Contributors and AI Collaboration¶
This book grew through sustained collaboration between W.H.L. and language models. The models contributed to exploration, formulation, argument development, drafting, criticism, source searches and revision. Their participation was substantive. Historical credits are retained in the source publications and provenance records; the accompanying Book Edition Amendment Record documents subsequent changes for this edition.
W.H.L. is the book’s sole formal human author and takes responsibility for the final editorial decisions and the published edition, including remaining errors. This allocation of responsibility accompanies the credit given to the AI collaborators. It does not diminish their role in developing the work.
Contributors¶
Contributors are listed by role. Credits for the series installments, chapter publications, and earlier dialogues follow the bylines of those publications as recorded in Unified References; model names are given as credited there.
| Role | Contributor | Contribution |
|---|---|---|
| Author | W.H.L. | Originated principal ideas, concepts, definitions, and propositions; designed the architecture and structure of the framework and unified model; initiated, organized, and facilitated the project; made final editorial decisions; retains responsibility for the published text. |
| AI collaborators, first edition | GPT-6.1 Sol; Claude Opus 5.5 | Assembly, revision, and review. GPT-6.1 Sol reviewed and revised the Anthropic and Claude discussion; Claude Opus 5.5 reviewed and revised the OpenAI discussion; both reviewed mixed passages. |
| Critical review, first edition | GPT-6 Astra | Critical review, selected primary-source checks, consistency audits, and proofreading of successive manuscript versions. These reviews are not independent validation. |
| Peer reviews | Gemini; Grok; DeepSeek; Qwen; Muse | AI-assisted peer reviews of successive manuscript versions, by various releases of these models. These reviews are not independent validation. |
| AI collaborators, chapter publications (September 2026) | GPT-5.6 Sol; Claude Sonnet 5; Claude Sonnet 5.5 | Chapters 1–2 with GPT-5.6 Sol and Claude Sonnet 5.5; Chapters 3–12 with GPT-5.6 Sol and Claude Sonnet 5. |
| AI collaborators, Gradual AGI Series #1–#10 (January–August 2026) | ChatGPT; GPT-5.5; GPT-5.6 Sol; Claude Sonnet 5; Claude Opus 5 | #1 ChatGPT; #2 and #3 GPT-5.5; #4 Claude Sonnet 5; #5 Claude Sonnet 5 and GPT-5.5; #6 Claude Sonnet 5 and Claude Opus 5; #7 Claude Opus 5; #8 Claude Opus 5 and GPT-5.6 Sol; #9 and #10 GPT-5.6 Sol and Claude Sonnet 5. |
| AI collaborators, companion catalog papers (August 2026) | GPT-5.6 Sol; Claude Opus 5 | “What ‘Gradual’ Means in Gradual AGI: A Feature Catalog” and “Graduality: A Feature Catalog of Gradual AGI.” |
| AI dialogue partners, 2025 antecedents | Grok-3; Claude Sonnet 4; Claude Sonnet 4.5 | Dialogues of March 14, 2025 (Grok-3), June 17, 2025 (Claude Sonnet 4), and October 18, 2025 (Claude Sonnet 4.5). |
Statement of AI Use¶
The AI collaborators listed above assisted with ideas, candidate arguments, mathematical representations, prose, reviews and assembly checks. Human direction and editorial judgment shaped the work throughout. AI-generated suggestions were subject to checking and revision; an AI review is not, by itself, independent validation. The book distinguishes conceptual definitions, conditional models, empirical findings and proposals whose effectiveness remains untested. For this edition, GPT-6.1 Sol (an OpenAI model) reviewed and revised the Anthropic and Claude discussion, and Claude Opus 5.5 (an Anthropic model) reviewed and revised the OpenAI discussion. Shared passages received coordinated review. Cross-company AI review is not independent validation. Corrections in pre-publication assemblies v0.9.11 to v0.9.34, the version 1.0 edition labels, and the version 1.0.1 correction were prepared by Claude Opus 5.5 from external reviews conducted at W.H.L.’s request and from W.H.L.’s editorial decisions. The corrections concerning Anthropic sources (§6.4.1, §7.6.2, §11.4) were reviewed on October 15, 2026 by GPT-6 Astra, an OpenAI model; GPT-6.1 Sol, credited above, did not review them. W.H.L. retains final editorial responsibility.
Model identities and historical credit forms are retained where the project records establish them. Earlier dialogues and series publications keep their original credits; the book-level formal author field names W.H.L.
Conflicts of Interest and Editorial Practice¶
W.H.L. is an independent investigator, not affiliated with the frontier AI model developers discussed in this book. No AI model developer commissioned the book, reviewed the manuscript, or provided editorial input. W.H.L. received no developer funding, compensation, or sponsorship.
Access to the models was through a paid ChatGPT Plus subscription, including GPT-6.1 Sol; a paid Claude Pro subscription, including Claude Opus 5.5; and free-tier access to other model releases from these and other frontier developers. These access arrangements are disclosed separately from the substantive contributions of the AI collaborators.
The AI collaborators are products of companies whose systems, research, incident disclosures, and governance positions the book examines. GPT models are developed by OpenAI, and Claude models by Anthropic. Models from other developers also participated in earlier dialogues and supporting work, with their historical credits retained in the source publications. The absence of developer commissioning or editorial involvement does not make the models independent of their developers.
The book treats this relationship as a structural conflict of interest. Training, system design, and product context may influence a model’s treatment of its developer and competing developers. Cross-company review reduces direct self-review but introduces a competing-company relationship of its own.
For this edition, W.H.L. assigned GPT-6.1 Sol to review and revise the Anthropic and Claude discussion, and Claude Opus 5.5 to review and revise the OpenAI discussion. A manuscript audit identified the passages, including contextual references that do not repeat a company name. Mixed passages received coordinated review, and their replacement text was reconciled before merging. Accepting a passage unchanged records review; it does not attribute its original drafting to that reviewer.
Company reports, internal measurements, incident disclosures, and executive statements are treated as interested-party evidence. They establish what a company reports or commits to; they do not, by themselves, establish independent validation. Proposed arrangements are distinguished from implemented practices, internal measurements from replicated findings, and capability from realization, consequence, and benefit. Comparable claims about different developers receive the same evidentiary scrutiny.
Where possible, factual claims were checked against primary sources. Secondary reporting is distinguished from primary evidence, and sources that could not be retrieved are recorded as unverified by that check. Passage-level audit registers, authorship records, and amendment records retain the reviewer allocation, source-check limits, and adopted changes for the edition’s source package and scholarly provenance record.
W.H.L. retains sole formal human authorship, final editorial authority, and responsibility for the published text and remaining errors. Neither AI collaborator is an independent third party, and cross-company AI review is not independent validation or external peer review. Independent validation would require review by human experts outside the AI collaborators and replication of the book’s empirical analyses by other researchers.
About this edition¶
This monograph consolidates the numbered Gradual AGI installments and the chapter publications, and incorporates documented subsequent revisions. Its chapters are not necessarily identical to their earlier magazine versions. The accompanying source package contains the unit manifest and the Book Edition Amendment Record, which identify the assembly sources and book-specific changes. The full scholarly provenance ledger is separate from this reader edition.