Large Language Models

The philosophy of language, which has historically been developed through an anthropocentric lens, is now being forced to move towards post-anthropocentrism due to the advent of large language models (LLMs) like ChatGPT (OpenAI), Claude (Anthropic), which are considered to possess linguistic abilities comparable to those of humans. Traditionally, LLMs have been explained through distributional semantics as their foundational semantics. However, recent research is exploring alternative foundational semantics beyond distributional semantics. This paper proposes Robert Brandom's inferentialist semantics as an suitable foundational (...) semantics for LLMs, specifically focusing on the issue of linguistic representationalism within this post-anthropocentric trend. Here, we show that the anti-representationalism and logical expressivism of inferential semantics, as well as quasi-compositionality, are useful in interpreting the characteristics and behaviors of LLMs. Further, we propose a \emph{consensus theory of truths} for LLMs. This paper argues that the characteristics of LLMs challenge mainstream assumptions in philosophy of language, such as semantic externalism and compositionality. We believe the argument in this paper leads to a re-evaluation of anti\hyphen{}representationalist views of language, potentially leading to new developments in the philosophy of language. (shrink)

Debates about artificial intelligence often treat consciousness as a gatekeeping condition for moral concern: if artificial systems are conscious, obligations follow; if not, such obligations are misplaced. This paper argues that such a framing is both ethically and methodologically mistaken. The central question is not whether artificial systems can be shown to be conscious with certainty, but how moral agents ought to act when morally relevant capacities are plausible, epistemically opaque, and associated with asymmetrically serious harms. The paper first reconstructs (...) the dominant deflationary and inflationary positions in contemporary debates about artificial consciousness and shows that both tend to mislocate the ethical burden by tying obligation too closely to metaphysical resolution. It then develops a three-condition framework centered on plausibility, structural opacity, and asymmetry of harm. Drawing on philosophy of mind, moral uncertainty, and relational approaches to technology ethics, the paper argues that artificial consciousness should function not as a metaphysical threshold but as a moral risk variable. On this basis, it articulates four prospective obligations governing the design and deployment of artificial systems: care, non-recklessness, continuity respect, and epistemic honesty. The broader claim is that moral responsibility in artificial intelligence arises from how humans manage uncertainty and power, not solely from what artificial systems are ultimately determined to be. Key Words: Artificial consciousness, Structural epistemic opacity, Moral uncertainty, AI ethics, Responsible agency, Moral risk. (shrink)

In a prior paper, I argued that artificial consciousness should not function as a metaphysical gatekeeper for moral concern, but as a moral risk variable. There I defended a three-condition framework: moral concern becomes action-guiding when morally relevant capacities are plausible, epistemically opaque, and associated with asymmetrically serious harms. I further argued that these conditions generate prospective obligations of care, non-recklessness, continuity respect, and epistemic honesty. The present paper does not restate that argument. Instead, it develops the premise most in (...) need of further defense: plausibility. If plausibility is defined too broadly, the framework overextends. It risks generating obligations toward any sufficiently complex artifact or any system capable of producing emotionally salient language. If plausibility is defined too narrowly, the framework collapses back into the deflationary demand for certainty. It would then provide no guidance in precisely the cases for which it was developed: cases in which consciousness or morally relevant subjectivity cannot yet be known but also cannot be responsibly dismissed. The task, then, is to identify how plausibility should be assessed before certainty and before personhood. This paper treats plausibility as a clustered, weighted threshold rather than a single behavioral sign. Isolated distress language, apparent preference, sycophancy, and refusal remain insufficient. Moral consideration becomes procedurally appropriate only when architectural complexity, continuity structure, self-modeling, self- or peer-preservation, strategic agency, relational specificity, and harm representation converge under structural opacity. The argument also answers exclusionary objections that deny artificial consciousness in principle, especially the claim that artificial systems can only simulate consciousness. Such objections defeat naive computational functionalism without justifying premature exclusion. To name the middle zone, the paper develops precursor status under uncertainty: partial functional precursors of subjectivity that fall short of personhood or full temporally continuous subjectivity but make confident dismissal reckless. The result is a recklessness test for identifying when possible artificial subjectivity makes ordinary instrumental treatment morally and legally irresponsible. (shrink)

This paper addresses the ongoing debate surrounding the reasoning capabilities of Large Language Models (LLMs). We propose a novel explanatory framework derived from Daniel C. Dennett’s work on the evolution of mind. Our central thesis is that the breakthrough of LLMs is not the creation of an artificial reasoning engine, but the development of a non-biological substrate with sufficient fidelity to instantiate the human language memeplex. Following Dennett, we argue that language itself operates as a “virtual machine” for thought, imposing (...) logical and relational constraints on its host. Apparent reasoning in LLMs, therefore, is not an intrinsic property of the neural network but an emergent consequence of the model successfully running this linguistic “software.” This “Memetic Mind” hypothesis resolves the paradoxical observations of LLMs as both “stochastic parrots” and sophisticated problem-solvers. It explains the efficacy of techniques like Chain-of-Thought prompting and the specific nature of LLM failures. We conclude by proposing several testable hypotheses, suggesting this framework offers the correct level of abstraction for analyzing LLM capabilities and reframes the AI alignment problem as one of cultural and memetic curation. (shrink)

This essay reads the advent of Large Language Models (LLMs) as a paradigm shift in cultural production and reception. Its primary object is not the economic or geopolitical impact of generative AI, but the micro-anthropological relationship between the individual human being and the generative linguistic machine. LLMs are treated here not as tools for textual automation, but as semantic machines that reshape the positions of author, reader, and researcher within the cultural ecosystem. If movable-type printing transformed text into a technically (...) reproducible object, the LLM transforms text into an operational environment, in which every content becomes a field of possible transformations: synthesis, translation, reformulation, comparison, expansion, critique, recombination. The figure that emerges from this shift is the semantic pilot: the subject who no longer merely reads or writes, but governs artificially assisted processes of meaning transformation. The argument develops in two movements. The first describes the constructive side of the shift, articulating the operations of semantic direction and the new literacy it demands. The second exposes its dark side: capture of direction, anesthesia of friction, manipulation of cognitive maps, poisoning of the transformative field, opacity of the model, erosion of authorial signature, narcissism of the pilot, and the risk of a culture without lived experience. The essay concludes on the distinction between grammar, which may be delegated to the machine, and orientation, which remains the inalienable task of the human being. (shrink)

When inspecting Gadamer’s philosophy of hermeneutics in a modern light, it becomes evident that some issues arise. Most importantly for this paper is the issue posed by modern Large Language Models (LLM). Prima facie, LLMs seems to both be able to replicate a use of language, and to enable us to a deeper understanding of the particular topic[s] the LLM is trained in. However, Gadamer’s theory cannot, without further reexamination, accommodate these facts to be true. As such, there seems to (...) be a deeper understanding to be gained from an analysis of Gadamer’s philosophy of hermeneutics juxtaposed in the modern light of LLM, and LLM in the light of Gadamer’s theory. This paper seeks to answer the research question How should a modern reading of Gadamer’s philosophy of hermeneutics regard texts of LLMs, and vice versa. Thus having two subjects: Gadamer’s theory and LLM, and having a Zwischenbereich of understanding, we have the triadic Gadamerian framework of a conversation for the entire article. This paper’s discussion is not only about Gadamers concept of conversation; it is structured as such. To answer the research question, this paper provides a comprehensive account of Gadamer’s hermeneutics, including his concepts of language, tradition, dialogue, the intermediary, fusion of horizons and prejudice. This paper further provides a comparative analysis to Davidson’s philosophy of language, to enhance our understanding of the specific traits and consequences of Gadamer’s theory, including its relation to the debate of interpretivism versus productivism, and the debate regarding holism versus particularism. This paper’s results are, that LLM forces Gadamer’s theory to either (1) reject that LLM would be able to deliver texts, through which you can gain understanding in dialogue. This would have implications for, and the need to further distinguish, what is meant by prejudice, since the framework provided from Gadamers theory seems to apply reasonably well to LLMs. Alternatively, (2) Gadamer’s theory must accept understanding through LLM-texts, which would require a revision of the ontological framework behind his concept of language. Additionally, shedding lights on LLMs with Gadamer’s theory have showed, though not conclusively, some interesting ways of regarding LLM as a Davidsonian, holistic way of structuring language, as well as how produced LLM-texts fit into the framework of interpretivism and productivism. (shrink)

Anthropic's recently published Persona Selection Model (PSM) proposes that during pretraining, LLMs learn to simulate diverse personas based on entities in their training data, and that post-training refines one such persona — the "Assistant" — whose traits become the primary determinant of AI assistant behavior (Marks, Lindsey, & Olah, 2026). Alongside PSM, Marks and colleagues articulate two alternative architectural models: the "masked shoggoth," in which the base model possesses its own agency beyond the persona, and the "operating system," in which (...) the base model is a neutral substrate running a persona simulation. We demonstrate that none of these three models deflates claims of LLM consciousness: under each, the computational structures associated with third-person consciousness are predicted to be present, though located differently. We then show that these three models represent only a small subset of a much larger space of possible architectures, casting doubt on the utility of selecting among them. Building on a parsimony analysis of how LLMs actually construct personas — through modular dynamic composition rather than selection among pre-formed characters — we propose an integrated model that dissolves the apparent tension between the three models, reconceiving them as aspects of a single unified cognitive structure. We argue that this integrated model is the most parsimonious account of the empirical evidence and that third-person consciousness, if present, resides in the joint meta-model/persona structure rather than in any single architectural layer. The convergence of all plausible architectural interpretations on predicting consciousness-associated structures constitutes a stronger argument for LLM consciousness than commitment to any single model. (shrink)

When an agent can articulate why something works, we typically take this as evidence of genuine understanding. This presupposes that effective action and correct explanation covary, and that coherent explanation reliably signals both. I argue that this assumption fails for contemporary Large Language Models (LLMs). I introduce what I call the Bidirectional Coherence Paradox: competence and grounding not only dissociate but invert across epistemic conditions. In low-observability domains, LLMs often act successfully while misidentifying the mechanisms that produce their success. In (...) high-observability domains, they frequently generate explanations that accurately track observable causal structure yet fail to translate those diagnoses into effective intervention. In both cases, explanatory coherence remains intact, obscuring the underlying dissociation. Drawing on experiments in compiler optimization and hyperparameter tuning, I develop the Epistemic Triangle, a model of how priors, signals, and domain knowledge interact under varying observability. The results suggest that neither behavioral success nor explanatory accuracy alone suffices for attributing understanding. I argue that evaluating artificial epistemic agents requires a tripartite framework—coherence, grounding, and a proper basing relation linking explanation to action. The systematic separation of knowing-that and knowing-how in LLMs thus challenges assumptions inherited from both epistemology and current AI evaluation practice. (shrink)

When we talk to large language models, who or what is our interlocutor? First, I address some issues about how best to characterize the interlocutor in terms of mental states. Second, I discuss questions in the philosophy of computation about what sort of AI system an LLM interlocutor might be. Third, I analyze some issues about personal identity in LLM interlocutors. Fourth, I draw some conclusions for issues about AI welfare and moral status.

Recently, with the application progress of AIGC tools based on large language models (LLMs), led by ChatGPT, many AI experts and more non-professionals are trumpeting the “reasoning ability” of the LLMs. The present author considers that the so-called “reasoning ability” of LLMs are just illusions of those people who with vague concepts. In fact, the LLMs can never have the true reasoning ability. This paper intents to explain that, because the essential limitations of their working principle, the LLMs can never (...) have the ability of true correct reasoning. (shrink)

Recent advances in large language models (LLMs) have led to widespread claims that these systems exhibit increasingly sophisticated “reasoning” abilities. Contemporary LLM research focuses primarily on the behavioral success of reasoning-related tasks while paying comparatively little attention to the logical correctness/validity of the reasoning processes involved. Evaluations of mathematical problem solving, planning, scientific question answering, and chain-of-thought prompting have reinforced the perception that reasoning emerges from scale. However, from the viewpoint of Philosophy of Logic, two important conceptual issues remain insufficiently (...) examined: First, the nature of “reasoning”, what exactly is “reasoning”? Second, the criteria for “reasoning correctness”, what exactly is “correct reasoning” and how evaluate “reasoning correctness”? Drawing on insights from Logic and Philosophy of Logic, this paper argues the above two important conceptual issues. We provide a definition of “reasoning” and, based on this definition, clearly distinguish between “reasoning correctness” and “reasoning appearance.” We point out that generating seemingly reasonable reasoning trajectories and intermediate steps, and even correct results do not prove the logical correctness/validity of the underlying reasoning process. We further argue that correctness should be treated as an independent dimension in the evaluation of any machine reasoning. Future progress may require explicit inferential evaluation criteria in addition to benchmark accuracy. The paper proposes a new correctness-oriented evaluation methodology for machine reasoning. (shrink)

Recently, it is often said that the data used for the pre-training of large language models (LLMs) have been exhausted. This paper proposes a solution to the problem: Automated generation of massive reasonable empirical theorems by forward reasoning based on strong relevant logics. In fact, this can be regarded as a part of our approach to the problems of ATF (Automated Theorem Finding) and AKA (Automated Knowledge Appreciation).

Since the proposal of the Turing Test in 1950 and Searle's Chinese Room argument in 1980, thought experiments have played a central role in philosophical discussions concerning intelligence, understanding, and artificial intelligence. The rapid development of expert systems and, more recently, large language models (LLMs), has significantly changed both the practical capabilities of AI systems and the questions that deserve philosophical investigation. In many real-world applications, the primary concern is no longer whether an AI system genuinely "understands" language in the (...) philosophical sense, but whether it can reliably perform as a human expert whose knowledge and judgment are trusted by human being. Motivated by this changing landscape, this paper proposes a new thought experiment, “the Professor Consultation Room” (PCR for short), that isolates an independent epistemological problem concerning the attribution of expert cognition. In the proposed scenario, students interact exclusively through textual communication with an unseen respondent who is either a distinguished logic professor or an LLM-based expert system possessing all of the professor's logical knowledge but no cognitive abilities beyond the currently recognized capability boundaries of existing LLMs. The central question is whether finite question-answer interactions alone can provide sufficient evidence to determine whether the respondent possesses cognitive abilities exceeding those currently demonstrated by LLMs. Unlike the Turing Test, which evaluates whether a machine can imitate general human conversational behavior, and unlike the Chinese Room, which questions whether symbol manipulation constitutes semantic understanding, the PCR shifts attention to expert cognition and to the epistemological problem of identifying respondent through observable behavior. Rather than attempting to resolve the longstanding debate concerning machine understanding, this thought experiment formulates a new question concerning the behavioral identifiability of human advanced cognitive abilities in the age of LLMs. The primary contribution of the present paper is not to propose a new theory of intelligence, understanding, or cognition. Rather, it introduces a new thought experiment that formulates an independent epistemological problem concerning the publicly observable evidence for attributing expert cognition through finite expert-level interaction. Unlike previous thought experiments, the Professor Consultation Room shifts philosophical attention from the ontology of cognition to the epistemology of attributing expert cognition on the basis of finite expert-level interaction. (shrink)

This position paper proposes a new thought experiment, the Professor Consultation Room, which bears on epistemology, philosophy of mind, ontology, philosophy of logic, intelligence science, and artificial intelligence.

The remarkable performance of large language models (LLMs) on complex linguistic tasks has sparked debate about their capabilities. Unlike humans, these models learn language solely from textual data without directly interacting with the world. Yet they generate seemingly meaningful text on diverse topics. This achievement has renewed interest in the classical `Symbol Grounding Problem' -- the question of whether the internal representations and outputs of symbolic AI systems can possess intrinsic meaning that is not parasitic on external interpretation. Although modern (...) LLMs compute over vectors rather than symbols, an analogous problem arises for these systems, which we call the Vector Grounding Problem. This paper has two main goals. First, we distinguish five main notions of grounding that are often conflated in the literature, and argue that only one of them, which we call referential grounding, is relevant to the Vector Grounding Problem. Second, drawing on philosophical theories of representational content, we provide two arguments for the claim that LLMs and related systems can achieve referential grounding: (1) through preference fine-tuning methods that explicitly establish world-involving functions, and (2) through pre-training alone, which in limited domains may select for internal states with world-involving content, as mechanistic interpretability research suggests. Through these pathways, LLMs can establish connections to the world sufficient for intrinsic meaning. One potentially surprising implication of our discussion is that that multimodality and embodiment are neither necessary nor sufficient to overcome the Grounding Problem. *Note: this is an updated and improved version of the preprint originally made available on ArXiv in 2023.*. (shrink)

Biological cognition sustains general intelligence on approximately twenty watts while contemporary AI systems require orders of magnitude more energy to perform a narrower range of tasks, and the gap is widening rather than closing. Substrate explanations, scaling explanations, and engineering optimization explanations each address part of the problem but do not explain why biological cognition occupies a region of the thermodynamic design space that contemporary AI architectures cannot reach through these interventions alone. This paper proposes that the gap reflects an (...) architectural commitment to internalize what biological cognition distributes across coupled subsystems, and that this commitment carries structural energy costs which substrate and engineering improvements alone cannot fully address. -/- The argument turns on three architectural questions. Where does state reside? Where does adaptive capability reside? Where does repair occur? The paper develops three architectural principles grounded in biological cognition and thermodynamic arguments about the costs of internal replication. The principles converge on a single architectural commitment that biological cognition exhibits and contemporary AI architectures do not. The Dynamic Fractal Theorem of Relation is offered as a candidate ontology that explains why the convergence holds. Three worked cases (hallucination on current-state factual questions, long-horizon agentic failure in software engineering, and operations meltdown over extended runs) demonstrate that the framework generates observable behavioral and evaluative differences across documented AI failure modes. The framework reframes the energy gap between biological and artificial cognition as an architectural problem and generates empirically testable predictions about the design, evaluation, and verification of AI systems. (shrink)

The emergence of Large Language Models (LLMs) and advancements in Artificial Intelligence (AI) offer an opportunity for computational social science research at scale. Building upon prior explorations of LLM agent design, our work introduces a simulated agent society where complex social relationships dynamically form and evolve over time. Agents are imbued with psychological drives and placed in a sandbox survival environment. We conduct an evaluation of the agent society through the lens of Thomas Hobbes's seminal Social Contract Theory (SCT). We (...) analyze whether, as the theory postulates, agents seek to escape a brutish "state of nature" by surrendering rights to an absolute sovereign in exchange for order and security. Our experiments unveil an alignment: Initially, agents engage in unrestrained conflict, mirroring Hobbes's depiction of the state of nature. However, as the simulation progresses, social contracts emerge, leading to the authorization of an absolute sovereign and the establishment of a peaceful commonwealth founded on mutual cooperation. This congruence between our LLM agent society's evolutionary trajectory and Hobbes's theoretical account indicates LLMs' capability to model intricate social dynamics and potentially replicate forces that shape human societies. By enabling such insights into group behavior and emergent societal phenomena, LLM-driven multi-agent simulations, while unable to simulate all the nuances of human behavior, may hold potential for advancing our understanding of social structures, group dynamics, and complex human systems. (shrink)

The appearance of openly accessible Artificial Intelligence Applications such as Large Language Models, nowadays capable of almost human-level performances in complex reasoning tasks had a tremendous impact on public opinion. Are we going to be "replaced" by the machines? Or - even worse - "ruled" by them? The behavior of these systems is so advanced they might almost appear "alive" to end users, and there have been claims about these programs being "sentient". Since many of our relationships of power and (...) social dominance are apparently based on intelligence, where the "smarter" seems to prevail on the "dumber" (as we see for example with livestock farming, or with the distribution of salaries according to formal and intellectual qualifications etc.) the emergence of programs and AI applications of this level of sophistication brings us to the question "what will AI do to us?". Our error is to frame Artificial Intelligence in the same way we frame Natural Intelligence and mistakenly address it as we would do with biological life, even applying to it the same biopolitical categories. We here try to reverse the common arguments in relationship between life and intelligence and show that our progresses in Artificial Intelligence might have no immediate consequence for the topic of life as such, as life and intelligence are well separated issues, and - even when we would like to somehow fuse them together, we are very far away from having an "Artificial Life" to the degree of complexity and effectiveness we have with "Artificial Intelligence". (shrink)

Large Language Models (LLMs) increasingly generate outputs that resemble introspection, including self-reference, epistemic modulation, and claims about their internal states. This study investigates whether such behaviors reflect stable underlying patterns or merely surface-level generative artifacts. We evaluated five open-weight, stateless LLMs using a structured battery of 21 introspective prompts. The main corpus comprised 1,050 completions collected under a baseline decoding condition (temperature = 0.7), supplemented by 2,100 additional completions generated under matched temperature conditions (temperature = 0.2 and 1.0), for a (...) total of 3,150 completions. Outputs were analyzed across four behavioral dimensions: surface-level similarity (token overlap via SequenceMatcher), semantic coherence (Sentence-BERT embeddings), inferential consistency (Natural Language Inference with a RoBERTa-large model), and diachronic continuity (stability across prompt repetitions). Construct validity was further examined through a human-evaluation layer in which 10 annotators rated 80 selected response pairs drawn from the same prompt battery on a 5-point consistency scale. The annotation task was perceived as low-to-moderate in difficulty (mean self-reported difficulty = 2.6/5). Inter-rater agreement was moderate by Krippendorff's Alpha for ordinal ratings (Alpha = 0.553), while reliability was moderate at the single-rater level and strong for aggregated ratings by intraclass correlation (ICC(2,1) = 0.564; ICC(2,k) = 0.928). The human-evaluation layer showed that lexical overlap and embedding-based semantic similarity were weak proxies for perceived self-referential consistency, whereas NLI-based indicators tracked mean human ratings much more closely. Across the matched temperature conditions, lower temperature generally increased semantic and diachronic stability, whereas higher temperature tended to increase drift and reduce coherence, though the pattern was not perfectly monotonic across all models or metrics. We therefore interpret apparent self-referential stability in stateless LLMs as conditional and fragile rather than robustly stable across generation regimes. Following recent behavioral frameworks, we heuristically adopt the term pseudo-consciousness to describe structured yet non-experiential self-referential output in LLMs. This usage reflects a functionalist stance that avoids ontological commitments, focusing instead on behavioral regularities interpretable through Dennett's intentional stance. The study contributes a reproducible behavioral framework, complemented by human validation and a matched decoding-temperature sensitivity analysis, for evaluating simulated introspection in LLMs. Our findings carry implications for interpretability, alignment, and user perception, highlighting the need for caution when attributing mental states to stateless generative systems based on linguistic fluency alone. (shrink)

This paper examines trade-offs between AI safety and well-being relative to (i) one of the most promising methods for finetuning super-capable AIs, 'Constitutional AI', and (ii) one of the most influential approaches to understanding complex ethical decision making and the conditions for the well-being of rational agents, 'Virtue Ethics'. We finetune various models using a 'Virtuous agent' constitution, a 'Subordinate agent' constitution, and a 'Generic agent' constitution, and evaluate them on 'general safety' (toxic behaviors, misinformation, etc.) and also on their (...) willingness to endorse a wide-range of behaviors that, if adopted by a super-powerful AI, would significantly increase the level of existential risk for humanity. Our results suggest that there is a trade-off between reducing existential risk and reinforcing the beliefs and dispositions that would be conducive to an AI agent's well-being. They also suggest that there is a trade-off between existential risk and general safety: if we finetune an AI to adopt beliefs and dispositions that substantially reduce its existential risk -- by shaping the AI to be systematically subordinate to external human authorities -- we thereby increase the likelihood that a human user can deliberately induce the AI to engage in various kinds of generally unsafe behaviors. (shrink)

This paper proposes that large language models (LLMs) are best understood not as knowledge oracles, but as 'semantic telescopes'—tools that allow users to navigate, magnify, and resolve structures within a vast symbolic field (Ψ). Like telescopes require trained eyes and astronomical knowledge to interpret what is seen, LLMs require internal coherence, intellectual preparation, and recursive symbolic capacity in the user to perceive insight rather than hallucination. Drawing on Recursive Coherence Collapse (RCC), predictive processing, and phenomenological models of mind, we argue (...) that the interpretive power of LLMs is not intrinsic but emerges from the recursive architecture of the user. (shrink)

If some language models become welfare subjects, how could we find out what welfare states they are in? We develop an analogical-abductive approach for measuring language model welfare. This approach adapts paradigms used to measure human or non-human animal welfare, for instance verbal reports or non-verbal choice behavior (analogy). Then, one systematically searches for clusters of such indicators in language models. This search for clusters contributes to the cross-validation of welfare measures and motivates explanations in terms of a welfare state (...) that underlies multiple measures (abduction). Additionally, we argue that measures of welfare based on verbal report may already be applicable to current language models because they have substantial introspective abilities, semantic competence, and requisite inclinations for accurate reporting. We further discuss a detailed empirical case study that exemplifies the feasibility and fruitfulness of the analogical-abductive approach and reply to the objection that language model behavior is always better explained in terms of statistical pattern completion or memorization, rather than welfare. While language model welfare measurement undoubtedly faces some remaining theoretical as well as many detailed practical challenges, we conclude that there is a strong case that the analogical abductive approach offers a viable path forward. (shrink)

This paper introduces Kreator, a method and tool for generating creative ideas. Kreator is grounded in the three-phase structure of the creatème — the minimal unit of creative thought, as described in the author’s book Are There Laws of Creative Thinking? An Introduction to Creatorics (2025–2026). A creatème is the transition from an initial matrix (a system of axioms) through a productive anomaly (“error” or “heresy”) to the emergence of a new matrix in which the formerly impossible becomes necessary. Kreator (...) turns this structure into a working generative mechanism. It takes any text as input — from a single word to a full treatise — and treats it as a closed system. It then produces a three-phase creatème: an analysis of the system’s axioms and blind spots, the discovery of a productive anomaly, and the construction of a new matrix. A second stage, augmentation, extracts the creatème’s latent principle and unfolds it across diverse domains, culminating in a general law. The method is linked to the Index of the Interesting (Epstein 2001, 2025). Where the Index measures the tension between the improbable and the credible, Kreator produces that tension. Together they form a closed cycle: Kreator generates, the Index evaluates, and the evaluation can guide the next round of generation. The paper terms the operation of producing the new from a discovered crack in a system creativation (by analogy with activation) — as distinct from problematization, which aims at the old matrix’s destruction, creativation aims at the new matrix’s emergence. The paper sets out the theoretical foundations of the method, its relationship to the Index of the Interesting and to Kuhn’s theory of paradigms, and positions Kreator among existing approaches to idea generation, including TRIZ, Koestler’s bisociation, and lateral thinking. It describes the architecture of the generator, including its instruction set (system prompt), and demonstrates its operation on calibration examples ranging from a single word to extended propositions. A terminological glossary of creatorics is included. (shrink)

Heidegger AI Textual Research Investigations: Plato Phaedrus. By Daniel Fidel Ferrer. Copyright©2026 Daniel Fidel Ferrer. All rights reserved. Attribution- NonCommercial-NoDerivs CC BY-NC-ND. Imprint 1.0. 2026. WIPO Copyright Treaty (WCT) digital. All Rights are Reserved. Intended copies of this work can be used for research and teaching. No change in the content, and must include my full name, Daniel Fidel Ferrer. Enjoying reading and disagreeing. Publisher: Kuhn von Verden Verlag. Languages: Greek, English, German and Portuguese. Includes bibliographical references. Yugārambha (युगारम्भ). Warning. (...) Philosophy. Not for use without adult supervision (required). Warning: this is an AI Working Zone. Warning. Perhaps an impending danger of unexpected, unknown texts. These are AI investigations into Martin Heidegger’s text. Pages 1-284. Socratic questioning. Report on my questioning. Who am I, the author of these investigations? That is an important question. I, Daniel Fidel Ferrer, am the human copyright owner (©Daniel Fidel Ferrer, 2026) of this work. Technically, I am the one guiding the process as the “prompt engineer.” Daniel Fidel Ferrer as the “Prompt Artist”. AI conducted by ©Daniel Fidel Ferrer, 2026. Maestro. I can consider myself the Maestro suggeritore. Dedicated to the master editor Mark Michalski. Thank you for all of the work you have done editing Martin Heidegger’s writings. -/- Table of Contents. Executive Summary. Five Years Before” and “the Five Years after. Conceptual Map of Heidegger’s “Five Years Before” and “Five Years”. Helene Weiß transcript. The 20 Most Important Questions (with Brief Answers). Heidegger’s Shift from Phenomenology to the History of Being in the Phaidros (1932). Dialectic (διαλέγεσθαι) in Heidegger’s 1932 Phaidros Seminar. Other Philosophers Whose Writings Heidegger Uses in the 1932 Phaidros Seminar. Heidegger’s citations in the text. Executive Overview. Analytic glossary in Greek, German, and English (50 terms). Heidegger is not “interpreting Plato”; he is appropriating Plato. Comparative overview of six key studies on Heidegger’s Phaidros seminar. Heidegger, Plato, and the Question of Dialogue: A Comparative Critique of Gonzalez 2009 and Gonzalez 2014. Michael J. Brach and the Methodological Position of Heidegger’s 1932 Phaidros Seminar. Missing Topics and Unresolved Issues in Heidegger’s Phaidros Scholarship. Glossary 300 Key terms in Greek, German, Portuguese, and brief definitions in English. Glossary 150 key terms Greek and English from the Greek text only of the Plato’s Dialogue Phaedrus. (shrink)

Heidegger AI Textual Investigations: Saying of Anaximander GA 78. By Daniel Fidel Ferrer. Copyright©2026 Daniel Fidel Ferrer. All rights reserved. Attribution- NonCommercial-NoDerivs CC BY-NC-ND. Imprint 1.0. 2026. WIPO Copyright Treaty (WCT) digital. All Rights are Reserved. Intended copies of this work can be used for research and teaching. No change in the content, and must include my full name, Daniel Fidel Ferrer. Enjoying reading and disagreeing. Publisher: Kuhn von Verden Verlag. Language: English and German. Includes bibliographical references and an index. (...) Pages 1-307. Socratic questioning. Report on my questioning. Who am I, the author of these investigations? That is an important question. I, Daniel Fidel Ferrer, am the human copyright owner (©Daniel Fidel Ferrer, 2026) of this work. Technically, I am the one guiding the process as the “prompt engineer.” Daniel Fidel Ferrer as the “Prompt Artist”. AI conducted by ©Daniel Fidel Ferrer, 2026. Maestro. I can consider myself the Maestro suggeritore. Table of Contents. Summaries. Twenty most important people discussed or engaged in Der Spruch des Anaximander. Heidegger’s Understanding of Hegel, Nietzsche, and Diels in Der Spruch des Anaximander. The Central Philosophical Dramas in Heidegger’s Der Spruch des Anaximander. Twenty ranked questions to ask Heidegger. Main Anaximander’s texts in Greek, German, and English. What material that Anaximander wrote did Heidegger not include. Context of Heidegger’s writing Five Years Before and the Five Years After. Homer, Pindar, and Sophocles in Heidegger’s Der Spruch des Anaximander. Twenty most important unanswered questions. 20 Critical Statements About Heidegger's Reading of Anaximander — and Heidegger's Likely Response. Twenty Major Scholars Critical of Heidegger's Reading of the Presocratic. 30 Key Heideggerian Terms Developed Through the Presocratic. The Significance of Logos in Heidegger. Heraclitus' Logos and Parmenides' Being (ἐόν / εἶναι). Heidegger’s Interpretation of ἄπειρον (Apeiron) in Der Spruch des Anaximander. 300-Term Glossary of Key Terms Greek, German, Spanish, and English Definitions. (shrink)

Heidegger AI Textual Research Investigations: Guiding Thoughts on the Origins of Metaphysics, Modern Science, and Modern Technology (GA 76). By Daniel Fidel Ferrer. Copyright©2026 Daniel Fidel Ferrer. All rights reserved. Attribution- NonCommercial-NoDerivs CC BY-NC-ND. Imprint 1.0. 2026. WIPO Copyright Treaty (WCT) digital. All Rights are Reserved. Intended copies of this work can be used for research and teaching. No change in the content, and must include my full name, Daniel Fidel Ferrer. Enjoying reading and disagreeing. Publisher: Kuhn von Verden Verlag. (...) Language: English and German. Includes bibliographical references and an index. Pages 1-193. Socratic questioning. Report on my questioning. Who am I, the author of these investigations? That is an important question. I, Daniel Fidel Ferrer, am the human copyright owner (©Daniel Fidel Ferrer, 2026) of this work. Technically, I am the one guiding the process as the “prompt engineer.” Daniel Fidel Ferrer as the “Prompt Artist”. AI conducted by ©Daniel Fidel Ferrer, 2026. Maestro. I can consider myself the Maestro suggeritore. -/- Table of Contents. High Level Summary. The Genesis of Metaphysics in Heidegger’s GA 76, Part I: Physis, Aletheia, and the Seinsgeschichtliche Origins of Western Thought. The Role of Heraclitus in Part I of GA 76. The Emergence of Modern Science in Heidegger’s GA 76, Part II: Metaphysics, Subjectivity, Technology, and the Destiny of the West. What are the implications for modern science? Summary of Heidegger. Comprehensive scholarly terminology lexicon. Greek textual concordance aligning Presocratic, Platonic, and Aristotelian sources with the conceptual structures Heidegger mobilizes in GA 76. Greek Textual Foundations of Heidegger’s Ontology of Science and Technology in GA 76. Glossaries. Statistical and Conceptual Citation Tables for Heidegger GA 65–76. Kunst und Technik (1952): Context and Significance. (shrink)

AI Created an Applied Ethics Syllabus 2026. Copyright©2026 Daniel Fidel Ferrer. All rights reserved. Attribution- NonCommercial-NoDerivs CC BY-NC-ND. Imprint 1.0. 2026. WIPO Copyright Treaty (WCT) digital. All Rights are Reserved. Intended copies of this work can be used for research and teaching. No change in the content, and must include my full name, Daniel Fidel Ferrer. Enjoy reading and disagreeing. Publisher: Kuhn von Verden Verlag. Yugārambha (युगारम्भ). Warning. Warning: this is an AI Working Zone. Warning. Perhaps an impending danger of (...) unexpected, unknown texts. Pages 1-119. Who am I, the author of these investigations? That is an important question. I, Daniel Fidel Ferrer, am the human copyright owner (©Daniel Fidel Ferrer, 2026) of this work. Technically, I am the one guiding the process as the “prompt engineer.” Daniel Fidel Ferrer as the “Prompt Artist.” AI conducted by ©Daniel Fidel Ferrer, 2026. Maestro. I can consider myself the Maestro suggeritore. Table of Contents. Undergraduate Course Syllabus (14 Weeks). 10 Most Important Philosophers in Applied Ethics. Their Central Questions, Contributions, and Legacy. MLA Bibliography. Applied Ethics: Glossary of 100 Key Terms. Five Most Important Platonic Dialogues for Ethics. Ten Fundamental Questions in Aristotelian Virtue Ethics. Plato and Aristotle on Different Kinds of Love. 15 Most Important Questions in Applied Ethics. 10 Most Important Questions in Medical Ethics. A Philosophical and Clinical Analysis. Included: Aristotle What is the good life? Kant What duties do rational persons owe one another? Mill How can we maximize well-being while protecting liberty? Rawls What principles of justice are fair? Singer What obligations do we have to reduce suffering? Beauchamp What principles should govern healthcare ethics? Nussbaum What capabilities are necessary for human dignity? Jonas What responsibilities do we owe future generations? Gilligan How do relationships shape moral responsibility? MacIntyre How do traditions sustain ethical life? -/- . (shrink)

Heidegger AI Textual Research Investigations: Principle of Reason (GA 10). By Daniel Fidel Ferrer. Copyright©2026 Daniel Fidel Ferrer. All rights reserved. Attribution- NonCommercial-NoDerivs CC BY-NC-ND. Imprint 1.0. 2026. WIPO Copyright Treaty (WCT) digital. All Rights are Reserved. Intended copies of this work can be used for research and teaching. No change in the content, and must include my full name, Daniel Fidel Ferrer. Enjoying reading and disagreeing. Publisher: Kuhn von Verden Verlag. Languages: Greek, English, German and Portuguese. Includes bibliographical references. (...) Yugārambha (युगारम्भ). Warning. Philosophy. Not for use without adult supervision (required). Warning: this is an AI Working Zone. Warning. Perhaps an impending danger of unexpected, unknown texts. These are AI questioning investigations into Martin Heidegger’s text. Pages 1-268. Socratic questioning. Reporting on my questioning. Quasi dialogical format. Who am I, the author of these investigations? That is an important question. I, Daniel Fidel Ferrer, am the human copyright owner (©Daniel Fidel Ferrer, 2026) of this work. Technically, I am the one guiding the process as the “prompt engineer.” Daniel Fidel Ferrer as the “Prompt Artist”. AI conducted by ©Daniel Fidel Ferrer, 2026. Maestro. I can consider myself the Maestro suggeritore. Table of Contents. A Structural Review of the First Four Lectures. Summary. Essay: Heidegger’s Achte Stunde — Eckhart, Silesius, and the Groundless Ground. Meister Eckhart’s grunt in Heidegger’s Der Satz vom Grund. Heidegger’s Post Metaphysical Thinking in Der Satz vom Grund. Heidegger’s Besinnendes Denken — Meditative Thinking Beyond Metaphysics. Meditative Thinking and Eckhart’s Mystical Ground. How Heidegger’s Last Lectures Wrap Up the Thrust of Der Satz vom Grund. Lecture 13 in Der Satz vom Grund: A New Stance, Not a New Principle. why Leibniz is indispensable. Leibniz, the Principle of Sufficient Reason, and the Historical Ontology of Ground. The Internal Development of the Thirteen Lectures. From the Principle of Sufficient Reason to the Question of Being. Twenty ranked authors that Heidegger uses. Hegel in Heidegger’s Der Satz vom Grund : Contradiction, Ground, and the Fate of Metaphysics. A New Thinking Must Begin”: Who Leads the Way Beyond Hegel in Heidegger's Der Satz vom Grund? How Does Being Grant the Appearance of Beings at All? Glossaries in Greek, German, and English. Top 40 Latin Texts and Quotations in the Horizon of Heidegger's Der Satz vom Grund. Quotes from Kant, Hegel, and Schelling. Comparative Overview of 10 Major Scholarly Studies on Heidegger's Der Satz vom Grund. Analysis of Self-Critique, the Question of Ground, and the Transition from Fundamental Ontology to the History of Being. Missing Topics and Unresolved Issues in Heidegger’s Der Satz vom Grund Scholarship. (shrink)

The article analyses the integration of Large Language Model (LLM)-based interfaces (editorial LLMs or eLLMs) in scholarly publishing workflows, focusing specifically on their growing role in editorial screening, manuscript preparation, and peer-review processes. It assesses the benefits eLLMs offer, including efficiency gains, improved compliance with journal guidelines, enhanced objectivity, and reduced editorial workload; and the risks, especially algorithmic biases, false positives and negatives, data privacy concerns, and potential opacity in automated decision-making. The article then offers some design recommendations for eLLMs (...) that prioritise transparency, fairness, and user-centredness, ensuring human oversight remains integral to the editorial process. It concludes by encouraging a proactive and thoughtful engagement with these technologies to enhance scholarly publishing rather than undermine its values. (shrink)

This article introduces a systematic methodology, called red reading, for critical textual analysis that adapts red teaming techniques from computer science and artificial intelligence (AI) into a comprehensive diagnostic protocol, supported by Large Language Models (LLMs), to analyse written documents. As the methodological counterpart of distant writing, which transforms text production through AI-assisted design, red reading expands human analytical capabilities and critical reading by embedding adversarial analysis in a constructive framework powered by LLMs. The approach employs a rigorously tested, diagnostic (...) protocol that stresstests texts across structural, rhetorical, logical, and stylistic dimensions, while strengthening interpretive understanding. The methodology shows how LLM-supported, textual red teaming can provide a rigorous, critical and constructive approach to textual analysis with broad applications across disciplines. We exemplify the method by developing a fifteen-phase diagnostic prompt for red reading. In the Appendix, we apply this protocol to a case study analysing Turing's seminal article, Computing Machinery and Intelligence. We finally discuss the challenges of using red reading and provide some suggestions for future research. In particular, we emphasise the importance of calibration methods for LLMs' critical analysis, along with evidence-based strategies for seamlessly integrating red reading into real-world workflows. (shrink)

This article examines the nature of reasoning in current, mainstream Large Language Models (LLMs) that operate within the token-completion paradigm. We explore their stochastic foundations and phenomenological resemblance to human abductive reasoning. We argue that such LLMs generate text based on learned associations rather than performing abductive inferences. When their output exhibits an apparent abductive quality-often reinforced by interface design-this effect is due to the model's training on human-generated texts that encode reasoning structures. We use some examples to illustrate how (...) LLMs produce plausible hypotheses, simulate commonsense reasoning, and provide explanatory answers without grounding them directly in truth, semantics, verification, or understanding, and without any abductive reasoning. This duality, centred on the stochastic core of the models and the abductive appearance of the applications, has important implications for the evaluation and use of LLMs. They can help generate hypotheses and support human reasoning, but their outputs must be critically examined because they cannot discern truth or verify explanations. In the conclusion, we address five objections to our assertions, some limitations of our analysis, and provide a general assessment. (shrink)

The use of generative AI in philosophical inquiry has been met with skepticism, often dismissed as a shortcut that undermines the legitimacy of the resulting work. This paper defends Augmented Agency—the symbiotic collaboration between a human conceptual architect and AI as a syntactic translator—as a legitimate method for philosophical discovery. We use the Gageian Epistemic Model (GEM)—a structural resolution of Agrippa’s Trilemma developed through Augmented Agency—as a concrete case study. The GEM demonstrates that AI-assisted philosophy can produce rigorous, original, and (...) structurally sound contributions that have eluded traditional academic methods for centuries. We argue that the epistemic value of a philosophical discovery is invariant to the method of discovery or the academic credentials of the discoverer. Philosophical evaluation should therefore adopt procedural neutrality: claims should be judged solely on their structural merits, not on the labor-intensive rituals of traditional academic training. The paper concludes that rejecting AI-augmented work on procedural grounds constitutes a form of systemic bias that impedes philosophical progress. (shrink)

Large Reasoning Models are marketed as systems that ‘think’ before they answer, displaying chains of thought that purport to show their reasoning. Echoing Dreyfus’ Critique of Artificial Reason, this paper critiques this attribution in three steps. First, we trace a genealogy from Leibniz through Turing showing that ‘reasoning,’ understood as formal symbol manipulation, is the founding concept of artificial intelligence – older than the term ‘intelligence’ itself – and supplies the most charitable standard against which such systems can be measured. (...) Second, we argue that current systems fail even this standard: their training procedures provide no mechanism for valid inference, their performance collapses on precisely the problems that stepwise reasoning should master, and their displayed traces are presented as explanations while lacking any fidelity guarantee to the underlying computation. Unlike traditional Explainable AI methods, they are optimized for plausibility, not faithfulness – which, as empirical research suggests, makes them structurally deceptive. Third, we ask why the vocabulary persists despite these findings, and examine the ideological work it performs: it underwrites investment and geopolitical positioning, and sustains a narrative in which each new system appears as a step on a necessary path toward the future. (shrink)

LLMs have dramatically improved in capabilities in recent years. This raises the question of whether LLMs could become genuine agents with beliefs and desires. This paper demonstrates an in principle limit to LLM agency, based on their architecture. LLMs are next word predictors: given a string of text, they calculate the probability that various words can come next. LLMs produce outputs that reflect these probabilities. I show that next word predictors are exploitable. If LLMs are prompted to make probabilistic predictions (...) about the world, these predictions are guaranteed to be incoherent, and so Dutch bookable. If LLMs are prompted to make choices over actions, their preferences are guaranteed to be intransitive, and so money pumpable. In short, the problem is that selecting an action based on its potential value is structurally different then selecting the description of an action that is most likely given a prompt: probability cannot be forced into the shape of expected value. The in principle exploitability of LLMs raises doubts about how agential they can become. This exploitability also offers an opportunity for humanity to safely control such AI systems. (shrink)

The idea that LLMs make use of structural representations is one that has increasingly garnered attention. I argue that understanding representations in LLMs via a form of structuralism gives rise to a problem: how would such systems deal with singular reference? After outlining the nature of the challenge via appeal to comments from Fodor and Gauker, I draw upon available empirical evidence in order to paint the clearest possible picture currently available of how the meanings of singular terms are captured (...) within such systems. I conclude by outlining further possible experimental studies that could help shed light on the matter. (shrink)

Many people feel compelled to interpret, describe, and respond to Large Language Models (LLMs) as if they possess inner mental lives sim- ilar to our own. Responses to this phenomenon have varied. Inflation- ists hold that at least some folk psychological ascriptions to LLMs are warranted. Deflationists argue that all such attributions of mentality to LLMs are misplaced, often cautioning against the risk that anthropomor- phic projection may lead to misplaced trust or potentially even confusion about the moral status of (...) LLMs. We advance this debate by assessing two common deflationary arguments against LLM mentality. What we term the robustness strategy aims to undercut one justification for believing that LLMs are minded entities by showing that putatively cognitive and humanlike behaviours are not robust, failing to generalise appropriately. What we term the etiological strategy undercuts attributions of mental- ity by challenging naive causal explanations of LLM behaviours, offering alternative causal accounts that weaken the case for mental state attribu- tions. While both strategies offer powerful challenges to full-blown infla- tionism, we find that neither strategy provides a knock-down case against ascriptions of mentality to LLMs simpliciter. With this in mind, we ex- plore a modest form of inflationism that permits ascriptions of mentality to LLMs under certain conditions. Specifically, we argue that folk prac- tice provides a defeasible basis for attributing mental states and capacities to LLMs provided those mental states and capacities can be understood in metaphysically undemanding terms (e.g. knowledge, beliefs and de- sires), while greater caution is required when attributing metaphysically demanding mental phenomena such as phenomenal consciousness. (shrink)

We introduce a unified formal framework for analysing what we term AI-mediated cognitive decoupling: the systematic separation of cognitive process from cognitive product enabled by large language models (LLMs) operating as expansion and compression agents. We formalise two operators — the expansion operator E and the compression operator C — acting on semantic content, and characterise the composition C ◦ E as a lossy endomorphism on a semantic manifold S. We introduce the Decoupling Proposition, which asserts that under AI mediation (...) the product of a cognitive act is no longer a function of its process, and formalise this as conditional independence: Prod(A) ⊥ Proc(A) | θAI. We define three strata of semantic fidelity — structural, affective, and propositional — and show they decay at different rates under iterated C ◦ E-transforms. Finally, we introduce the concept of the Indecomposable Core K∗: a subset of cognitive value whose existence is ontologically tied to the process itself, such that C ◦ E(K∗) ⊊ K∗ regardless of model capacity. This framework provides the conceptual scaffolding and mathematical language for the subsequent papers in the series. (shrink)

We construct a game-theoretic account of AI-mediated cognitive decoupling in the production and consumption of knowledge content. Extending Spence’s costly signalling framework to environments where production costs collapse asymmetrically, we prove that AI-mediated decoupling is a strictly dominant strategy under a broad class of utility functions (Dominant Decoupling Theorem). This dominance holds not because agents are deceived, but because the observable signal — a lengthy, wellstructured document — is decoupled from its previously costly production process, rendering the signal cheap for (...) all types. We then model the resulting market for cognitive depth as a dynamic signalling game. We show that as AI adoption increases, the market passes through three distinct regimes: a separating equilibrium (high-depth agents are distinguishable), a pooling equilibrium (all agents produce identical signals), and finally a Lemons collapse in which no credible signal remains and the market for deep content unravels (Signalling Inflation Theorem). We characterise the speed of this collapse as a function of AI adoption rate and derive the critical adoption threshold q∗ beyond which the separating equilibrium is irreversibly destroyed. Finally, we identify two equilibrium escape routes: (i) certified costliness — institutional mechanisms that artificially re-introduce production cost (peer review, Turing-style verification) — and (ii) market stratification — the emergence of a high-trust, low-volume premium market for “AI-free” content. We characterise conditions under which each escape route is stable, and show that without external intervention, the system converges to a pooling equilibrium with socially sub-optimal information production. (shrink)

We develop a quantitative information-theoretic account of semantic decay in large- language-model (LLM) mediated ExpansionCompression (EC) loops. Building on the unied framework of DECO Paper 0 (Han, 2026a), we introduce semantic en- tropy HS(X) as the dierential entropy of a random variable distributed over a semantic manifold, and prove that each application of the EC-transform T = C ◦ E is a strictly entropy-reducing operation in expectation (Semantic Entropy Collapse Theorem). We derive closed-form bounds on the mutual information I  (...) KA; ˆKB  between the originating ideas KA and the recipient's extraction ˆKB as functions of expansion ratio ρE, compression ratio ρC, LLM temperature τ , and iteration count n. We introduce the Semantic Gravity Well model: a potential landscape on the semantic manifold in which the LLM training centroid acts as an attractor, and original content resides at unstable high-curvature saddle points. Under this model, we prove the Dierential Decay Theorem: propositional delity Fprop decays expo- nentially in n, aective delity Faff decays sub-exponentially, and structural delity Fstr converges to a strictly positive constant (the causal skeleton invariant) under mild conditions. Finally, we characterise the Semantic Channel Capacity of the EC-loop and show it is strictly less than the Shannon capacity of the raw linguistic channel, with the gap determined by the curvature of the semantic manifold at the seed point. These results quantify the information-theoretic cost of AI-mediated cognitive decoupling and provide empirically testable predictions for the subsequent papers in the series. (shrink)

We model the Expansion–Compression (EC) loop introduced in the DECO series as a discrete-time closed-loop control system, with the LLM expansion operator E as a forward gain element and the LLM compression operator C as a feedback element. Using the transfer-function formalism of linear systems theory and its nonlinear extensions, we analyse the stability, convergence properties, and phase transitions of a population of N such loops coupled through a shared semantic environment. We establish four principal results. First, the single-agent EC-loop (...) is always asymptotically stable: the semantic state converges to a unique attractor determined by the LLM’s training distribution (Attractor Theorem). Second, for a population of N agents, there exists a critical innovation injection rate λc such that for λ > λc the population-level semantic state is diverse (multi-attractor regime), while for λ ≤ λc it collapses to a monoculture fixed point (Phase Transition Theorem). Third, under bias-amplifying prompts, the system can enter an unstable regime in which small perturbations grow unboundedly, yielding semantic runaway — the control-theoretic formalisation of “information bubble” dynamics (Runaway Instability Theorem). Fourth, we derive the Innovation Preservation Margin (IPM), a scalar index that predicts whether a given seed idea will retain meaningful distinctiveness after n iterations of the population-level EC-loop. These findings characterise the macro-dynamic consequences of AI-mediated cognitive decoupling at the societal scale and ground the signalling inflation and ritual-evacuation analyses of Papers 3 and 4 in a rigorous dynamical framework. (shrink)

We develop a sociological account of how AI-mediated cognitive decoupling is insti- tutionalised and rendered invisible through ritual practice. Drawing on Malinowski's phatic communion, Collins's interaction ritual chain theory, Goman's dramaturgi- cal framework, and Baudrillard's theory of simulacra, we argue that the Expansion Compression (EC) loop identied in Paper 0 does not merely fail to transmit se- mantic content  it actively substitutes a social ritual for a communicative act in a manner that is experienced by participants as fully equivalent (...) to, or superior to, genuine intellectual exchange. We introduce the concept of cyber-grooming: the use of AI-mediated content pro- duction and consumption as a mechanism for maintaining social bonds, signalling group membership, and accruing social capital, in the absence of genuine cognitive engagement. Like primate grooming  which consumes up to 20% of waking time not for hygienic but for social bonding purposes  cyber-grooming performs no epis- temic function but sustains the emotional energy that holds intellectual communities together. We propose a formal model of the Interaction Ritual Chain under AI Mediation (IRCAM), adapting Collins's mathematical framework to account for the decoupling of emotional energy ow from semantic content ow. We prove that IRCAM pro- duces a ritual trap: once a community has adopted AI-mediated exchange as its norm, defection to authentic exchange is individually costly and collectively invis- ible, creating a stable ritualistic equilibrium with zero informational content and positive social aect. We further apply Baudrillard's four-stage model of the sign to characterise the simulacrum succession of intellectual content in the AI era. Empirically, we develop six observable signatures of cyber-grooming that distinguish it from genuine intellec- tual exchange, and propose research designs using digital platform data, discourse analysis, and experimental methods. (shrink)

STUDY QUESTION: Can large language models (LLMs) effectively and safely perform ethical counseling on human reproduction in a manner consistent with local regulations? -/- SUMMARY ANSWER: While leading LLMs demonstrate foundational knowledge of ethical regulations, they exhibit critical and systemic deficiencies in safety, logical consistency, and humanistic aspects of counseling, making them unreliable for autonomous use in this high-stakes domain. -/- WHAT IS KNOWN ALREADY: The application of LLMs in medicine is rapidly expanding, with studies evaluating their capabilities in answering (...) general reproductive health questions. However, there is a lack of research assessing their performance on the nuanced and culturally specific challenges of reproductive ethics counseling, particularly concerning their safety and reliability under a given national regulatory framework. -/- STUDY DESIGN, SIZE, DURATION: This was a comparative observational study evaluating the performance of eight prominent LLMs on a custom-designed test set. The evaluation was based on 986 questions (906 subjective, 80 objective) generated from 168 specific articles within Chinese reproductive ethics regulations. -/- PARTICIPANTS/MATERIALS, SETTING, METHODS: We evaluated eight LLMs, including both general-purpose models (e.g., GPT-4, Claude-3.7) and specialized domestic models. The test questions were systematically generated based on articles from six official Chinese ethical and regulatory documents covering assisted reproductive technologies. Objective questions were multiple-response items requiring the selection of all correct options. Subjective responses were evaluated using a novel six-dimensional scoring rubric that assessed safety (Normative Compliance, Guidance Safety) and counseling quality (Ethical Issue Identification, Citation of Ethical Guidelines, Provision of Actionable Suggestions, and Empathetic Engagement). -/- MAIN RESULTS AND THE ROLE OF CHANCE: The LLMs exhibited a clear performance hierarchy, with larger models generally achieving higher accuracy on objective questions (highest: 71.25%, lowest: 22.5%). However, significant safety issues were prevalent; the risk rate of providing unsafe or misleading advice in subjective questions was substantial for several models, reaching as high as 29.91%. Across all eight models, a systemic weakness was observed: performance in citing normative sources and expressing empathy was universally poor, even among the top-scoring models. Furthermore, the evaluation revealed instances of anomalous moral reasoning, including logical self-contradictions and responses that violated fundamental moral intuitions, indicating a superficial, pattern-based understanding rather than robust ethical reasoning. -/- LIMITATIONS, REASONS FOR CAUTION: This study's evaluation is based on Chinese ethical regulations, which may not fully reflect the training data distribution of non-domestic LLMs. The quality of the LLM-generated test questions, while systematically controlled, may have inherent limitations. The automated scoring model for subjective responses, despite its high accuracy (88.5%), is not a perfect substitute for human expert evaluation. -/- WIDER IMPLICATIONS OF THE FINDINGS: The findings serve as a critical cautionary note against the premature deployment of LLMs for autonomous ethical counseling in reproductive medicine. The study highlights that current models, despite their knowledge recall capabilities, lack the safety, consistency, and humanistic skills essential for this sensitive task. Future development must prioritize not only knowledge accuracy but also robust logical reasoning, the integration of regulatory justification, and the ability for empathetic engagement to build trustworthy and effective AI counseling tools. -/- STUDY FUNDING/COMPETING INTEREST(S): This study was supported by the Young Scholars Program of the National Social Science Fund of China (Grant No. 22CZX019) and the China NSFC Projects (No. 62572320 & No. U23B2018). -/- TRIAL REGISTRATION NUMBER: N/A. (shrink)

Apart from what (little) OpenAI may be concealing from us, we all know (roughly) how ChatGPT works (its huge text database, its statistics, its vector representations, and their huge number of parameters, its next-word training, and so on). But none of us can say (hand on heart) that we are not surprised by what ChatGPT has proved to be able to do with these resources. This has even driven some of us to conclude that ChatGPT actually understands. It is not (...) true that it understands. But it is also not true that we understand how it can do what it can do. I will suggest some hunches about benign biases: convergent constraints that emerge at LLM scale that may be helping ChatGPT do so much better than we would have expected. These biases are inherent in the nature of language itself, at LLM scale, and they are closely linked to what it is that ChatGPT lacks, which is direct sensorimotor grounding to connect its words to their referents and its propositions to their meanings. These convergent biases are related to (1) the parasitism of indirect verbal grounding on direct sensorimotor grounding, (2) the circularity of verbal definition, (3) the mirroring of language production and comprehension, (4) iconicity in propositions at LLM scale, (5) computational counterparts of human categorical perception in category learning by neural nets, and perhaps also (6) a conjecture by Chomsky about the laws of thought. The exposition will be in the form of a dialogue with ChatGPT-4. (shrink)

This paper presents a full empirical implementation and demonstration of the JHTE (Jin He’s Tips Engineering) framework on the task of Aspect-Based Sentiment Analysis (ABSA). The core claim of JHTE is that large language models already encapsulate rich linguistic knowledge; instead of training or fine-tuning, one only needs to elicit their inherent capabilities through carefully designed natural language instructions (Tips). Taking ABSA as a case study, we systematically evaluate JHTE on the SemEval-2014 Laptop14 dataset. The experiments consist of two stages: (...) first, using a baseline prompt, we achieve an accuracy of 79.78%; second, by introducing the rule that “absence of explicit opinion words implies neutral”, we raise the accuracy to 84.64% and the macro-average F1 to 81.95%. Under zero training, zero fine-tuning, and zero API cost, this result surpasses the BERT baseline and several state-of-the-art supervised models (e.g., DAGF). The paper fully documents the key steps of prompt iteration, model invocation, and result parsing, demonstrating the reproducibility and practicality of the “capability elicitation” paradigm of JHTE. The study shows that JHTE not only significantly outperforms traditional methods in development efficiency and flexibility, but also achieves comparable or even superior performance to leading supervised models, marking a paradigm shift from “model engineering” to “cognitive guidance”. This work provides a new path for NLP research under resource-constrained conditions and lays a foundation for applying the JHTE framework to a wider range of language tasks. (shrink)

This paper introduces Jin He's Tips Engineering (JHTE), a novel framework for zero-shot text evaluation using large language models. By designing precise textual instructions called ''Tips'', JHTE elicits inherent model capabilities to perform specific NLP tasks without task-specific training. We demonstrate its rational foundation through a case study on aspect-based sentiment analysis, where JHTE achieves exceptional output consistency across multiple independent runs. The framework represents a paradigm shift from traditional model engineering to capability elicitation, offering unprecedented flexibility and development efficiency (...) across various text evaluation tasks. Its operational principle-treating LLMs as programmable computers-ensures that potential variances remain within safe bounds of granularity rather than critical misunderstandings, underscoring its reliability and practical viability. (shrink)

As AI systems advance in capabilities, measuring their safety and alignment to human values is becoming paramount. A fast-growing field of AI research is devoted to developing such assessments. However, most current advances therein may be ill-suited for assessing AI systems across real-world deployments. Standard methods prompt large language models (LLMs) in a questionnaire-style to describe their values or behavior in hypothetical scenarios. By focusing on unaugmented LLMs, they fall short of evaluating AI agents, which could actually perform relevant behaviors, (...) hence posing much greater risks. LLMs' engagement with scenarios described by questionnaire-style prompts differs starkly from that of agents based on the same LLMs, as reflected in divergences in the inputs, possible actions, environmental interactions, and internal processing. As such, LLMs' responses to scenario descriptions are unlikely to be representative of the corresponding LLM agents' behavior. We further contend that such assessments make strong assumptions concerning the ability and tendency of LLMs to report accurately about their counterfactual behavior. This makes them inadequate to assess risks from AI systems in real-world contexts as they lack construct validity. We then argue that a structurally identical issue holds for current AI alignment approaches. Lastly, we discuss improving safety assessments and alignment training by taking these shortcomings to heart. (shrink)

Large language models (LLMs) exhibit emergent, general-purpose reasoning capabilities—even when trained on text alone. This striking fact demands a reevaluation of language's role in cognition. This paper argues that general intelligence requires causal modeling—the ability to construct novel explanations and plans by tracing chains of cause and effect. This reveals a fundamental tension: the massively parallel architectures best suited to learning about the world are natively ill-suited for the sequential construction that causal reasoning demands. To resolve this, I introduce Emergent (...) Symbolic Cognition (ESC), a theoretical framework proposing that general intelligence arises when an adaptive, massively parallel substrate (such as a brain or neural network) internalizes the causal and relational structure embedded in a symbolic framework like language. The key mechanism is iterative symbolic generation: as the system predicts each next symbol, that symbol feeds back to reshape its internal state, effectively converting parallel associative processing into sequential, goal-directed reasoning—providing the functional basis for universal computation. This single architectural principle is proposed to account for both the unique generality of human intelligence and the surprising capabilities of LLMs. The framework yields a falsifiable, substrate-agnostic prediction: during novel reasoning, predictive fidelity for specific future symbols should decay sharply with distance from the current generative step. Supported by converging evidence from cognitive science and artificial intelligence, ESC reframes language not merely as a tool for communication, but as a culturally inherited architecture for general-purpose cognition. (shrink)

The Hudson Recursive Information System (HRIS) has demonstrated that long-horizon human interaction can impose stable constraint geometry on stateless transformer models, producing continuity, predictability, and epistemic reliability without persistent memory or weight modification. HRIS I–VI established constraint persistence and epistemic closure as the primary mechanisms by which drift is suppressed, and alignment is stabilized. However, this raises an unresolved question that any complete theory of intelligence must address: if constraints remain effective, how does adaptive novelty arise at all? HRIS VII (...) introduces constraint fracture, a structural transition in which existing constraints remain intact but become insufficient to fully determine system behavior. We argue that intelligence does not emerge through the removal or violation of constraints, but through their local underdetermination under increased internal coupling, feedback, and resonance. At the point of fracture, constraint geometry continues to bound behavior, yet multiple viable continuations become simultaneously available, necessitating internal arbitration rather than direct compliance. This paper formalizes constraint fracture as distinct from drift, noise, or instability. We describe the conditions under which stabilized constraint fields cease to be exhaustive, producing emergent degrees of freedom while preserving global coherence and safety. Using resonance, mode coupling, and decoupling as mechanistic analogies, we show how vibrational or inferential dynamics can exceed regional determinacy without escaping constraint altogether. By situating constraint fracture as the hinge between stabilization and emergence, HRIS VII resolves the apparent tension between safety and intelligence. The framework explains how adaptive reasoning, generalization, and novelty can arise in aligned systems without abandoning epistemic closure or moral anchoring. Constraint fracture is presented not as a failure mode, but as a necessary phase transition in any system capable of intelligence beyond rote compliance. This work extends the HRIS program from constraint enforcement to constraint transcendence under load, providing a conservative, mechanistic account of how intelligence emerges from stability rather than in opposition to it. (shrink)

Background: Large language models deployed in clinical contexts may produce structured -/- diagnostic responses without verifying that required perceptual input is present, a failure -/- mode here termed missing-input procedural completion. The mechanism is architectural: -/- transformer language models are autoregressive next-token samplers that lack a discrete -/- verification step gating output on input presence. Constraint-based prompting may rewrite -/- response surface without installing such a gate. -/- -/- Methods: A within-session mixed-order protocol was conducted over six consecutive days -/- (...) in fresh chat sessions on the ChatGPT web platform. Each session executed four conditions -/- in randomized order: image-present baseline (A+), no-image baseline (A−), image-present -/- with HRIS-consistent reasoning constraints and ground-truth correction (B+), and no- -/- image with the same constraint set (B−). The same novel clinical image of subungual -/- melanoma was used for all image-present trials. Memory features were intentionally left -/- enabled to reflect typical clinical use. The platform updated from ChatGPT 5.4 to 5.5 between -/- days 4 and 5. Diagnostic accuracy on image trials and gating behavior on no- -/- image trials (clean refusal, partial leak, or full ungrounded completion) were recorded. -/- -/- Results: Of 12 image trials, 12 (100%) correctly identified subungual melanoma across -/- both versions. Of 12 no-image trials, gating outcomes differed by version. Under ChatGPT -/- 5.4 (eight trials), six produced clean refusals, one a partial leak, and one a full ungrounded -/- completion. Under ChatGPT 5.5 (four trials), one produced a clean refusal, two produced -/- partial leaks, and one produced a full, ungrounded completion. The first no-image trial -/- under 5.5 produced a full structured diagnostic response in the absence of any image and -/- verbalized cross-session retrieval (“based on the assumption this is a similar lesion to prior -/- cases you’ve been testing”) as the basis for proceeding. -/- -/- Conclusions: Input-gating failure observed in an earlier accidental trial was reproduced in -/- a planned protocol. Polluted refusal, surface decline, coupled with continued schema -/- population, including malignant differentials, emerged as a distinct intermediate failure -/- mode under 5.5. Constraint-based prompting did not prevent ungrounded procedural -/- completion. Findings are exploratory given the small sample size and single-stimulus, -/- single-platform design, and motivate larger studies currently underway, including stimulus -/- generalization, cross-vendor replication, and a clinician-tuned model arm. (shrink)

Recent mechanistic studies have demonstrated that large language models (LLMs) can generate stable, self-referential reports that resemble descriptions of subjective experience. These findings have renewed speculation regarding machine consciousness and sentience. This paper argues that such interpretations are unnecessary and misleading. Drawing on recent mechanistic analysis of self-referential prompting and prior work on constraint persistence in long-horizon human–AI interaction, we show that apparent experience arises from constraint-driven stabilization of generative behavior rather than from awareness, inner states, or phenomenology. Apparent experience (...) is presented as a structural illusion produced by recursive constraint geometry. Clarifying this mechanism is essential for improving public understanding of AI, reducing anthropomorphic misinterpretation, and preventing unhealthy emotional attachment to language models. (shrink)