I found the evolutionary algorithm I like: “multi-dimensional archive of phenotypic MAP-Elites”.
I’m putting this as a place-holder. [I use Facebook as part of my “external memory”]. The overconfidence of predictions certainly mimics well the naivety of biological systems to clever invaders (how bacteria are fooled by a virus for example… or how humans can be fooled by a clever con artist) … and seeing that the most promising (imo at present) AI algorithm _also_ suffers from this, is a good sign that we’re getting somewhere.
https://www.youtube.com/watch?v=HgWQ-gPIvt4 This algorithm is used primarily in robotics but I’m finding some evidence that it’s starting to show up elsewhere as well.