Crowdsourcing has undoubtedly achieved a great deal of success, but coordinating the inputs of many random people, each with disparate aims and ambitions can be a real challenge.
A team from Carnegie Mellon believe that the future may lie in automating the management of the crowd via a system they’re calling the Knowledge Accelerator.
Managing the Crowd
The system utilizes machine learning to sort and organize information created by the crowd to try and turn it into something coherent. It will then use this ‘big picture’ view to create new assignments that aim to fill in the gaps.
The tool is already being tested by Bosch, who have used it to gather diagnostic and repair information for a range of complex projects.
“In many cases, it’s too much to expect any one person to maintain the big picture in his head,” the team said.“And computers have trouble evaluating or making sense of unstructured data on the Internet that people readily understand. But the crowd and the machine can work together and learn something.”
It’s this ability to learn that is a key selling point of the system, with Bosch keen to exploit it to synthesize knowledge and apply it to a range of domains.
During the test period, the automatically curated solutions provided better results than those compiled via sources written by renowned experts in the field.
The results promise to help crowdsourcing to provide its thought diversity at a huge scale without running into diminishing returns as organizations struggle to coordinate and manage the inputs provided by the crowd.
The machine-based approach is especially valuable as it appeared to rapidly gain the subject matter insights required to make smart decisions, which is something that human supervisors often struggle with.
Keeping Humans in the System
Suffice to say, this isn’t a case of humans being scrubbed from the system, and they will remain crucial in deciphering unstructured data, but even here there are signs that automation will begin to make inroads.
As machines gain the ability to recognize patterns in information so that it can categorize and sort it, this kind of automated management of the crowd is only likely to intensify.
Tapping into the wisdom of the crowd is only likely to intensify, so having better means of managing those inputs is likely to be an increasingly important role.