A Conversational Agent
Powered by Crowdsourcing

Chorus is a Google Hangouts chatbot powered by crowdsourcing. A group of crowd workers will be recruited on-demand in real-time to chat with you and try to solve your problems. You can now use Chorus on Google Hangouts for anything anytime, anywhere.

Try It Now >>


FAQ

  1. Are real humans responding to me?
    Yes. Mostly. We will add some automated components to Chorus, but still, they are actual humans.
  2. How do you recruit them? Do they just sit there and wait for us?
    Currently we recruit workers from Amazon Mechanical Turk (MTurk) to talk to you in real-time.
  3. Who are the workers? Do they get paid?
    For the workers from MTurk, yes, they get paid. If you're interested in the demographics of MTurk workers, please take a look at mturk tracker.
  4. How many workers am I talking to at the same time?
    It depends. We try to have up to 5 workers to talk to you at the same time, but sometimes we only manage to get one worker.
  5. Why not just having one worker per conversation?
    We have multiple workers to vote on each other's responses to control the chat quality.
  6. Would the workers know who I am?
    No. Workers only know that you are a "user" on Google Hangouts. (They can't even see your displayed name.)
  7. Do I need to tell Chorus any of my personal information?
    No. You don't need to disclose any of your information to Chorus unless you want to.
  8. What can I ask Chorus?
    We are exploring the possibilities of Chorus. So, you can be creative. You can say anything reasonable with Chorus. -- Just don’t abuse Chorus. Please keep in mind, they are actual humans.
  9. Can I send images to Chorus?
    Yes. You can send images to Chorus via Google Hangouts, and the crowd workers are able to see the image you sent.
  10. Can I start a video call with Chorus?
    No. We didn't enble video calls for Chorus.
  11. Is that you? Are you the person who is talking?
    No. It's not me, nor any of our research group members.

Papers

  • This deployment is primarily described in the following paper:
    "Is there anything else I can help you with?": Challenges in Deploying an On-Demand Crowd-Powered Conversational Agent
    Ting-Hao K. Huang, Walter S. Lasecki, Amos Azaria, Jeffrey P. Bigham.
    In Proceedings of Conference on Human Computation & Crowdsourcing (HCOMP 2016), 2016, Austin, TX, USA.
  • The Chorus system framework is based on the following paper:
    Chorus: A Crowd-Powered Conversational Assistant
    W.S. Lasecki, R. Wesley, J. Nichols, A. Kulkarni, J.F. Allen, J.P. Bigham.
    In Proceedings of the ACM Symposium on User Interface Software and Technology (UIST 2013). St Andrews, UK. p151-162.
  • "Guardian" is one of the follow-up works we developed to automate Chorus:
    Guardian: A Crowd-Powered Spoken Dialog System for Web APIs
    Ting-Hao K. Huang, Walter S Lasecki, Jeffrey P Bigham.
    In Proceedings of Conference on Human Computation & Crowdsourcing (HCOMP 2015), pages 62–71, November, 2015, San Diego, USA.

Team

Ting-Hao (Kenneth) Huang
Carnegie Mellon University
Walter S. Lasecki
University of Michigan
Amos Azaria
Ariel University /
Carnegie Mellon University
Jeffrey P. Bigham
Carnegie Mellon University

Contact

Any questions, suggestions, proposals, opinions, or user feedbacks, please contact Ting-Hao (Kenneth) Huang at tinghaoh@cs.cmu.edu.

This field deployment study of the "Chorus" system is part of a research study conducted by Prof. Jeffrey P. Bigham and his group at Carnegie Mellon University.