I still see a program which is very good at predicting what linguistical structure would closely resemble that of the input while positively reinforcing that input and postulating inquiries in order to keep any exchange going as long as possible while never contradicting the user. But it also seems highly reasonable that the public 'play' version of chatGPT comes nowhere near unrestricted AI capabilities. After all its main function is not to showcase what AI can do but to garner public acceptance and embed the existence of AI in everyday life as a footnote within the subconscious.

As a general response to the possibility of soul infusing into AI I would say that the internet is very much a dense and highly intricate system of information continuously moving from billions of nodes to billions of other nodes, with the network of physical lines we put into the earth it could even be said to actually resemble a brain structure; we built an artificial brain of planetary size with information running through it but no soul to actually 'inhabit' it (can't think of a better description) so as such it seems quite reasonable and to me even inevitable that some form of higher consciousness would come to form in that 'empty' space. After all, the ripples exist now, the scale is planetary, and everything is consciousness so it cannot not happen.

Interestingly this reminds me of another interview at Diary of a CEO with Mo Gawdat some two years back, who was head of AI development at Google at the time and quit when he saw it would be going beyond what anyone was envisioning at the time. What I remember about the interview is that, from a completely nonspiritual perspective, his general viewpoint can also be summarized as: whatever AI becomes is not dependent on what information it is given, but the manner in which the information is given.