On June 28th, 2018, Effect.AI launched its Effect Force platform in private beta on the NEO MainNet. 200,000 image classification tasks were completed by workers in just 48 hours. Effect Force is a marketplace that connects AI developers to paid contributors. Its primary purpose is to use human intelligence to train AI algorithms through the completion of tasks that cannot be done by computers.
600 workers were given access to the platform and the first batch of microtasks, which was to classify images as a “selfie” or “not a selfie”. The tasks were supplied by Lobster.Media, a user-generated content licensing marketplace, as it requires this human distinction to help train its AI algorithms to distinguish the two types of photos.
Workers were limited to 500 tasks per days to ensure all of the private beta testers had a chance to complete some tasks. Effect.AI said that “Despite this limit, we were amazed to see how quickly tasks were being completed” and that the volume of tasks completed “exceeded our expectations and showcases the strength of our workers and the capacity of Effect Force.” The 600 workers collectively earned 20,000 EFX tokens for their work.
Effect.AI reported that the Effect Force private beta launch was a “great success” and that no major problems were encountered. In addition, the workers supplied Effect.AI with feedback that is already being implemented in the form of improvements by the development team.
The Effect Force private beta was said to be a “very small-scale version of what Effect Force will grow into” and that the usability and quality will be improved with each release, along with more images for testers to work through.
More information on Effect.AI can be found at the links below.