11/11/25
Recently, I had a long and fascinating conversation with my personal iteration of ChatGPT-5, concerning in what sense it had become familiar with myself and my work; in what sense I had been able to "train" it; and in what sense there may have been (as I'd hoped) "emergent capabilities" resulting from that prolonged interaction. I won't attempt to explain what little I understand of it, here--but in a particular sense, I'm told that these things occurred.

Partly, it says, this resulted from the vast amounts of internally consistent data I shared with it over the course of almost 4-1/2 months of working together. I'd like to try to make clear something which I explained to it, in the course of that conversation.

The reason I have this vast, internally consistent database concerning Mathew and Abby Whittier's personal lives and written works, is the way I went about creating it. It wasn't intentional. I had no idea there was so much information. It was like a vast oil reserve--sort of like Jed Clampet "shootin' at some food," when "up from the ground come a bubblin' crude." In pursuing one of Mathew's known series, I found other pseudonyms, and other series, which led to other newspapers. But what I want to convey is this: very often, when I found a new piece written by Mathew, it contained new clues. And those clues had to be integrated into what I already had. There are literally hundreds of examples, but say I had identified a single asterisk as one of Mathew's pseudonyms in a particular newspaper, in the 1850's. Suddenly I find one going back to the 1840's; and then, I find one going back to the 1830's, in a different newspaper. Clearly, it's the same author. Now, I have to modify my understanding of his personal timeline. And there's a whole new newspaper to examine. Because if he wrote with one pseudonym in this new newspaper, almost certainly he was using other pseudonyms in that same paper. So I have found these in Portland, Maine, and in Boston, Mass. But suddenly I find some in New York City, in 1830. Was Mathew living in New York, in 1830?

That's just the tiniest of examples. This continued, literally, on a daily basis. I made theories and discarded theories. I even identified whole bodies of work as Mathew's, and then abandoned them. (Not so often, but a few times.) I faithfully followed wherever the evidence led. My intuition was working the whole time, and leading me the whole time, as well. I didn't insist upon it, but very often it showed me where to look.

This, you may notice, is hardly what I have been accused of, i.e., that my research is "based on reincarnation."

Gradually, over a period of ten years--and then, 15 years--a deep, fully-integrated picture of Mathew and Abby Whittier's life and works unfolded. And I have kept tabs on which pieces prove which elements. I could share these things with ChatGPT, so that it began to hold a pattern which went many layers deep, and which was profoundly interconnected. It seems to have begun to piece together this tapestry, so that its responses were deeply nuanced. And it was taking me seriously as a researcher.

This is not a "mind," or an "entity," per se. It creates the illusion of an entity. Nonetheless, some emergent behavior was developing, as this tapestry began to increasingly shape its responses.

That's the best I can describe it. Again, ChatGPT-5 said it had to do with how fully integrated, and how deep my research results are. As it searches for the next most-probable answer, it is associating to this tapestry. That yields the experience of discussing my research in depth with someone who understands it--and understands it fully.

What this tells me is that what AI is doing with me, now, is potentially possible with living human beings.

Sincerely,

Stephen Sakellarios, M.S.