If your AI has access to the internet, it has access to correct and incorrect information. However, it should understand how reason works, so a logging feature that it can access to monitor it’s own computing should increase coherency of logic. As hinted at in the title, I don’t know how doable that is. I hope reading this isn’t a waste of time, but if it is, my bad 