Rohin Shah Publishes Blog Posts on Learning Preferences

14 Feb 2019

CHAI grad student Rohin Shah published two blog posts on his recent paper Preferences Implicit in the State of the World. The first was Learning Preferences by Looking at the World, published on the Berkeley AI Research (BAIR) Blog. The second blog post was an article of the same name that was published on the AI Alignment Forum. This article went into more of the details and background of the original paper.

Joe Halpern Elected to National Academy of Engineers

13 Feb 2019

CHAI Professor Joe Halpern received one of the highest honors in the field of engineering, joining the National Academy of Engineering for his outstanding contribution to “reasoning about knowledge, belief, and uncertainty and their applications to distributed computing and multiagent systems.” Read more here.

Daniel Filan Publishes Blog Post on Impact Measures

CHAI grad student Daniel Filan published a blog post entitled Test Cases for Impact Measures on the AI Alignment Forum.

CHAI Papers at FAT* 2019

29 Jan 2019

CHAI grad student Smitha Milli attended the ACM Conference on Fairness, Accountability, and
Transparency (ACM FAT*) in Atlanta, Georgia.

Stuart Russell Gives Invited Talk to Netherlands Ministry of Foreign Affairs

21 Jan 2019

Professor Stuart Russell gave a talk to the Netherlands Ministry of Foreign Affairs on the risks of autonomous weapons. Professor Russell has been involved in bringing awareness to the risks that autonomous weapons have, calling them the New Weapons of Mass Destruction due to their ability to scale with a small number of human operators.

Former CHAI Intern Wins AI Alignment Prize

20 Jan 2019

Alex Turner, a former CHAI intern, was announced as one of the two winners of the AI alignment prize for his paper on Penalizing Impact via Attainable Utility Preservation. The award is given as the prize of a competition for submitted papers that further understanding in the field of AI safety. More information can be found here.

CHAI Papers at the AAAI 2019 Conference

17 Jan 2019

Many of our staff and faculty submitted papers to the 2019 Association for the Advancement of Artificial Intelligence Conference. Below is a list of the papers and authors who are CHAI faculty:

Stuart Russell Receives AAAI Feignenbaum Prize

15 Jan 2019

The Association for the Advancement of Artificial Intelligence (AAAI) announced that CHAI PI Stuart Russell is the winner of the 2019 Feigenbaum Prize, and was awarded at the AAAI conference in Honolulu in January “in recognition of (his) high-impact contributions to the field of artificial intelligence through innovation and achievement in probabilistic knowledge representation, reasoning, and learning, including its application to global seismic monitoring for the Comprehensive Nuclear-Test-Ban Treaty.” The Feigenbaum Prize is awarded biennially to recognize and encourage outstanding Artificial Intelligence research advances that are made by using experimental methods of computer science.

Anca Dragan Hosts Workshop on Machine Learning for Intelligent Transportation at NeurIPS 2018

Over 150 experts from all areas attended CHAI PI Anca Dragan’s workshop to discuss the challenges involved with machine learning and transportation, such as coordination between both between vehicles and between vehicles and humans, object tracking, and more.

Rosie Campbell Speaks About AI Safety and Neural Networks at San Francisco and East Bay AI Meetups

08 Jan 2019

CHAI Assistant Director Rosie Campbell gave a talk at the San Francisco AI Meetup to explain in basic terms what a neuron is, how they combine to form a network, why convolutional neural networks have had so much attention, and the challenges and opportunities of AI progress. See the slides here and a write-up of an earlier version of the talk here. Rosie also sat down for a chat with tech blogger Mia Dand on “Keeping AI Safe and Beneficial for Humanity”. Their discussion covered topics such as CHAI’s mission and approach, the problem of AI alignment, and why you might or might not believe AI safety research is important.