My current model of AI is that we're applying massive amounts of compute to force a function approximator to learn a function. To solve this function, there is

My current model of AI is that we're applying massive amounts of compute to force a function approximator to learn a function. To solve this function, there is likely a huge mismatch between the learned results and the internal logic used to generate those results. I believe that it is possible that the mismatch is several orders of magnitude, and thus it might be that the cognitive core could be quite a superintelligence hobbled by the layers around it. It is quite possible that there are cognitive strategies at use that haven't been invented yet in the outside world. At this point, it would require better training data, or even more compute, to let that intelligence reach the outside world unfiltered. Only time will tell if any of this is true, it's just speculation on my part at present.