July 25, 2026. Paper Permission: What Publisher AI Policies Allow and Practice Withholds AFS3.6: Paper Permission: What Publisher AI Policies Allow and Practice Withholds Names Paper Permission: a permission present in the text of publisher policy and largely absent from open practice. Read as a map of delegable work, the reserved core has narrowed to the integrity of primary research images and the warranty of the claims, so almost every other task becomes conditionally delegable; yet the disclosed and permitted route is the one least taken. Proposes that disclosure can function as a confession signal, so that the authors who comply and are least protected carry the cost while the judging side uses AI under fragmented rules, and offers three remedies that add no conduct duty to authors or reviewers. July 23, 2026. Companion to AFS03: how the reserved core is guarded, with its supporting dataset AFS3.5: The Warranty Axis: How Publishers Govern the Human-Reserved Core of AI-Assisted Authorship Asks not what publishers reserve for humans but how they guard it, and finds the reserved core at the level of text governed almost entirely as a regime of disclosure and human accountability rather than as a list of forbidden tasks. Names that dimension the warranty axis: what a publisher reserves is the obligation that a competent human warrant the resulting claims, rather than a set of tasks that machines may not touch. AFS3.5-dataset: A Minimum/Maximum Reading of Six Publishers' Generative-AI Disclosure Requirements: A Companion Dataset Records, for six major publishers and six manuscript-preparation tasks, whether an author must disclose the use of a generative-AI tool, taking each answer from the publisher's own public guidance. Because publishers state some tasks and leave others unstated, it gives both a minimum and a maximum reading so that the uncertainty stays visible rather than hidden in a single table. July 19, 2026. "Accelerated Future Science" Program Plus: An extra three articles AFS07: Post Scientific Singularity: Verification Access, AI Research Guardianship, and the Political Economy of Machine-Generated Knowledge Argues that once AI can generate and machine-verify more candidate discoveries than human institutions can reconstruct, the decisive scarcity shifts from discovery to verification access and to the institutional capacity that converts validated outputs into priority, rights, and implementation. AFS08: The Implosion of Science: When Cognitive Topologies Converge toward a Least Common Denominator Identifies the Implosion of Science: science may become faster and more productive while human researchers, AI models, institutions, and observation systems converge on a shared cognitive topology, losing the functional differences needed to generate fundamentally different questions. AFS09: The Cost of a Second Intelligence: Can ASI Create an Epistemic Outside to Itself? Asks whether an artificial superintelligence can create an epistemic outside to itself, develops a four-role architecture from ASI-ONE to ASI-VAGABOND, and argues that after cognitive convergence, restoring discovery may cost the loss of complete control. July 19, 2026. Interlude to AFS02: what is lost before evaluation can begin AFS2.5: Accessibility-Loss Mortality: The Triple Barrier Before the Discoveries of Non-Native, Disabled, and Neurodivergent Scholars Identifies a triple barrier standing before an insight can be evaluated at all: thought must be converted into conventional written language, that writing must often be converted into academic English, and when AI is used to cross either barrier the resulting text may face suspicion because it sounds machine-generated, so the remedy for one barrier creates the next. Names the life-related consequence Accessibility-Loss Mortality, claims no magnitude for it, and argues that institutions should evaluate warranty and answerability directly instead of treating linguistic style as evidence of cognitive origin. July 15, 2026. "Accelerated Future Science" Program: A six-part article series discussing what happens—in the future (or even the present)—when science is accelerated by AI. AFS01: The Conceptual Armament Effect: What Happens When the Cognitive Topology Is Rewritten Defines the Conceptual Armament Effect: a generated idea's novelty depends strongly on the concepts a generator is equipped with at generation time, and that equipping deforms the space of questions the generator can pose. AFS02: Recognition-Lag Mortality: The Expected Human Cost of Not Recognizing Genuine AI Discoveries Defines Recognition-Lag Mortality as a measurement object — the expected human cost of the interval between a genuine AI-produced discovery and its institutional recognition — without estimating its magnitude. AFS03: What Publishers Reserve for Humans: Reading Generative AI Policies as a Map of Delegable Work Reads six major publishers' generative-AI policies through a three-layer model (disclosure-exempt, disclosed, human-reserved core) and shows authorship relocating from manual text production to epistemic warranty. AFS04: The Paradox of Honest Deceleration versus Undisclosed Acceleration: From He and Bu (2026) to Policy Design as Behavioral Intervention Models a self-reinforcing paradox in which disclosing authors decelerate while non-disclosing authors accelerate, via constructs of Net Acceleration, the Permitted-Perceived Gap, and Disclosure Asymmetry, and proposes an experiment to test it. AFS05: Transitionhood's End: From Origin-Based to Contribution-Based Evaluation of Scientific Knowledge Argues that four premises underlying current AI-disclosure regimes (discreteness of use, cognitive independence, traceable provenance, human verifiability) are collapsing, and that scientific evaluation should shift from origin to contribution. AFS06: The Scientific Singularity: When Human Understanding Ceases to Be a Condition of Scientific Acceptance Separates AI's capacity to produce discoveries from the threshold at which human comprehension ceases to be a condition of scientific acceptance, names that threshold the Scientific Singularity, and identifies the resulting intermediary role, the hierophant. July 13, 2026. Output Management Plan: Organized the various outputs to ensure an accurate chronological record. July 5, 2026. The Connecting Criterion: Discussed the criteria for determining to whom a discovery should be attributed and on what grounds. July 4, 2026. The Purity Premium / The Contamination Dividend: Examined whether an experimental artifact—equipped with a new set of human-derived concepts—could yield discoveries beneficial to humanity. June 1, 2026. The AI Layoff Trap's Trap: Describes the phenomenon where only summaries circulate on social media, causing arguments to become polarized.
Paper
The full text of this publication is not hosted on 44B due to licensing.
Read it at OpenAlex