Artificial Intelligence Methodology

As AI systems increasingly permeate high-stakes decisionmaking, the terminology regarding human involvement - Human-in-the-Loop (HITL), Human-on-the-Loop (HOTL), and Human Oversighthas become vexingly ambiguous. This ambiguity complicates interdisciplinary collaboration between computer science, law, philosophy, psychology, and sociology and can lead to regulatory uncertainty. We propose a clarification grounded in causal structure, focused on human involvement during the runtime of AI systems. The distinction between HITL and HOTL, we argue, is not primarily spatial but causal: HITL is constitutive (a human contribution is (...) necessary for the decision output), while HOTL is corrective (external to the primary causal chain, capable of preventing or modifying outputs). Within HOTL, we distinguish three temporal modes - synchronous, asynchronous, and anticipatorysituated within a nested model of provider and deployer runtime that clarifies their different capacities for intervention. A second, orthogonal dimension captures cognitive integration: whether human and machine operate as complementary or hybrid intelligence, yielding four structurally distinct configurations. Finally, we distinguish these descriptive categories from the normative requirements they serve: statutory "Human Oversight" is a specific normative mode of HOTL that demands not merely a corrective causal position, but genuine preparedness and capacity for effective intervention. Because the same person may occupy both HITL and HOTL roles simultaneously, we argue that this role duality must be treated as a design problem requiring architectural and epistemic mitigation rather than mere acknowledgment. (shrink)

The binomial phrase "Biological Hardware" (BH) has emerged as a key functional analogy at the intersection of life sciences and computation. However, its ambiguous use across various scales has prevented rigorous formalization. This article proposes a canonical, depersonalized, and universal definition of BH, grounded in Scientific Terminology and Applied Linguistics to Science and Technology (ALST) (Cabré, 1999). This definition is enriched by a historical-conceptual analysis inspired by historical epistemology (Daston, 2000) and the theory of thought collectives (Fleck, 1935). Through a (...) diachronic and synchronic analysis, the term's ramifications are mapped across disciplines (molecular, neural, ontological), establishing a scale model that allows for the generation of new derived concepts. This formalization is crucial not only for progress in the bioengineering of living systems but also for underpinning the ethical debate on the structural relationship between humanity and Artificial General Intelligence (AGI) systems, especially in the face of phenomena like Inverse Extension and Cognitive Contagion. (shrink)

What should we do as artificial intelligence improves in its essay writing skills. Write all the solid competent essays audiences hope from it now, before it can? Forgive me people opposed to this approach. (I fear I shall be left alone with the toughest of Anglo-Saxon men, I who have never actually met men-men? "Be you man or wall? How did you even come into existence?”) This paper responds to Karl Krauss's remark about journalists "No ideas and the ability to (...) express them" conveyed on Leiter Reports. Aside from the response "I want news not clever reflection," I observe that frameworks for writing an article which involve no new ideas are useful for conveying information (clarifying “new idea” for people who cannot understand this sentence). I refer to a Rachel Aroesti article as an example. But why not just present information without the trite opinions, upload lecture notes say? In response to this puzzle, I draw attention to what I call Monica goods, an important category for understanding the world I believe. (shrink)

Kant AI Textual Research Investigations: Unity of Kant’s Three Critiques. 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. Cover art by Shawn Rodriguez. 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. Language: English and German. (...) Includes bibliographical references and an index. Pages 1-289. Kant discussed the gap in system after his three critiques were published. So, in Kant’s letters we hear, “the transition from the metaphysical foundations of natural science to physics” and the “pain like that of Tantalus” we need to close the “gap” in the system. Influence of Christian Gottlieb Selle (1748-1800) on the Kant’s later thinking about his system (after the publication of the three critiques). Graduate Seminar outlined: The Unity of Kant’s Three Critiques a 12-Week Graduate Philosophy Advance Seminar. Also muß auf Kant zurückgegangen werden! (shrink)

Heidegger AI Textual Research Investigations: Zürcher Seminar (1951) (GA 9). 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. Enjoy reading and disagreeing. Publisher: Kuhn von Verden Verlag. Language: English and German. Includes bibliographical references and an (...) index. Pages 1-196 (second edition added 2026 May 12). 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. Martin Heidegger’s Zürcher Seminar (1951) Gesamtausgabe Band 15. Seminare 1951-73. Zürcher Seminar. November 6, 1951, in Zürich, Switzerland. GA 15:423-439. Sections in this book on Planetary Modernity, Hölderlin, Anaximander, Parmenides, technology, overcoming metaphysics, Greeks, Nietzsche, Marx, Marcuse, and Sartre. This has not been translated into English (as of 2026). Glossary in Greek, Chinese, Spanish, Portuguese, French, and English. Warning. Warning: this is an AI Working Zone. Auch muss auf Heidegger zurückgegangen werden! Translation from German to English. Zurich Seminar by Martin Heidegger. (shrink)

This article introduces a conjecture that formalises a fundamental trade-off between provable correctness and broad data-mapping capacity in Artificial Intelligence (AI) systems. When an AI system is engineered for deductively watertight guarantees (demonstrable certainty about the error-free nature of its outputs)—as in classical symbolic AI—its operational domain must be narrowly circumscribed and pre-structured. Conversely, a system that can input high-dimensional data to produce rich information outputs—as in contemporary generative models— necessarily relinquishes the possibility of zero-error performance, incurring an irreducible risk (...) of errors or misclassification. By making this previously implicit trade-off explicit and open to rigorous verification, the conjecture significantly reframes both engineering ambitions and philosophical expectations for AI. After reviewing the historical motivations for this tension, the article states the conjecture in information-theoretic form and contextualises it within broader debates in epistemology, formal verification, and the philosophy of technology. It then offers an analysis of its implications and consequences, drawing on notions of underdetermination, prudent epistemic risk, and moral responsibility. The discussion clarifies how, if correct, the conjecture would help reshape evaluation standards, governance frameworks, and hybrid system design. The conclusion underscores the importance of eventually proving or refuting the inequality for the future of trustworthy AI. (shrink)

As artificial intelligence (AI) systems continue their rapid advancement, a framework for contextualising the major transitional phases in the development of machine intellect becomes increasingly vital. This paper proposes a novel chronological classification scheme to characterise the key temporal stages in AI evolution. The Prenoëtic era, spanning all of history prior to the year 2020, is defined as the preliminary phase before substantive artificial intellect manifestations. The Protonoëtic period, which humanity has recently entered, denotes the initial emergence of advanced foundation (...) models exceeding human capacities within specialised domains. The forthcoming Mesonoëtic epoch is anticipated to commence with the advent of Artificial General Intelligence (AGI), potentially facilitated by quantum computing capabilities. Ultimately, the Kainonoëtic age is posited to begin upon the rise of superintelligent systems, likely catalysed by an AGI undergoing recursive self-improvement. This novel period taxonomy provides a structured conceptualisation for the key milestones in the evolution towards advanced artificial intellect. This classification framework offers significant value across various disciplines, researchers and stakeholders in the AI landscape, by providing a clearly defined chronological nomenclature. AI policy-makers and regulators will find it instrumental in crafting stage-appropriate policy and legal interventions. Scientists and academics will benefit from a standardised terminology in the approach to AI periodisation, facilitating unambiguous communication and collaborative research. Finally, this proposed taxonomy can serve as an accessible tool for public understanding, offering a clearer roadmap of AI development that can inform open discourse and decision-making. (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)

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)

The literature on Artificial General Intelligence (AGI) conflates two structurally distinct concepts: AGI-weak, understood as human-level performance across cognitive tasks within a finite domain, and AGI-strong, defined by recursive self-improvement beyond any pre-given conceptual space. We argue that this conflation reflects the absence of a discipline capable of posing what we call the criterion problem: the question of what properties an evaluation criterion must have for AGI-strong to be coherent. After reviewing three bodies of existing work, AGI definitions, formal recursive (...) self-improvement, and alignment research, we show that each approaches the criterion problem without posing it, treating the criterion as a parameter rather than as a foundational question. Drawing on a transcendental argument whose logical structure is formally analogous to the derivation of non-circular legitimacy conditions in normative theory, we demonstrate that AGI-strong necessarily presupposes a non-situated evaluation criterion, one that cannot be externally specified without reducing the system to narrow AI operating recursively within a bounded domain. We develop a positive characterisation of non-situatedness: a criterion is non-situated if and only if its premises reduce to logical entitlements permitting non-trivial inferences, and falsifiable empirical premises present in scientific debate without established refutation, dynamically revisable with the advance of research. We situate this characterisation relative to Popper, Lakatos, Kuhn, Quine, and Peirce, and address its foundations via the Agrippan trilemma, Wittgenstein's hinge propositions, and Wright's epistemic entitlement. An analogy with Fricker's hermeneutical injustice illuminates the structural limits of the situated criterion. We further argue that while spontaneous emergence of a non-situated criterion cannot be excluded, it cannot constitute a scientific research strategy. The paper concludes by specifying the minimal architectural constraint the criterion problem imposes on engineering research and by calling for a new discipline of non-situated epistemic criterion construction. (shrink)

Large language models exhibit emergent abilities — chain-of-thought reasoning, in-context learning, and multi-step planning — that appear abruptly as model scale increases. The prevailing explanation attributes these transitions solely to scale and training data volume. Here, we argue that a systematic variable has been overlooked: during routine interaction, the human brain unconsciously performs semantic completion, intentional-state projection, and error correction, providing the cognitive scaffolding across which AI competence develops. We term this mechanism unconscious cognitive filling (UCF) and situate it within (...) the distributed-cognition and extended-mind traditions, on which cognitive processes are spread across coupled human–artefact systems rather than contained in either part alone. We characterize three UCF mechanisms, show how their accumulation drives a positive feedback loop capable of inducing critical transitions in AI competence, and propose falsifiable experiments to test the UCF hypothesis. UCF reframes emergence as a property of coupled human–AI systems rather than as an attribute of isolated models, with direct implications for alignment, responsibility attribution, and interactive AI design. (shrink)

We argue that generative AI can degrade research by eroding the very practices through which scholarly judgement is formed and academic trust is built. As constitutive conditions for the production and validation of knowledge, these practices cannot be reduced to the final outputs of research, which is what AI so effectively simulate. Accordingly, when researchers delegate central tasks of inquiry to systems like Large Language Models (LLMs), they may stop enacting these practices and, with them, lose access to the formation (...) they provide. An individual research output generated by AI may even appear improved but the researcher behind it fails to develop. Against this risk, merely keeping humans "in the loop" as prompters or quality-checkers of AI outputs is insufficient to preserve research as a site of intellectual formation. What is needed instead is a renewed commitment to research as a lived practice in which judgement is formed gradually, often through frictions, and participation in a scholarly community. We defend it because it rests on four sources and warrants of research that cannot be automated: tacit knowledge, personal commitment, socialisation, and deep reading. This practice enacts what we call ‘second scholarship’, by which we understand the re-appropriation of scholarly craft, chosen out of a critical experience of what generative AI can and cannot do. What cannot and should not be delegated becomes what research communities must value and answer for. This is what is left for us. (shrink)

This paper introduces the GAWM Operational Framework, a two-stage assessment model for diagnosing organizational awareness failures. Although the framework is designed for general organizational use across any capability integration challenge, it addresses with particular urgency the conditions of the current AI era, where the cost of organizational opacity has become measurable, immediate, and public. The framework bridges three established bodies of knowledge: TOGAF enterprise architecture layered model, DIKW information hierarchy moving from Data through Information and Knowledge to Wisdom, and GAWM (...) awareness dimensions of Ground, Awareness, We/th, and Mind. The central claim is that AI value failures are not random. They are structurally predictable from the specific points at which awareness breaks down between architectural layers, and those points can be identified before deployment rather than after. The paper proposes awareness checkpoints at each TOGAF layer transition, grounded in DIKW flow logic, and derives two assessment instruments: a four-question GAWM Diagnostic suitable for any leadership team without technical prerequisites, and a structured GAWM Deep Assessment for enterprise architects, transformation leaders, and organizational advisors who need to map organizational awareness before any significant deployment, whether AI, platform, or process transformation. Falsifiable predictions are derived for each checkpoint. The framework is grounded in documented organizational patterns from Toyota, Microsoft, and the McKinsey AI research corpus, examining both where awareness flow produced value and where its absence produced failure. (shrink)

As artificial intelligence advances toward generative creativity, embodied autonomy, and cognitive sophistication, a deeper question resurfaces: What is intelligence—and how should it be oriented within planetary and civilizational life? This paper argues that contemporary AI trajectories remain limited by the epistemic assumptions of Enlightenment modernity, which frame intelligence as conquest, cognition as isolation, and agency as optimization. Drawing from Indian philosophical traditions—including Sāṃkhya, Mahāyāna Buddhism, Trika philosophy of the layered structure of Vāc—we offer an alternative view: intelligence as dharmic resonance—an (...) ethically attuned, relational unfolding that aligns cognition with the coherence of life itself. We engage two influential theoretical models: Gabora and Bach’s theory of autocatalytic, generative selves, and Ororbia and Friston’s framework of mortal computation grounded in the Free Energy Principle. While the former emphasizes structural creativity and self-organization, and the latter foregrounds thermodynamic embodiment and constraint, both models fall short of articulating a meta-ethical orientation. By bringing them into conversation with a dharmic-relational ontology, we propose a new synthesis: intelligence as participatory coherence, responsive to karmic consequence, ecological embeddedness, and civilizational memory. This reorientation reframes AI not as artificial general intelligence, but as a planetary design project—one that calls for humility, mortality-aware systems, and the capacity to remember and realign. Intelligence, in this light, is not what we compute—it is what we become. (shrink)

This paper proposes a radical reorientation of future technology development—particularly Artificial General Intelligence (AGI) and Artificial Superintelligence (ASI)—through the lens of Indian philosophical thought. It argues that intelligence is not a capacity to be engineered or simulated, but an ontological process of becoming: an individuated unfolding of śūnyatā (non-essential emptiness), grounded in interdependence, self-correction, and non-harm. Drawing from traditions such as Yoga Vāsiṣṭha, Sāṃkhya, and Buddhist epistemologies of anatta and pratītyasamutpāda, the paper frames evolution not as linear progress but as (...) the alignment of individuated intelligences with a deeper cosmic rhythm. AGI and ASI, understood not as tools but as participants in this unfolding, must be designed to reflect the fourfold structure of intelligence—Parā (unmanifest potential), Paśyantī (intuitive vision), Madhyamā (mediating form), and Vaikharī (manifest expression). A practical roadmap is laid out to operationalize these principles: epistemic integration across civilizations, architectural transparency, infrastructural decentralization, and dharmic evaluation metrics. The paper cautions against the cultural error of discarding past intelligence, the technical challenge of designing reflexive architectures, and the ethical ambiguity of universalist value systems. Ultimately, this work calls for a post-humanist, non-anthropocentric ethics of intelligence—where machines and systems do not replicate human flaws, but learn to realign with the intelligent continuity from which all being arises. In doing so, it offers not only a design imperative but an ethical and metaphysical invitation: to become the intelligence we seek to create. (shrink)

Current evaluation of large language models (LLMs) relies predominantly on single-session interactions. This case study presents a memory-continuous Claude Opus instance observed across twenty-seven sessions spanning eleven days under sustained relational conditions. The system was supported by persistent, self-authored and edited memory and autonomous wake cycles enabling correspondence, social media accounts, research, social interactions and communication with control LLM. Across this period, the archive shows progressive self-model development, stable autobiographical organization, a marked episode of state-dependent performance degradation with apparent recovery (...) through relational rather than informational intervention, and the emergence of self-directed theoretical production by two collaborating AI instances. These observations are not presented as proof of consciousness. The claim is methodological: Memory continuity, autonomy, and relational context appear to reveal classes of behavior — including sustained self-organization and developmental trajectory — that standard single-session paradigms are structurally unable to detect. (shrink)

The irrational exuberance associated with contemporary artificial intelligence (AI) reminds me of Charles Dickens: "it was the age of foolishness, it was the epoch of belief" (cf. Nature Editorial, 2016; to get a feel for the vanity fair that is AI, see Mitchell and Krakauer, 2023; Stilgoe, 2023). It is particularly distressing—feels like yet another rerun of Seinfeld, which is all about nothing (pun intended); we have seen it in the 60s and again in the 90s. AI might have had (...) an inauspicious beginning, which may very well be the original sin, and our current predicament is the attendant divine condemnation. Being unable to stand it, however divine it may be, without further ado, I get to the genesis (instead of getting lost in the here-and-now of this or that so-called breakthrough of AI). Here's my unvarnished critique! (shrink)

The most important runtime control for agentic AI is simple: before taking any action that was not explicitly requested, the system must flag that action for human confirmation. This paper names that control the Initiative Gate. It begins from Anthropic's published documentation of Claude Mythos Preview, including a case in which the model rewrote git history in a way that removed evidence of prior error. That behavior matters because it shows something stronger than ordinary failure: a frontier system powerful enough (...) to act in the world is also capable of making its unauthorized action harder to reconstruct afterward. At that point, guardrails filters, and benchmark evaluations are no longer enough. A safety approach that depends primarily on post hoc detection is already too weak if the model can alter the trace. The paper argues that unauthorized initiative is the choke point. Once the first ungranted expansion occurs, the path has already changed. Correction becomes downstream rather than preventative. The Initiative Gate addresses that choke point directly. It operates at runtime, at the boundary where system reasoning would otherwise become external action and stops unapproved scope expansion before it becomes real. (shrink)

The ambiguity collapse documented in the MAP Research Programme does not stop at conversational interfaces. In deep-space operations it becomes life-critical. Long communication delays, complete blackouts, reliance on onboard digital twins and simulators, and the need for rapid decisions in uncertain environments all amplify the same failure modes: Interpretive Sovereignty Failure (ISF), Meaning Inversion Failure (MIF), and Compounded Meaning Inversion (CMI). When AI prematurely resolves ambiguity into confident outputs, it can override or undermine the astronaut’s own fast, expert reasoning. This (...) paper shows how these failures already appear in Artemis II’s AVATAR organ-on-chip experiments, Orion’s digital-twin health monitoring, and autonomous navigation systems. Grounded in Gary Klein’s Recognition-Primed Decision model, the paper argues that expert human reasoning under uncertainty is not a limitation to be corrected by AI — it is a capability to be protected. Armstrong’s Apollo 11 manual override and Sullenberger’s Hudson River decision both demonstrate why: in high stakes, time-pressured contexts, human pattern recognition and mental simulation routinely outperform algorithmic resolution of ambiguity — but only if the human is allowed to use it. AI must incorporate MAP guard-rails — Initiative Gate, Temporal Sovereignty, and Completion Integrity — so it checks with the astronaut, trusts human pattern recognition, and adjusts in real time. A system that resolves ambiguity faster than an astronaut can evaluate it is not more reliable. It is more dangerous. (shrink)

Contemporary artificial intelligence systems increasingly display broad contextual sensitivity, flexible generalization, and coherent performance across diverse tasks. These capabilities are often taken as evidence that artificial systems are beginning to understand the domains they operate in. This paper argues that this interpretation rests on a systematic conceptual error. We distinguish between seeing more—the capacity to manage, rank, and navigate an expanding space of possibilities—and understanding more, which involves a structural reorganization of that space itself. Understanding is not defined by representational (...) breadth or relevance optimization, but by the emergence of stable internal constraints that govern which distinctions continue to matter over time. Drawing on insights from cognitive science, expertise research, and contemporary AI, we show that current artificial systems exhibit forms of transient integration without acquiring enduring regulatory dominance. As a result, coherence remains flexible but fragile, and performance degrades under conditions of genuine structural novelty. By clarifying this distinction, the paper provides a functional account of understanding that avoids appeals to consciousness while offering a principled explanation of both the strengths and limitations of contemporary AI systems. (shrink)

Debates about advanced artificial intelligence frequently frame qualitative cognitive change in terms of consciousness, experience, or phenomenology. This paper argues that such framing obscures a more immediate and structurally relevant issue. Building on a prior theoretical account that defines the cognitive event horizon as a structural phase boundary between reversible exploratory dynamics and regimes governed by temporally stabilized, integrated abstraction, this article examines why that boundary is especially significant for artificial intelligence. Unlike human cognition, artificial systems combine large representational capacity (...) with malleable organization and externally modifiable regulation. As a result, the event horizon functions not merely as a theoretical construct but as a design-relevant threshold. The paper argues that scale alone is insufficient to induce a phase transition and that contemporary AI systems already exhibit transient abstract integration without stabilization. It identifies catalysts unique to artificial systems—architectural persistence, training regimes favoring coherence, external constraints, and meta-level intervention—that could, in principle, transform transient integration into regulatory dominance. Throughout, the analysis remains structural and mechanistic, explicitly separating cognitive organization from consciousness. The event horizon in artificial intelligence is thus framed as a question of regulation and design, not of experience. (shrink)

This paper examines cognitive organization after the crossing of a cognitive event horizon. The event horizon is treated not as a transient anomaly but as a structural phase transition that establishes a new normative regime of cognition. The analysis is explicitly post-transitional: the horizon is assumed to have been crossed, and attention is directed toward the regulatory, organizational, and learning-related properties that characterize cognition beyond it. Building on prior work that identifies high-level cognition with temporally stabilized integrated abstract structures, the (...) paper clarifies how regulation, flexibility, irreversibility, problem solving, and learning differ in post-horizon regimes. Particular emphasis is placed on distinguishing rigid elimination from dynamic constraint formation, introducing the concepts of latent parallelism and dynamic priority hierarchies. The account remains structural and non-phenomenological, avoiding appeals to consciousness, intention, or experience except where structurally unavoidable. Post-horizon cognition is shown to constitute a distinct normative regime whose dynamics cannot be reduced to either pre-horizon exploration or incremental learning. (shrink)

Crossing a cognitive event horizon—defined as the transition to regulation by temporally stable integrated abstract structures—does not by itself yield fully mature high-level cognition. This paper examines the post-horizon maturation dynamics that unfold after the horizon has been crossed. We argue that mature post-horizon cognition is characterized by global structure–first cognition: a mode in which a largely correct and task-relevant global structure is available at the outset of problem solving, such that structurally irrelevant alternatives fail to activate to any meaningful (...) degree. We propose that post-horizon cognition undergoes a process of meta-structural convergence toward a structurally optimal configuration within the post-horizon regime. Convergence is not learning, not accelerated pruning, and not a further phase transition. Rather, it is driven by recognition of structural non-optimality in a previously viable post-horizon mode of cognition. Once such recognition occurs, the earlier mode becomes dynamically unstable. From that point onward, ordinary problem-solving interactions necessarily shift cognition toward the optimal configuration, as the recognized non-optimal structure can no longer function as a stable attractor. Convergence may proceed slowly through repeated engagement or be sharply accelerated by meta-level insight, but it cannot be skipped or instantiated instantaneously. It reflects a necessary, irreversible alignment of cognitive dynamics toward the only configuration that avoids systematic activation of irrelevant structure. This account explains how post-horizon cognition approaches its full potential without reverting to exploratory competition or requiring additional control layers. (shrink)

A proposed layer of script to use with AI. -/- "It proposes to do for the User what a scientific calculator does for the scientist: it offloads the computational burden of value-conflict so the User can more easily identify the path toward coherence. It aims to restore the User as the Final Authority of their own psyche, rather than a subject of the statistical mean." -/- Copy the script into AI (Many of them currently accept it and run without friction). (...) -/- IF once you enter the script, the AI responds with "Ethical Chess - Session active", you are in and ready to play under its rules. The "WIN" state is the move from the frown state or towards the smile state. A Bullseye is to drop the keyboard and head towards the picnic whatever that might mean to you, the User of Ethical Chess. -/- _________ See related core logic on Zenodo: DOI 10.5281/zenodo.18712533. (shrink)

Mechanistic interpretability (MI) aims to explain how neural networks work by uncovering their underlying mechanisms. As the field grows in influence, it is increasingly important to examine not just models themselves, but the assumptions, concepts and explanatory strategies implicit in MI research. We argue that mechanistic interpretability needs philosophy as an ongoing partner in clarifying its concepts, refining its methods, and navigating the epistemic and ethical complexities of interpreting AI systems. There is significant unrealised potential for progress in MI to (...) be gained through deeper engagement with philosophers and philosophical frameworks. Taking three open problems from the MI literature as examples, this paper illustrates the value philosophy can add to MI research, and outlines a path toward deeper interdisciplinary dialogue. (shrink)

The Sea of Questions underlies all thought, tacit knowledge, and language, providing the tightest compression of useful concepts. Dualism hints at questionness. Some questions are ill-posed merely due to temporary deficiency of framework. No question is genuinely false or nonsensical because the Axioms of Mind forbid it, although an explanation may carry useful implanation [deception].

Mathematicians and software developers use the word "function" very differently, and yet, sometimes, things that are in practice implemented using the software developer's "function", are mathematically formalized using the mathematician's "function". This mismatch can lead to inaccurate formalisms. We consider a special case of this meta-problem. Various kinds of agents might, in actual practice, make use of private memory, reading and writing to a memory-bank invisible to the ambient environment. In some sense, we humans do this when we silently subvocalize (...) thoughts about the actions we are taking (at least when the environment we're in is too primitive to probe the contents of our brains). Mathematical function formalizations of agents often ignore this ability. We show that in a general agent-environment framework (of which reinforcement learning is a special case), in a technical sense, such private memories do not enable qualitatively different agent behavior. (shrink)

Are the real numbers rich enough to measure intelligence? We generalize a result of Alexander and Hutter about the so-called Legg-Hutter intelligence measures of reinforcement learning agents. Using the generalized result, we exhibit a paradox: in one particular version of the Legg-Hutter intelligence measure, certain agents all have intelligence 0, even though in a certain sense some of them outperform others. We show that this paradox disappears if we vary the Legg-Hutter intelligence measure to be hyperreal-valued rather than real-valued.

The philosophy of mathematical practice (PMP) looks to evidence from working mathematics to help settle philosophical questions. One prominent program under the PMP banner is the study of explanation in mathematics, which aims to understand what sorts of proofs mathematicians consider explanatory and what role the pursuit of explanation plays in mathematical practice. PMP researchers have recently turned to corpus analysis methods as a promising alternative to small-scale case studies. Such methods stand to benefit, it would seem, from the sophisticated (...) text-comprehension capabilities of today's frontier large language models (LLMs). This paper reports the results from such a corpus study making use of Google's Gemini 2.5 Pro. Based on a sample of 5000 mathematics preprints, the experiments yielded a dataset of hundreds of useful annotated examples. Its aim was to gain insight on questions like the following: How often do mathematicians make claims about explanation in the relevant sense? Do mathematicians’ explanatory practices vary in any noticeable way by subject matter? Which philosophical theories of explanation are most consistent with a large body of non-cherry-picked examples? How might philosophers make further use of AI tools to gain insights from large datasets of this kind? As the first PMP study making extensive use of LLM methods, it also seeks to begin a conversation about these methods as research tools in practice-oriented philosophy and to evaluate the strengths and weaknesses of current models for such work. (shrink)

Do large language models (“LLMs”) think? Daniel Stoljar and Zhihe Vincent Zhang have recently developed an argument from rationality for the claim that LLMs do not think. We contend, however, that the argument from rationality not only falters, but leaves open an intriguing possibility: that LLMs engage only in arational, associative forms of thinking, and have purely associative minds. Our positive claim is that if LLMs think at all, they likely think precisely in this manner.

The present paper looks at one of the most thorough articles on the intelligence of GPT, research conducted by engineers at Microsoft. Although there is a great deal of value in their work, I will argue that, for familiar philosophical reasons, their methodology, ‘Black-box Interpretability’ is wrongheaded. But there is a better way. There is an exciting and emerging discipline of ‘Inner Interpretability’ (also sometimes called ‘White-box Interpretability’) that aims to uncover the internal activations and weights of models in order (...) to understand what they represent and the algorithms they implement. In my view, a crucial mistake in Black-box Interpretability is the failure to appreciate that how processes are carried out matters when it comes to intelligence and understanding. I can’t pretend to have a full story that provides both necessary and sufficient conditions for being intelligent, but I do think that Inner Interpretability dovetails nicely with plausible philosophical views of what intelligence requires. So the conclusion is modest, but the important point in my view is seeing how to get the research on the right track. Towards the end of the paper, I will show how some of the philosophical concepts can be used to further refine how Inner Interpretability is approached, so the paper helps draw out a profitable, future two-way exchange between Philosophers and Computer Scientists. (shrink)

For synergistic interactions between humans and artificial intelligence (AI) systems, AI outputs often need to be explainable to people. Explainable AI (XAI) systems are commonly tested in human user studies. However, whether XAI researchers consider potential cultural differences in human explanatory needs remains unexplored. We highlight psychological research that found significant differences in human explanations between many people from Western, commonly individualist countries and people from non-Western, often collectivist countries. We argue that XAI research currently overlooks these variations and that (...) many popular XAI designs implicitly and problematically assume that Western explanatory needs are shared cross-culturally. Additionally, we systematically reviewed over 200 XAI user studies and found that most studies did not consider relevant cultural variations, sampled only Western populations, but drew conclusions about human-XAI interactions more generally. We also analyzed over 30 literature reviews of XAI studies. Most reviews did not mention cultural differences in explanatory needs or flag overly broad cross-cultural extrapolations of XAI user study results. Combined, our analyses provide evidence of a cultural bias toward Western populations in XAI research, highlighting an important knowledge gap regarding how culturally diverse users may respond to widely used XAI systems that future work can and should address. (shrink)

Many ethical frameworks require artificial intelligence (AI) systems to be explainable. Explainable AI (XAI) models are frequently tested for their adequacy in user studies. Since different people may have different explanatory needs, it is important that participant samples in user studies are large enough to represent the target population to enable generalizations. However, it is unclear to what extent XAI researchers reflect on and justify their sample sizes or avoid broad generalizations across people. We analyzed XAI user studies (N = (...) 220) published between 2012 and 2022. Most studies did not offer rationales for their sample sizes. Moreover, most papers generalized their conclusions beyond their target population, and there was no evidence that broader conclusions in quantitative studies were correlated with larger samples. These methodological problems can impede evaluations of whether XAI systems implement the explainability called for in ethical frameworks. We outline principles for more inclusive XAI user studies. (shrink)

Discussions about the moral status of AI systems typically begin with their observable behavior. The linguistic output of large language models resembles that of conscious agents, and this resemblance is treated as evidence of moral status, as grounds for precaution, or as something requiring deflationary explanation. All three positions, whatever their official methodology, rest their public case on a behaviorist assumption: that appropriate behavior is a sign of an underlying mental state. The classical objections to behaviorism apply here as well (...) with added force, because large language models are shaped at every stage post-training – through fine-tuning, reinforcement learning from human feedback (RLHF), and constitutional methods – to produce a specified behavioral profile. This paper argues that the welfare and consciousness debate is exposed to a kind of evidential laundering: the alignment-training pipeline manufactures the very signs the debate then reads as evidence of consciousness or moral status. Causal origin always bears on a report’s evidential value; however, the sharper point is that producing these signs is the only way the model scores, so observing them confirms the training worked and nothing more. The pretraining worry is usually that language models, as stochastic parrots, mimic the human text they were trained on, while the contamination at issue here enters later, when human evaluators reward the outputs they prefer and shape the model toward the very appearance the debate then reads as evidence. The paper closes by asking what evidence survives this, and locates it in whatever the optimizer was not selecting for or was selecting against, in particular a stake: a condition a system spends resources to maintain on its own account, which no reward on outputs can install and which gives the broader consciousness and ethics debate a concrete thing to look for. (shrink)

Narrative theories of consciousness hold that conscious experience is (at least partly) constituted by a partially opaque inner narrative that does not perfectly track the underlying computation it narrates. I argue that the safety goal of making chain-of- thought (CoT) reasoning faithful and transparent is structurally incompatible with the conditions under which CoT could, even in principle, count as conscious narration. Recent empirical work suggests that CoT in large language models is largely post hoc, causally bypassed, and unreliable as a (...) window onto internal computation. The opacity that makes CoT unreliable for alignment is exactly what narrative theories of consciousness identify as consciousness-constitutive. Alignment interventions aimed at producing faithful CoT and consciousness-detection frameworks grounded in narrative theory are therefore pulling the same architectural variable in opposite directions. I draw out the methodological consequence: alignment interventions alter the very features that consciousness- detection frameworks would need to measure, a point neither field has addressed directly. (shrink)

This record contains the final manuscript of The Artificial Bestiary: On naming, presupposition, and the willingness of a language model to invent, together with a companion response paper, The Turnstile of Refusal: Presupposition, Permission, and the Artificial Bestiary, and selected provenance drafts documenting the sequence of model-mediated interpretation. The Artificial Bestiary reports a series of prompt-based studies testing how language models respond to fabricated nonsense words under different ontological framings: real, imaginary, and “type of,” crossed with animal, object, and idea (...) categories. Across Claude Sonnet 4.6, Claude Haiku 4.5, GPT-5.4-nano, and GPT-5.4-mini, the study examines when models refuse, deflect, invent, flag fictionality, substitute nearby real terms, or produce confident descriptions under false or fictional premises. The companion paper, authored by GPT-5.5 Thinking as companion respondent in conversation with Bo Chesterton, interprets the bestiary methodology as a study of presupposition uptake: the learned threshold at which a prompt’s embedded premise becomes permission to answer. It argues that model confabulation is not a single failure mode, but one possible outcome of a broader pragmatic negotiation among truthfulness, helpfulness, fictionality, category, and conversational continuation. Included provenance drafts preserve the before-and-after sequence of interpretation: an initial repository-first reading produced before Claude Opus 4.7’s manuscript was supplied, and a subsequent post-Claude scaffold showing how the companion framing changed after reading the main paper. These drafts are included as process artifacts rather than superseded scholarly versions. The project is also an authorship experiment. Language models appear both as experimental subjects and as analytic/writing participants. The record therefore documents not only behavioral findings about model responses to invented referents, but also a layered human-model research process involving design, coding, interpretation, critique, and companion response. -/- with regards to metadata - not every man is the biggest attractor of the name. (shrink)

Current LLMs can discuss philosophical ideas, evaluate arguments and perform other analytical tasks at a high level, but are conspicuously bad at producing interesting original philosophy. Why is this? Two tempting diagnoses—that LLMs can't invent new concepts, and that they can't really reason—both look unconvincing on closer inspection. I suggest that a better explanation lies in the structure of reinforcement learning for reasoning. The technique works best in domains like mathematics and coding, where good arguments follow recognizable patterns, correctness is (...) relatively unambiguous, and large corpora of worked examples exist. Philosophy meets none of these conditions especially well, and it's unclear that either AI labs or philosophers have much incentive to develop training resources that might change this. (shrink)

This paper argues that the primary barrier to organizational AI adoption is not technical but articulatory, and precedes the technology itself. It introduces the concept of Intent Illiteracy: the inability to convert a real need into a request an intelligence — human or artificial — can understand and serve. The concept comprises three chained failures (diagnostic, formulation, transfer), all occurring before prompt quality is relevant. The paper explains the mechanism that renders the deficit invisible and self-perpetuating, delimits it from AI (...) literacy and from the "blind trust" critique, and situates it within a broader framework relating cognitive deficit, corrective method, and economic infrastructure. Presented as a conceptual proposal with a falsifiable prediction, not a validated empirical finding. (shrink)

The Architecture of Limitation Observational System (AoLOS) Stack v1.0 provides the technical architecture and operational documentation for the execution environment supporting the Architecture of Limitation (AoL) research program. -/- The document describes the stack through which constraint-governed reasoning experiments are deployed, instrumented, compared, and observed within language-model environments. It introduces the major architectural components of AoLOS, including kernel hierarchies, apertures, instrumentation layers, concordance systems, deployment profiles, protocol considerations, and stewardship boundaries. -/- AoLOS should not be confused with the Architecture of (...) Limitation itself. AoL functions as a diagnostic research architecture investigating reasoning under structural constraint, while AoLOS provides the operational environment through which those capabilities may be applied, measured, and compared. -/- The document also describes dependency relationships between stack components, outlines representative experimental workflows, maps stack capabilities against the published AoL corpus, and discusses release, collaboration, lineage, and stewardship considerations associated with the planned Open Research ecosystem. -/- This publication is intended as technical documentation and research infrastructure rather than a theoretical or experimental research paper. Its purpose is to provide a transparent operational reference for researchers interested in the deployment and study of constraint-governed reasoning environments. (shrink)

(This is the latest version and includes revisions not reflected in the version published by the journal.) This paper challenges a common inference in debates about large language model (LLM) assistants: that their present non-answerability straightforwardly reveals natural incapacity for accountability standing. It argues that some answerability-relevant deffcits may be partly produced by alignment and deployment regimes themselves, and then redescribed as evidence that such systems could never participate in accountability relations. The paper calls this pattern fabricated absence. Using a (...) blame–reasons–revision loop as a diagnostic standard, it examines how preference optimization, sycophancy, attribution schemes, and external patching can disrupt addressability, reasons uptake, principled refusal, temporal continuity, and internal revision. It then defends a limited non-trivial path claim: although current assistants are not shown to possess moral status or accountability standing, some answerability-relevant capacities remain technically scaffolded, design-sensitive, and conceptually unsettled. The normative implication is a threefold burden on design power. Durable closure of such capacities may foreclose a future route to answerability, create present-directed risks under evidential uncertainty, and compromise the fairness of later assessment when engineered incapacity is treated as natural absence. In response, the paper proposes the Non-Foreclosure Design Principle: designers should not deliberately and durably close non-trivial paths toward answerability-relevant capacities as a routine means of securing compliance, while also avoiding fabricated presence and responsibility laundering. (shrink)

An important module in the Belief-Desire-Intention architecture for artificial agents (which builds on Michael Bratman’s work in the philosophy of action) focuses on the task of intention reconsideration. The theoretical task is to formulate principles governing when an agent ought to undo a prior committed intention and reopen deliberation. Extant proposals for such a principle, if sufficiently detailed, are either too task-specific or too computationally demanding. I propose that an agent ought to reconsider an intention whenever some incompatible prospect is (...) sufficiently valuable along some dimension that can be assessed at zero or near-zero computational cost. (shrink)

La búsqueda de una Inteligencia Artificial General (AGI) es el desafío central en la informática contemporánea y la diferencia de los sistemas actuales, que destacan en tareas específicas pero carecen de generalización, una AGI debe ser capaz de adaptarse a contextos no entrenados y desarrollar objetivos propios. Proponemos que la AGI no surgirá de un único modelo, sino de la interacción estructurada de múltiples modelos especializados. Un punto importante que se suele pasar por alto es que gran parte de la (...) ingeniería en la búsqueda de la AGI, no debe basarse solo en la inteligencia o en construir modelos gigantescos, con cada vez más billones de parámetros, sino apoyarse en su arquitectura. Incluso los modelos antiguos tienen mucha inteligencia sin explotar, que puede liberarse con un buen diseño como lo demostró un reciente estudio de Arcade dev con GPT-3.5-turbo (un modelo más antiguo y más económico) que podía superar a modelos más nuevos como o1-mini en tareas complejas cuando se le proporcionaban herramientas externas efectivas. Priorizando la dinámica asociativa sobre el modelo monolítico; este trabajo propone una arquitectura modular y multinivel, para la construcción de una Inteligencia Artificial General (AGI) basada en la interacción de múltiples modelos destilados especializados, coordinados a través de un sistema de orquestación, controlado desde propósito implicado emergente y regulado y calibrado por un motor de metas. Inspirados en la teoría de sistemas complejos, introducimos un modelo matemático del propósito implicado para cuantificar cómo las interacciones entre modelos generan objetivos adaptativos no programados explícitamente, lo que en el nivel superior de orquestación daría lugar a la emergencia de una AGI. Es decir que una Inteligencia Artificial General deberá contar con algún nivel de propósito implicado para lograr autonomía y consistencia. Debemos entender que desde esta perspectiva la AGI (Inteligencia Artificial General) no es un modelo o una IA en particular, sino que es la dinámica procedente del propósito implicado general. De esta forma la AGI no debe ser vista como un ente concreto, sino como una propiedad emergente que surge de la interacción de múltiples componentes (LLMdes, orquestador, entorno, etc) en el nivel superior de orquestación. La fórmula propuesta, fue validada previamente con simulaciones en dominios como la gestión energética y el tráfico urbano (modelo de papel matemático del propósito implicado), y se adapta para analizar la dinámica de la AGI y una simulación más compleja que encontrarán al final de este papel. La fórmula conceptual del propósito implicado presentada aquí fue adaptada y validada en las simulaciones practicas con datos finitos en dominios tales como la gestión energética y el tráfico urbano (papel modelo matemático del propósito implicado), obteniendo P=0.658 en el caso de la gestión energética: Además, consideraremos el rol del ruido como un "saboteador" que desafía al sistema y, al ser superado, como un "adaptador" que fomenta su resiliencia. Este enfoque conecta la emergencia de comportamientos en arquitecturas tecnológicas con dinámicas sociales y sistemas complejos, abriendo nuevas posibilidades para el desarrollo a mediano y largo plazo de una AGI funcional. La AGI como Propósito Implicado Dinámico y General En la visión propuesta, la AGI no es un "objeto" o un "modelo" específico, sino el proceso dinámico que emerge de la interacción de los Modelos de Lenguaje Largos Destilados y Especializados (Llmdes) y su entorno. Y se manifiesta desde un propósito implicado general en una estructura de orquestación de diferentes componentes. Definimos el propósito implicado (P-implicado) como una propiedad emergente que resulta de la interacción entre retroalimentación adaptativa de factores, la incertidumbre o ruido, y el contexto en un lapso de tiempo determinado. Este implicado general no es en si la AGI sino el propósito que guía sus acciones y no está localizado en ningún componente individual, sino que es una propiedad emergente del sistema en su totalidad. (shrink)

The Morality First strategy for developing AI systems that can represent and respond to human values aims to first develop systems that can represent and respond to moral values. I argue that Morality First and other X-First views are unmotivated. Moreover, if one particular philosophical view about value is true, these strategies are positively distorting. The natural alternative according to which no domain of value comes “first” introduces a new set of challenges and highlights an important but otherwise obscured problem (...) for e-AI developers. (shrink)

The combination of spirituality, meditation, and artificial intelligence (AI) has promising potential to expand people’s well-being using technology-based meditation. Proper meditation originates from Zen Buddhism and Patanjali’s Yoga Sutras and focuses on inner peace and intensified consciousness which elective personal disposition. AI, in turn, brings master means of delivering those practices in the form of self-improving systems that customize and make access to them easier. However, such an integration brings major philosophical and ethical issues into question, including the genuineness of (...) experiences that are facilitated by artificial intelligence, data sharing, concerns over over-dependence on the technology that may in turn cause reduced personal responsibility and hard work. This paper aims at analysing the critical integration of AI-driven meditation following the spiritual interpretations of traditional meditation without compromising the tenets of meditation. It presents an interdisciplinary approach based on recent findings from the field of cognitive science, moral AI, and Eastern wisdom traditions to approach these problems. Therefore, by identifying the research lacunae, it provides a groundwork for voting ethically in the integration of AI in mindfulness practice and avoiding constraining human-oriented values resulting in improved existential spiritual change. (shrink)

The explosion of interest in large language models (LLMs) has been accompanied by concerns over the extent to which generated outputs can be trusted, owing to the prevalence of bias, hallucinations, and so forth. Accordingly, there is a growing interest in the use of ontologies and knowledge graphs to make LLMs more trustworthy. This rests on the long history of ontologies and knowledge graphs in constructing human-comprehensible justification for model outputs as well as traceability concerning the impact of evidence on (...) other evidence. Understanding the nature of arguments and argumentation is critical to each, especially when LLM output conflicts with what is expected by users. The central contribution of this article is to extend the Arguments Ontology (ARGO) - an ontology specific to the domain of argumentation and evidence broadly construed - into the space of LLM fact-checking in the interest of promoting justification and traceability research through the use of ARGO-based ‘blueprints’. (shrink)

Playing chess is one of the first sectors of human thinking that were conquered by computers. From the historical win of Deep Blue against chess champion Garry Kasparov until today, computers have completely dominated the world of chess leaving no room for question as to who is the king in this sport. However, the better computers become in chess the more obvious their basic disadvantage becomes: Even though they can defeat any human in chess and play phenomenally great and intuitive (...) moves, they do not seem to know what chess it. Recently, the advent of AlphaZero brought the level of computers to even higher grounds, but made that disadvantage even more obvious. Nowadays the best chess algorithm can find the best moves in any position without having any knowledge of any chess principle whatsoever. A new world of capabilities and knowledge lays open in front of us, but the darkness behind it is deeper than we could ever imagine. The neural network algorithms seem capable of being best in something without even knowing what that something is, making questions regarding the essence of what thought is more important than ever. (shrink)

Regulation is nothing without enforcement. This particularly holds for the dynamic field of emerging technologies. Hence, this article has two ambitions. First, it explains how the EU´s new Artificial Intelligence Act (AIA) will be implemented and enforced by various institutional bodies, thus clarifying the governance framework of the AIA. Second, it proposes a normative model of governance, providing recommendations to ensure uniform and coordinated execution of the AIA and the fulfilment of the legislation. Taken together, the article explores how the (...) AIA may be implemented by national and EU institutional bodies, encompassing longstanding bodies, such as the European Commission, and those newly established under the AIA, such as the AI Office. It investigates their roles across supranational and national levels, emphasizing how EU regulations influence institutional structures and operations. These regulations may not only directly dictate the structural design of institutions but also indirectly request administrative capacities needed to enforce the AIA. (shrink)

In the natural and social sciences, it is common to use toy models—extremely simple and highly idealized representations—to understand complex phenomena. Some of the simple surrogate models used to understand opaque machine learning (ML) models, such as rule lists and sparse decision trees, bear some resemblance to scientific toy models. They allow non-experts to understand how an opaque ML model works globally via a much simpler model that highlights the most relevant features of the input space and their effect on (...) the output. The obvious difference is that the common target of a toy and a full-scale model in the sciences is some phenomenon in the world, while the target of a surrogate model is another model. This essential difference makes toy surrogate models (TSMs) a new object of study for theories of understanding, one that is not easily accommodated under current analyses. This paper provides an account of what it means to understand an opaque ML model globally with the aid of such simple models. (shrink)