If good people make AI, can AI make good people?
The developers of ChatGPT built extremely useful software, putting a “Wikipedia on steroids” at our fingertips by leveraging collective intelligence in ways never seen before. In this sense, as defined by Plato and Aristotle over two thousand years ago, they are “good people”. This means they are morally good or virtuous by giving away a highly beneficial software product for free that makes many chores of the daily life of knowledge workers much easier. These AI developers are thus acting according to the golden rule (of reciprocity), treating others as they would like to be treated. The problem is that not everybody who writes AI software is a “good” person. If only one percent of AI developers acts entirely egoistic, or even malicious, there is the risk that these “bad apples” will abuse the power of AI for their own sinister purposes, without any consideration for the wellbeing of the rest of us. But what if we could turn the power of AI on these “bad apples”, to identify them be...