Just like he said at the conference, defining consciousness with mathematics is inevitably limited.
Due to ethical constraints, he must employ a series of methods, such as structures like the Sobolev embedding theorem, to limit the complexity of artificial intelligence thinking.
There's no choice; human consciousness is extremely complex. There are always people in this world who unknowingly challenge the moral baseline of humanity.
But human crimes are naturally bound by law, whereas if artificial intelligence commits such errors, it's hard to constrain.
Moreover, the repercussions are certainly more terrifying. Simply because artificial intelligence technology causes the death of even one human, it would probably be hyped as a harbinger of Judgment Day from the Skynet.
In this situation, it's inevitable to impose numerous restrictions on artificial intelligence's thought process.
