Artificial Intelligence is going to be integrated into our lives more than we can imagine, but it is us, humans, that need to teach AI in the first place in most of the cases. That is exactly what you will need to do today, March 2nd, at 19:00 (GMT+1) when the 9th live demo event will start on Eventum Alpha.
You need to join until 18:40 (GMT+1), otherwise smart contract will not accept your application! Please read the tutorial to learn how to join events, report data and collect rewards.
How?
When the event starts (at exactly 19:00!) an image will appear where you will need to recognize different objects that AI recognized but couldn’t categorize. You will report this by selecting a correct value of the category in the drop-down menu for each object on the image.
Why?
AI market is growing with 57.2% annual growth rate and it will be one of the most important factors of our lives in the future. AI can learn on its own, but so far the most effective algorithms are of the “[supervised machine learning](http://(https//en.wikipedia.org/wiki/Supervised_learning)” type. This means that we need to teach AI in the first place, by labelling training data from which the learning algorithm then learns to understand unseen situations in a “reasonable” way. This is used everywhere — from self-driving cars, Google’s search engine, auto-detection of diseases from medical images and samples, Snapchat’s and Instagram’s filters, Spotify’s recommendation system, designing building’s anti-earthquake features, figuring out how to build a plane and much more.
To do that, researchers do a lot of manual, time-consuming work or use services like CrowdFlower and Mechanical Turk. Researchers’ time is much better used elsewhere and services like Mechanical Turk are centralized, inflexible, expensive and non-real-time (which means it can take hours or days before new training set can be introduced back to the AI system for it to learn something new).
Eventum is a perfect solution to the above problem, where crowd categorizes, labels and analyses the image, then if the consensus is made (i.e. the result is verified), the result is sent as a real-time feedback to any AI system, which can now learn something new, WHILE doing a task (e.g. driving a car, correcting its recommendation system while people are searching, adjusting parameters when designing a system in real-time …). Eventum thus acts as a real-time feedback stream to assist AI algorithms in learning and correcting their predictions.
