Microsoft scientists develop AI system for hiring the perfect crowdsourced team
Thu 13 Aug 2015
Microsoft researchers are investigating algorithmic solutions to tackle the crowdsourcing recruitment challenge of hiring effective teams and co-workers for specific tasks.
The project, led by Microsoft computer scientists Eric Horvitz and Pushmeet Kohli and ETH Zurich researchers Adish Singla and Andreas Krause, is looking to develop a computational solution for selecting the most effective team members from a pool of marketplace applicants. The researchers evaluated the methods using synthetic data as well as data collected from the global online collaborative work platform oDesk.
‘The recruiter needs to learn workers’ skills and expertise by performing online tests and interviews, and would like to minimize the amount of budget or time spent in this process before committing to hiring the team,’ reads the published research paper Learning To Hire Teams.
The research aims to make it easier to chose the ‘right’ people for a project – an effective selection of team members increasing performance, productivity, engagement and employee retention. The paper quotes Apple co-founder Steve Jobs: “A small team of A+ players can run circles around a giant team of B and C players.”
The proposed technology involves intelligent algorithms with PAC (probably approximately correct) bounds which analyse information such as skill-specific expertise, reputation, social ties and commonalities to support the recruitment of a ‘near-optimal’ team in line with required budgets.
These algorithms built into computation platforms, the paper suggests, could also facilitate the optimisation of team-building by connecting recruiters and workers across the globe without the need for advance contact.
Considering the future direction of the research, the paper reads: ‘We see promise in extending the results to incorporate more complex relations among team members, such as the matching of task types within teams to balance the workload, capturing diminishing returns of growing teams, learning and representing costs associated with communication and coordination among people with different skills and abilities […] and other combinatorial constraints.’