Interactive Crowds: Real-Time Crowdsourcing and Crowd Agents

People can now connect instantly via nearly ubiquitous broadband connections, enabling interactive computational systems that are no longer constrained to machine-based automation alone. Instead, they can work in concert with the ondemand labor of people available on the web (the crowd), recruited on-demand and working synchronously together to complete tasks in real-time. The resulting model resembles a Distributed AI (discussed in Chapter ), but with a mix of human and machine agents composing the network. Crowdsourcing workfl ows typically involve dividing tasks into smaller, separable tasks (Little et al. 2010 ; Dai et al. 2010 ). Using independent tasks provides an effective means of leveraging human intelligence to solve discretized problems, such as image labeling, offl ine transcription, handwriting recognition, and more. However, this model cannot handle acquiring consistent input for an ongoing task from workers. In order to expand the power of crowd algorithms, new models have been introduced that present approaches for continuous real-time crowdsourcing. This allows the crowd to be used to generate responses within the few-second time window needed for interactive tasks (Nielsen 1993 ). In these workfl ows, workers are engaged for longer periods of time, allowing them to receive feedback from the system as the task evolves due to their own input, as well as the input of others. Interactive Crowds: Real-Time Crowdsourcing and Crowd Agents