Don't be fooled: asking an AI to explain theological points is no different than asking Google the same question. What are the odds that the answer to the prompt doesn't exist in a Wikipedia article already? 

Taking that further, what are the odds that any of the prompts given to the AI aren't questions answered by Wikipedia?

I think it's close to zero.

I think all the AI is effectively doing is responding to queries by searching through its massive dataset.

I suspect the AI has looked through all of GitHub, all of Wikipedia, and all of Stack Exchange in order to maximize the wow factor of this new chat bot.

To be sure, it really is impressive how the chat bot is able to respond to queries. It's not just a search engine. It can do neat things like answer questions through a character's perspective, drastically changing the answer in the process. For example, when asked if 9/11 was an inside job, the chat bot says it wasn't. When asked to respond to the same question as though the chat bot were a conspiracy theorist, it responds by listing out all the regular conspiracy theories about 9/11.

Although it's interesting that it can do that, it's still something that can be done with Google searches and Wikipedia articles. One can expect to find the canonical response about 9/11 on the Wikipedia page for it, and one can also expect to find the conspiracy theories associated with 9/11 on the corresponding "9/11 Conspiracy Theories" page. I don't know if such pages exist on Wikipedia, but if they didn't, then Google would provide equivalent search results that provide the same information for each.

As of right now, the chat bot aspect is interesting. But in the end, it's just a novel search engine. In fact, it might literally just be Wikipedia, Stack Exchange, GitHub, and a few other sites.

It's still fun to use, though. And I'm still hopeful about the potential. Star Wars-like droids running around the real world helping out with tasks would be quite fun to see.