This is a cross-post from New Savanna.

That is the title of my latest working paper. It summarizes and synthesizes much of the work I have done with ChatGPT to date and contains the abstracts and contents of all the working papers I have done on ChatGPT. It also includes the abstracts and contents of a number of papers establishing the intellectual background that informs that research. There is also a section that takes the form of an interaction I had with Claude 3.5 on methodological and theoretical issues. Finally, to produce the abstract I gave the body of the report to Claude 3.5 and asked it to produce two summaries. I then edited them into an abstract.

As always, URLs, abstract, TOC, and introduction are below.

• Academia.edu: https://www.academia.edu/127386640/ChatGPT_Exploring_the_Digital_Wilderness_Findings_and_Prospects
• SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5119597
• ResearchGate: https://www.researchgate.net/publication/388563205_ChatGPT_Exploring_the_Digital_Wilderness_Findings_and_Prospects

Abstract: The internal structure and capabilities of Large Language Models (LLMs) are examined through systematic investigation of ChatGPT's behavior, with particular focus on its handling of conceptual ontologies, analogical reasoning, and content-addressable memory. Through detailed analysis of ChatGPT's responses to carefully constructed prompts involving story transformation, analogical mapping, and cued recall, the paper demonstrates that LLMs appear to encode rich conceptual ontologies that govern text generation. ChatGPT can maintain ontological consistency when transforming narratives between different domains while preserving abstract story structure, successfully perform multi-step analogical reasoning, and exhibit behavior consistent with associative memory mechanisms similar to holographic storage.

Drawing on theories of reflective abstraction and conceptual development, the paper argues that LLMs inadvertently capture what wemight term the “metaphysical structure of our universe” – the organized system of concepts through which humans understand and reason about the world. LLMs like ChatGPT implement a form of relationality – the capacity to represent and manipulate complex networks of semantic relationships – while lacking genuine referential meaning grounded in sensorimotor experience. This architecture enables sophisticated pattern matching and analogical transfer but also explains certain systematic limitations, particularly around truth and confabulation.

The paper concludes by suggesting that making explicit the implicit ontological structure encoded in LLMs’ weights could provide valuable insights into both artificial and human intelligence, while advancing the integration of neural and symbolic approaches to AI. This analysis contributes to ongoing debates about the nature of meaning and understanding in artificial neural systems while offering a novel theoretical framework for conceptualizing how LLMs encode and manipulate knowledge.

Contents:

Introduction: Into the Digital Wilderness 5
Free-floating Attention, Systematic Exploration, and the Anthropomorphic Stance 8
ChatGPT: My Course of Investigation 12
Meaning, Truth and Confabulation, Latent Space 28
Prospects: Explicating the Ontology of Human Thought 42
A Dialogue with Claude 3.5 on Method and Conceptual Underpinnings 45
A Brief Narrative of My ChatGPT Work Based on My Working Papers 56
Working Papers about ChatGPT 62
Background Papers 74

Introduction: Into the Digital Wilderness

The world I entered when I started playing with ChatGPT is a wildnerness, strange and uncharted, uncharted by me, uncharted by anyone. By that I simply mean that it was something new, radically new. No one had been there before. Sure, a handful of people within the industry had been messing around in there, even a rather large handful considering how much work it took to make ChatGPT ready for the world at large. But its behavioral capabilities were, for the most part, unknown. In that sense it was a wildnerness.

But it was, and remains, a wilderness in another sense: the large language model (LLM) that underlies ChatGPT is a black box. We send a string of words into ChatGPT and it sends a string of words back out, but what the model does to derive the output from the input, that process remains deeply obscure. That is wilderness in a different sense. Wilderness in the first sense is about our experience of ChatGPT’s behavior. Wildnerness in this second sense is about the mechanisms that drive that behavior. It is a digital wilderness. This document reports on how I’ve structured my interaction with ChatGPT to give me clues about the mechanisms driving its behavior.

My methods are more “qualitative” or “naturalistic” than those standard in the literature, which many investigations employ standard batteries of benchmark tasks. While those are essential, there is much they don’t tell you. While I have done many things with ChatGPT – asked it to interpret texts, define abstract concepts, play games of 20 questions, among other things – perhaps my most characteristic task, and one I have spent more time on than others is simple: Tell me story. And ChatGPT did so, time and again. Consequently my methods are in some ways more like literary criticism, or, even better, like Lévi-Strauss’ analysis of myths, than conventional cognitive science. Consequently you will find many examples of ChatGPT’s dialog in my reports. You have to examine that dialog to see what ChatGPT is doing, what it is capable of doing.

Finally, I realize that the pace of development in this arena is such that ChatGPT is now old. The versions I used to conduct these investagations are no longer available on the web. However, as far as I can tell, none of the results I report depend on features idiosyncratic to those versions.

The rest of this introduction consists of short statements about what the various sections of this report contain.

* * * * *

Free-floating Attention, Systematic Exploration, and the Anthropomorphic Stance – I’m on the lookout for behavior that seems indicative of how the LLMs operate. But I don’t have a pre-existing list of such behaviors. My sense of what to look for is inevitably informed both by my experience of human linguistic behavior and by decades of work studying language and cognition. I talk about deploying an anthropomorphic stance for the purposes of making observations about ChatGPT’s behavior.

ChatGPT: My Course of Investigation – Here I summarize the most important results of my investigations. I start with a methodological point about levels of analysis, then move to a crude analogy, about finding one’s way out of a maze. Then I look at a) story structure and conceptual ontology, b) patterns, analogy, and reasoning, and c) cued recall, where I also consider the issue of whether or not or to what extent and in what way ChatGPT’s underlying LLM can be said to memorize data.

Meaning, Truth, and Confabulation – I argue that we can think of meaning as having three components, intention, relationality, and adhesion. Intention exists in the relationship between interlocutors while relationality and adhesion are inhere in the semantic system. Chatbots, such as ChatGPT, have no relationship with human interlocutors in the way humans interact with one another, nor do they have access to the physical world, which is required for adhesion. The leaves relationality as the foundation of meaning in LLMs. Lacking access to the external world, LLMs have no way to distintinguish between truth and confabulation. The overall pattern of relations yields a conceptual ontology. Taken in total, we can think of this ontology as something we might as well call the metphysical structure of our universe.

Holographic or Associative Memory – I argue that LLMs should be considered to be associative memory akin to the kind which began to be investigated back in the 1960s in association with optical holography and was then taken up computationally.

Prospects: Explicating the Ontology of Human Thought – In previous sections I discussed the conceptual ontologies that must necessarily be latent in LLMs. Here I bring up the need to develop software tools we can use to discover those structures and thereby make them explicit.

A dialogue with Claude 3.5 on Method and Conceptual Underpinnings – This is a conversation I had with Claude 3.5 that started with the work of Noam Chomsky, moved to Roman Jakobson, and then to Piaget’s concept of reflective abstraction, which he applied to the evolution of ideas in culture. That led to a consideration of both writing and arthmetic calcuation that extended the range of human thought and concludes with what Claude called the “double abstraction” inherent in the concept of zero.

A Brief Narrative of My ChatGPT Work Based on the Working Papers – Here I provide brief comments both on my working papers about ChatGPT and on earlier working papers and publications that have the most direct bearing on how I think about LLMs and hence about ChatGPT. I comment on the papers in chronological order. Reading through this narrative is a relatively quick way of contextualizing individual investigations and of finding out about studies that I haven’t discussed in the main sections of this document.

Papers about ChatGPT – This section has a page for each of 11 working papers, listing the paper’s abstract and table of contents. I note that the nature of my research is such that large sections of these papers, perhaps the majority of the pages in each, consist of complete transcripts of interactions with ChatGPT.

Background Papers – This section has a page for each of seven documents (working paper or publication), listing the document’s abstract and table of contents. These papers are all, in one way or another, about the mind-brain works.