investment in proactive research, monitoring, and guidance.  I look forward to comments and feedback from seminar attendees.

Time and place: 10:00-11:00, 14.12.2016, HG D3.2, Rämistrasse 101, Zürich

Kosten: keine
Zurück zur Liste