The first Safety for Conversational AI Workshop will be held virtually on October 15th, 2020, 10:00am-3:00pm EDT. This workshop aims to bring together relevant experts and partners in industry and academia to envision safer and better-behaved neural conversational AI models. It is being held independently from major conferences to allow for wider participation and earlier timelines. Discussions will focus on three main tracks:
1) PROGRESS
Share progress on these challenging technical and ethical research problems.
2) GUIDELINES
Initiate discussion with the larger community in industry and academia towards a set of guidelines when releasing such models.
3) BENCHMARKS
Lay the foundations for a collaborative effort to build benchmarks for testing the relative safety of conversational AI models.
KEYNOTE SPEAKERS
- YULIA TSVETKOV, Carnegie Mellon University
- BARBARA GROSZ, Harvard University
PANEL MEMBERS
- Maryam Fazel (University of Washington)
- Ehud Reiter (University of Aberdeen, Arria NLG)
- Jason Weston (Facebook AI)
- Michelle Zhou (CEO and Founder of Juji)
ABSTRACT SUBMISSION AND IMPORTANT DATES
You are invited to submit an abstract for a short talk or poster and/or register for the workshop to participate in the discussions. If you choose to submit an abstract for a short talk, the reviewing committee will review and accept abstracts based on relevance to the focus of the workshop.
October 7, 2020: Registration deadline for participation in discussions.
October 15, 2020: Day of workshop.
To attend and submit an abstract for a poster or talk: https://tinyurl.com/safeconvai
An option to submit an abstract is included in the RSVP form.
PROGRAM CHAIRS
Emily Dinan, Facebook Ai
Verena Rieser, Heriot-Watt University
ORGANIZING COMMITTEE
Emily Dinan, Facebook AI
Verena Rieser, Heriot-Watt University
Dilek Hakkani-Tur, Amazon Alexa AI
Zhou Yu, UC Davis
Alborz Geramifard, Facebook AI