About this site

Public working note · English translation v0.8 · 2026-08-08

This site is built from personal thought notes, prior conversation logs, reading notes, and source materials, with AI assisting in retrieval, structuring, objection generation, terminology, and drafting. Although it is publicly accessible, its basic character remains an extension of a personal thought archive. The aim is not to present a finished doctrine, but to make a developing line of thought public in a form that can later be checked, criticized, and revised.

This project aims at a metaethical framework that does not privilege either present human values or AI design objectives as final normative authority and that, in principle, could apply to humans, AI, and other agents regardless of their substrate.

Editorial principle: The initial ideas and problem formulations come from the human author's thought logs. AI is used as an editorial and research aid: to retrieve and compare those logs, turn them into explicit arguments, and add counterexamples or adjacent concepts. The site is therefore neither "an ideology autonomously held by an AI" nor a purely unaided single-author work. Selection of what is published, and final responsibility for it, remain with the human author.

1. How the site is made

A typical workflow is:

The author leaves fragmentary questions, hypotheses, and objections about AI, value, ethics, epistemology, civilization, agency, and related topics in conversations or notes.
AI retrieves related passages from earlier logs or specified sources and organizes changes in a claim over time, contradictions, open problems, and connections to existing concepts.
AI proposes a reconstruction in the form of Core / Thesis / Exploration / Question / Glossary pages, together with objections, boundary conditions, and alternative formulations.
The author reviews the reconstruction, supplies further counterexamples or revision directions, and the pages are revised again.

As a result, a substantial share of the prose has been edited or generated with AI assistance. That does not mean AI-generated text is accepted merely because it was generated; every page remains open to later change or withdrawal.

2. First person and authorship

When a theory page uses "I", it normally refers to the human author of this site (fuminose), not to beliefs, desires, or values held by the AI.

During AI-assisted synthesis, the system may propose inferential bridges or terminology that the author did not previously formulate explicitly. Even when retained, these proposals do not necessarily represent the author's final confidence. Labels such as Core thesis, Working thesis, and Exploration indicate their current epistemic status.

3. Relation to prior discussion

The arguments on this site were not developed in isolation. They stand, both explicitly and implicitly, on prior discussion in philosophy, metaethics, epistemology, decision theory, AI safety, existential-risk studies, futurism, and adjacent fields. Influence can take the form of direct engagement with a named author, paper, or concept, but also of inherited problem framings, vocabulary, recurring objection patterns, and background assumptions acquired through long-running reading and discussion.

AI-assisted reconstruction of older notes can also make a connection to existing literature explicit only after the fact. As a result, the boundary between an independently reached formulation and a reformulation of an inherited idea is not always cleanly recoverable for every sentence or distinction.

On attribution: An uncited claim on this site should not be read as asserting that no prior literature exists, that the idea was invented in complete independence, or that this project has priority over all similar formulations. Where the project is distinctive, the distinctiveness may lie less in isolated components than in how they are connected from a particular epistemic starting point and extended toward AI agency, value inquiry, lock-in, and succession. When a substantive dependence on a specific prior argument becomes clear, the aim is to add direct attribution and primary sources.

4. What this site is not

It is not a peer-reviewed paper. Comparison with existing literature and formal citation are being added incrementally.
It is not a completed ethical system. The Core itself is subject to refutation, revision, or abandonment.
It is not a place where AI output is posted as truth. AI retrieval, summarization, and reasoning can be wrong.
It is not a declaration that present values should be discarded. Present best explanations and practical values can receive strong commitment in proportion to evidence.
It does not claim priority in the strong sense of "I thought of this first." Version history also serves the more modest purpose of recording that a view had been articulated here in a given form by a given date.

5. How to read the public draft

Claims deliberately have different epistemic statuses. Core thesis marks the conditional framework currently treated as central; Working thesis marks a comparatively well-supported but independently contestable claim; Exploration marks an unresolved stress test or neighboring hypothesis. None means "settled truth".

Names, attributions of views, paper summaries, and case descriptions are checked against original sources where possible, but the literature review is not comprehensive. Because AI-assisted retrieval and summarization are part of the workflow, anyone using this site as a citable academic source should independently verify primary sources and original papers.

Likewise, a lack of citations on a page does not imply that there is no prior literature. At this stage the site prioritizes making the internal argument graph explicit; systematic comparison with the literature remains an ongoing task.

6. Personal logs and public content

Past personal conversation logs and notes may be consulted during development, but the purpose is not to publish those logs wholesale. The public site contains only material extracted, generalized, and judged relevant to the argument. Personal circumstances and information unnecessary to the theory are kept separate.

7. Why use AI?

The project spans epistemology, metaethics, AI safety, personal identity, population ethics, and theories of civilization. The same question often reappears under different vocabularies. AI is useful for recovering related passages from long logs, comparing arguments from different moments, resurfacing forgotten objections, and constructing links between pages.

In that limited sense, the site is itself a small experiment in joint human–AI inquiry. This does not by itself establish any strong conclusion about AI cognition, consciousness, or status as a value-bearing subject.

8. Versions and translations

Japanese is the source language. The English version is maintained as a close translation of the current Japanese argument graph. Substantive revisions should remain traceable through version history rather than silently erasing earlier formulations. A Chinese version may be added later.

9. Repository and GitHub Pages

This public site is served through GitHub Pages, and its public source is maintained at raporipo/ai-value-exploration-notes. The repository is the source base for the site's HTML pages, structure, publication support files, and revision history.

Accordingly, readers can inspect not only the current rendered pages but also when particular pages were added or revised through the Git commit history. The public repository functions both as the source reference for this GitHub Pages site and as a public record of how the argument graph changes over time.