Error message

Warning: Use of undefined constant openphil_social - assumed 'openphil_social' (this will throw an Error in a future version of PHP) in openphil_social_block_block_view() (line 90 of /var/www/html/openphil/live/sites/all/modules/custom/openphil_social_block/openphil_social_block.module).

UC Berkeley — AI Safety Research

Organization Name 
Award Date 
Grant Amount 
To support a four-year AI safety project led by Anca Dragan and Sergey Levine.

Published: October 2017
Grant investigator: Daniel Dewey

This page was reviewed but not written by the grant investigator. UC Berkeley staff also reviewed this page prior to publication.

The Open Philanthropy Project recommended two gifts totaling $1,450,016 to the University of California, Berkeley to support a four-year research project on AI safety. The work will be led by Professors Sergey Levine and Anca Dragan, who will each devote approximately 20% of their time to the project, with additional assistance from four graduate students. They initially intend to focus their research on how objective misspecification can produce subtle or overt undesirable behavior in robotic systems, though they have the flexibility to adjust their focus during the grant period.

Our broad goals for this funding are to encourage top researchers to work on AI alignment and safety issues in order to build a pipeline for young researchers; to support progress on technical problems; and to generally support the growth of this area of study.

This funding falls within our focus area of potential risks from advanced artificial intelligence.


Levine and Dragan, Project Narrative, 2017Source