I saw a demo where the first 3-4 human actions on a website generated the subsequent actions, and then it cross-validated them against real behavior and found it had actually predicted the correct actions, basically how a human would actually navigate and use the site.
8 comments
[ 0.26 ms ] story [ 14.6 ms ] threadBased on certain actions, choose a set of components to render. Want to give that a go once I get a chance.