Human computation
PLOW: a collaborative task learning agent
AAAI'07 Proceedings of the 22nd national conference on Artificial intelligence - Volume 2
VizWiz: nearly real-time answers to visual questions
UIST '10 Proceedings of the 23nd annual ACM symposium on User interface software and technology
Real-time crowd control of existing interfaces
Proceedings of the 24th annual ACM symposium on User interface software and technology
Adjunct proceedings of the 25th annual ACM symposium on User interface software and technology
| Hi-index | 0.00 |
When people work together, they converse about their current actions and intentions, building a shared context to inform their collaboration. Despite decades of research attempting to replicate this natural form of interaction in computers, the capabilities of conversational assistants are still extremely limited. In this paper, we investigate how human and machine intelligence can be combined to create assistants that work even in real-world situations. We introduce a crowd-powered conversational interface, called Chorus, that allows users to interact with a group of crowd workers as if they are a single conversional partner. We use Chorus as a personal assistant, and show that our incentive mechanism enables workers to hold consistent conversations and answer 84% of questions accurately. We then discuss a number of potential improvements that can be made by integrating artificial intelligence, and future systems that our work enables.