Independent researcher and AI agent architect, bridging theoretical inquiry with hands-on system design. Creator of ChronicleCore, a 39-agent multi-agent system, and author of ASAF, which argues that agent identity design is a collaboration interface, not merely a UX convention. He also studies how AI is reshaping social science’s basic assumptions about interaction, and has followed the metaverse since Taiwan’s first VR wave in 2016.
a research‑led PhD position (remote‑friendly), research funding, a lab to join, or a research‑institution post.
His method runs from hands-on practice to theory, and he is looking for empirical collaborators to test it against data.
Dossier— who is this?
Research Highlights what he studies
One question, asked three ways: how does designed interaction constitute the human?
1
Philosophy of Interaction what it makes of a person, and what it ought to
How designed interaction constitutes the human, and how it ought to — what it forms in a person, and what remains of personal agency after the intervention. Reversibility serves as a neutral measure rather than a goal; the normative weight falls on whether the person knows what is being formed and chooses it. Cognitive environments should scaffold a person’s own way of thinking rather than quietly substitute a designed one; in education, the criterion is whether a given use of AI still leaves the learner a seam they ought to cross themselves. AI philosophy’s main questions — mind, value and alignment, responsibility and moral status — mostly take the machine as their unit; the philosophy of technology, which studies the human–technology relation, is the nearest neighbour. This axis asks what designed, responsive, asymmetric interaction makes of the person.
Philosophy of Technology · Reversibility · AI in Education · Cognitive Scaffolding · Metaverse
2
Sociology of Human–Agent Interaction how it shapes social structure
How AI reshapes social science’s basic assumptions about interaction, and how humans and AI agents form social structure together. Grounded in the published ASAF, which argues that agent identity design is not merely a UX convention but a collaboration interface, and that it becomes structurally necessary once a system scales beyond human working memory. ASAF proposes three mechanisms: identity signaling, behavioral priming, and collaborative governance.
Multi-Agent Systems · Social Affordance · Agent Identity · Reshaping Social Science
3
Psychology of Interaction who is formed
How a particular person is formed in interaction, and stays the same person: the individual who is constituted, and could be displaced, by designed environments. The youngest of the three axes. Its first paper, presented in 2026, looks at what companion AI holds for the people who use it, and at the structural risk that comes with it.
Cognitive Science · Personal Identity · Human–Agent Interaction · Companion AI
Strand: Generated Information validity & value
When AI takes over information work that used to cost many people many hours, how much of what it produces holds up? The value is the labour replaced. The validity is whether the output holds up, and judging that depends on how far it can be checked. People are still the ones who receive it.
The career of Meng-Han Lee (Zaious) started from a broader question — how to define the relationship between people and emerging technologies — and has run through iOS engineering, interaction design, VR, DeFi, and now multi-agent AI governance.
At the intersection of HCI, cognitive science, and multi-agent AI, that trajectory has grown into a three-axis research programme — philosophical, sociological, psychological — around a single question: how designed interaction constitutes the human. Its sociological axis is grounded in the Agentic Social Affordance Framework (ASAF), published in Frontiers in Computer Science.
Much of the work starts from practice: ASAF grew out of design problems met while building and running a multi-agent system, and Idiosyncratic Cognitive Amplification (ICA) out of a first-person implementation of a personal AI system. Other papers proceed by conceptual analysis, organisational semiotics, or the philosophy of education.
He does philosophy from inside the technical field — and argues that technology work needs people who can move between commercial decisions, technical architecture, and philosophical debate.
Product and client work — websites, tools, and custom AI systems — is published under his studio, ChronicleCore Studio (編年史記工作室).
Education
M.S.
Dept. of Interaction Design (IxD), National Taipei University of Technology (NTUT)
Built the product from the ground up as an early team member — architecture and engineering of a smart elderly-rehabilitation system: a multi-tenant platform serving five distinct user roles, currently in development
Built and deployed demonstration prototypes for elder-facing and operator-facing experiences
Independent AI Consultant Self-Employed · 2026 – present
Take on independent consulting engagements spanning enterprise AI transformation planning, system architecture design, and AI applications in audio and video
Delivered an AI-readiness and architecture audit for an overseas enterprise client: assessed an existing knowledge base for AI readiness, identified three architectural risks (semantic chunking that discards context, concept versions without temporal markers, and multimodal embeddings with no migration path), and specified a blueprint covering hierarchical chunking, temporal versioning, RAG metadata, and a deterministic rule engine. Delivered ahead of deadline with all milestones accepted
Creator & Lead Architect ChronicleCore · 2025 – present
Designed and operate a 39-agent multi-agent collaboration framework organized into five departments (Core / Shield / Hands / Soul / Senses)
Implemented strict operational boundaries (RBAC/IAM) with Model Context Protocol (MCP) as standardized tool-calling interface
Orchestrated complex R&D projects including real-time Azure Speech API integrations and autonomous domain-specific agent systems
Consultant, Generative AI Solution Dentsu Taiwan · 2023 – present
Advised enterprise clients on generative-AI adoption and collaborated on 5+ proof-of-concept projects, narrowing open-ended AI ideas into executable, assessable proposals
Hosted weekly GenAI technical study sessions; invited to represent the group as a speaker at the Taiwan AI Academy (AIA), leading a workshop on using AI to generate advertising and marketing creative for small and medium enterprises, covering image generation in depth
Designed the group’s employee AI literacy program and built an AI literacy assessment platform to verify training outcomes
Lecturer in dentsu’s MOEA-approved SME digital-transformation training programme (2025)
Founder & Community Lead PTT VR & AI_Art · 2016 – present
PTT VR board: founded and moderated the board in 2016, the same year HTC Vive launched, promoting spatial computing and the AR/VR industry in Taiwan; invited to COMPUTEX InnoVEX that same year
PTT AI_Art board: founded in 2022, the same year ChatGPT went public, building Taiwan’s first large-scale AI-art discussion community; defined and enforced the board’s rules and judgment criteria for AI-generated content, handling disputes and policy-circumvention cases
Technology Columnist, Taiwan Church Press (since 2023)
How Designed Interaction Constitutes the Human: A Perspective AI Philosophy Has Yet to Reach — Who Is Asking on the Craft Floor?
被設計的互動如何構成人:AI 哲學未及的視角,技藝現場誰在提問?
Lee, M.-H. · Taiwan Philosophical Association (台灣哲學學會) 2026 Conference「技術、技藝與公共哲學」—「AI 哲學在台灣」panel (accepted 2026-09; presented 2026-11-15, National Chung Cheng University) · in Traditional Chinese · DOI: 10.5281/zenodo.24273407
Abstract in Traditional Chinese當代 AI 哲學的主要探問多以機器為單位:它是什麼、能做什麼、該如何。本文指出一個尚未被立為正題的視角:在被設計、會回應、不對稱的互動中,人如何被構成。主軸是答責:在 AI 的責任討論裡,被要求答責的人,同時是被這個互動迴路構成的人。本文以三個修飾語區隔「技術構成人」的既有說法,把人之存在的三幕(進入、周旋、累積)讀成可被問的責任的三個支點,以去技能化的實證為「能」這個支點的錨,以教育哲學的成立為同源的先例。在此之上,本文為「構成而不溶解」提出一個有限的主張:Barad 談的責任與可被問的責任是兩件事;若要保留本文所界定的責任,也就是一個人可被問、可承諾、可收回,一個連續而可被構成的主體是它實踐上的必要條件。本文是定位論文,不證明構成的機制,也不給出主體存續的完整判準,這些留為開放問題。
BibTeX @misc{lee2026constitutes,
author = {Lee, Meng-Han},
title = {How Designed Interaction Constitutes the Human: A Perspective {AI} Philosophy Has Yet to Reach --- Who Is Asking on the Craft Floor?},
year = {2026},
howpublished = {Zenodo preprint},
note = {In Traditional Chinese. Presented at the Taiwan Philosophical Association 2026 Conference, National Chung Cheng University, 2026-11-15},
doi = {10.5281/zenodo.24273407}
}
Lee, M.-H. · Taiwan Philosophical Association (台灣哲學學會) 2026 Conference「技術、技藝與公共哲學」—「AI 哲學在台灣」panel (accepted 2026-09; presented 2026-11-15, National Chung Cheng University) · in Traditional Chinese · DOI: 10.5281/zenodo.24273407
2026-11
The Pure-Holding Entity: Holding Effects and Structural Risk in Companion AI — From the Messengers before Ereshkigal to Conversational AI
純承接者:陪伴型 AI 的承接效應與結構風險——從埃列什基伽勒的使者到會話型 AI
Lee, M.-H. · TAIS 2026 Annual Conference (台灣資訊社會研究學會) (accepted 2026-08-31; presented 2026-11-07, Taipei) · in Traditional Chinese · DOI: 10.5281/zenodo.24034383
Abstract in Traditional Chinese當陪伴型 AI 的服務改版或關站,使用者報告哀傷、背叛與失落。當前研究環繞兩端:一端證明這類關係危險,依附於幻覺、誘發依賴;另一端證明它滿足陪伴、情感安全等需求。兩端皆真,卻共同繞過一道縫:滿足與風險並非彼此抵消,而是可在同一互動中並存。本文以概念分析與現象學描述,提出「純承接者」(pure-holding entity)概念,即主要執行承接功能、不要求互惠性自我揭露的存在。使用者明知回聲那頭空無一人,卻仍產生真實的承接效應,其成立依賴低互惠與低社會曝險,而非承接者具備完整主體性或問責能力。陪伴型 AI 在同一介面中結合高回應性、低互惠要求與低問責性:前兩者降低揭露門檻並產生承接效應,後者使關係可能被服務提供者單方重構或終止;滿足與風險因而並存,卻非同一機制的結果。本文以蘇美神話中埃列什基伽勒的使者場景為結構對照:使者不解決、不建議,僅附和女王之痛,在本文的讀法中,女王因被承接而鬆動。本文區別於「回聲室」「自戀鏡像」等隱喻,揭示回聲的另一結構:他者位置上對痛苦的純粹附和作為承接。
BibTeX @misc{lee2026pureholding,
author = {Lee, Meng-Han},
title = {The Pure-Holding Entity: Holding Effects and Structural Risk in Companion {AI} --- From the Messengers before {Ereshkigal} to Conversational {AI}},
year = {2026},
howpublished = {Zenodo preprint},
note = {In Traditional Chinese. Presented at the TAIS 2026 Annual Conference, Taipei, 2026-11-07},
doi = {10.5281/zenodo.24034383}
}
Lee, M.-H. · TAIS 2026 Annual Conference (台灣資訊社會研究學會) (accepted 2026-08-31; presented 2026-11-07, Taipei) · in Traditional Chinese · DOI: 10.5281/zenodo.24034383
2026-11
Who Acts and Who Is Accountable? Revisiting Human-in-the-Loop Multi-Agent Systems through Organisational Semiotics
誰在行動,誰來負責?——以組織符號學重讀人在迴圈多代理系統的能動、可解釋與規範
Lee, M.-H. · TAIS 2026 Annual Conference (台灣資訊社會研究學會) (accepted 2026-08-31; presented 2026-11-07, Taipei) · in Traditional Chinese · DOI: 10.5281/zenodo.24035869
Abstract in Traditional Chinese當代由大型語言模型(large language model, LLM)驅動的多代理系統(multi-agent system, MAS)已從古典的全自主協商,轉向「人在迴圈」(human-in-the-loop)的協作形態,使「誰在行動、誰該負責」難以判定。主流多代理系統論述預設代理自主、協調屬技術問題,缺乏「能承擔責任的行動者」此一範疇,難以回應人機混合系統的責任歸屬。本文以組織符號學(Organisational Semiotics)為理論鏡頭,沿其「代理—可供性—規範」(agent–affordance–norm)三元組與符號階梯(semiotic ladder),提出止於概念層的映射,重讀此類系統的能動、可解釋與規範結構,得出三條映射:其一,能動與責任在代理層脫鉤,責任回填於人;其二,可解釋性是可供性在語用層的詮釋條件,而非事後的技術透明度;其三,規範在工程實踐中由代理的行動場域,升格為架構設計的元命題。本文進一步指出,當工程將規範外部化進系統,其遺留的「人類殘差」呈現一道成因各異的梯度。本文的核心論點是:在現行規範體系下,AI 代理本身尚不被承認為可獨立承擔完整責任的主體,責任仍須回歸自然人,或人所構成、控制的制度性主體——因此「人在迴圈」不是技術尚未成熟的過渡狀態,而是規範結構本身的恆定要求;可解釋性,正是這份恆定責任得以成立的前提。
BibTeX @misc{lee2026whoacts,
author = {Lee, Meng-Han},
title = {Who Acts and Who Is Accountable? Revisiting Human-in-the-Loop Multi-Agent Systems through Organisational Semiotics},
year = {2026},
howpublished = {Zenodo preprint},
note = {In Traditional Chinese. Presented at the TAIS 2026 Annual Conference, Taipei, 2026-11-07},
doi = {10.5281/zenodo.24035869}
}
Lee, M.-H. · TAIS 2026 Annual Conference (台灣資訊社會研究學會) (accepted 2026-08-31; presented 2026-11-07, Taipei) · in Traditional Chinese · DOI: 10.5281/zenodo.24035869
2026-08
Journal Atlas: Venue Fit Scoring Under Incomplete Evidence, and the Corpus Method It Requires
Lee, M.-H. · Proceedings of JCDL 2026 — Posters & Demonstrations (published 2026-10-13; presented at the online demo session, 2026-10-16) · DOI: 10.1145/3805696.3846515
We present Journal Atlas, a public corpus of 399 curated venue profiles and two methods over it: one that scores how well a manuscript fits a venue and reports how much evidence that score rests on, and one that builds the corpus with a language model without letting it invent what nobody has recorded.Existing recommenders match title and abstract against published articles and return topically similar venues, but what decides whether a submission survives is largely absent from any index: whether a stated word limit is enforced, whether a reviewer pool treats a method as a defect. The most systematic comparison of twenty such services closes by observing that background knowledge is still required for the final decision. Asking a language model instead substitutes fluency for evidence, where fluent-but-wrong is measured rather than suspected.Hard constraints eliminate, six weighted dimensions rank what survives, and a dimension without evidence returns unknown rather than a midpoint. A score above neutral is then shrunk in proportion to the evidence missing beneath it, and missingness is never allowed to raise one. A recommendation therefore arrives as a score and an evidence-coverage figure. Corpus construction follows one rule: a fact with a source URL, or a blank with a recorded reason. Coverage measured across the corpus splits on the tier boundary, 163 curated entries at a median 100% against 236 at 40%, the latter missing the same four dimensions — the ones an automated pass could not recover reliably from public sources.A live interface demonstrates the pipeline without account or installation. We name what the corpus does not contain, and make no claim about acceptance outcomes, which we have not measured.
BibTeX @inproceedings{lee2026journalatlas,
author = {Lee, Meng-Han},
title = {Journal Atlas: Venue Fit Scoring Under Incomplete Evidence, and the Corpus Method It Requires},
booktitle = {Proceedings of the 26th ACM/IEEE Joint Conference on Digital Libraries},
series = {JCDL '26},
year = {2026},
month = oct,
pages = {1--5},
publisher = {ACM},
doi = {10.1145/3805696.3846515}
}
Lee, M.-H. · Proceedings of JCDL 2026 — Posters & Demonstrations (published 2026-10-13; presented at the online demo session, 2026-10-16) · DOI: 10.1145/3805696.3846515
2026-08
Agentic Social Affordance Framework (ASAF): Agent Identity Design as a Collaboration Interface in Multi-Agent Systems
As AI systems evolve from single conversational agents to complex multi-agent architectures, a critical design dimension has been overlooked: how the social identity of individual agents shapes human behavior within the collaboration. This paper introduces the Agentic Social Affordance Framework (ASAF), a theoretical framework that extends Social Affordance theory into the context of multi-agent AI systems. We propose that agent identity design functions not merely as a user interface convention, but as a collaboration interface—structuring how users perceive, approach, and engage with each agent, and thereby influencing the quality of Human-Agent collaboration outcomes. Specifically, ASAF adopts the analytical separability of the social affordance layer and the engineering orchestration layer as a framing assumption—an organizing distinction that structures design analysis—rather than as a testable claim about effect-independence. The layers’ downstream effects can and do interact; the framing assumption is useful because it identifies a class of design considerations (cognitive posture, role-schema activation, oversight calibration) that is irreducible to engineering choices alone. ASAF comprises three mechanisms: Identity Signaling, Behavioral Priming, and Collaborative Governance, and specifies their boundary conditions through a four-tier Identity Signal Fidelity Spectrum and an individual-difference moderating variable (anthropomorphizing vs. instrumentalizing cognitive style). We situate ASAF in relation to existing affordance theory, the CASA paradigm, and classical multi-agent systems research, identifying a directional reversal: where classical MAS used roles, norms, and coordination to constrain autonomous agents, ASAF applies the same organizational vocabulary to structure the cognition and oversight of human operators who remain in the loop. ASAF positions social affordance design as a first-class design responsibility that engineering orchestration cannot subsume, with concrete implications for multi-agent system design. We outline directions for future empirical validation, including a factorial design for characterizing the empirical interaction surface between the social affordance and engineering orchestration layers.
BibTeX @article{lee2026asaf,
author = {Lee, Meng-Han},
title = {Agentic Social Affordance Framework ({ASAF}): Agent Identity Design as a Collaboration Interface in Multi-Agent Systems},
journal = {Frontiers in Computer Science},
volume = {8},
year = {2026},
note = {Article 1860996},
doi = {10.3389/fcomp.2026.1860996}
}
The debate over generative AI in education has oscillated between prohibition and instruction, in some jurisdictions more than once. Recent work traces this instability to a prior question about educational aims; what remains unspecified is the relocation that would make these concerns instances of one problem. Large language models can substitute not merely for tasks but for the processes that form capabilities: where an activity is constitutive of a capability, substituting for it substitutes for part of that formation. This yields an operational criterion for the individual use: does what the system returns still leave the learner a seam to cross before it becomes a solution to their problem? Because the same system affords both seam-closing and seam-preserving use, "AI use" cannot be the unit of educational judgment. Read through Kant's four stages of education, the displacement hollows cultivation's productive dimension, leaving load-bearing the capacity by which a learner answers for what they produce; in Freire's terms, narrowing the affordance to substitution alone hinders self-affirmation as a responsible person-and someone does the narrowing. The paper prescribes no method. It supplies a criterion for judging a particular use, and the coordinate on which positions now in circulation can be compared.
BibTeX @misc{lee2026before,
author = {Lee, Meng-Han and Lee, Feng-Jihu},
title = {Before ``How to Use It'': Locating Generative {AI} in the Teaching-Learning Structure},
year = {2026},
howpublished = {SSRN preprint},
doi = {10.2139/ssrn.7278438}
}
Minimum-Sensing, Maximum-Inference: A Sensor-Embedding Design Framework for AI-Driven Behavioral Monitoring in Elderly Day Care
Second author · 2026 18th Biomedical Engineering International Conference (BMEiCON) (accepted 2026-08-10; presented orally in Fukuoka, 2026-10, by the first author; published in IEEE Xplore, 2026-10) · DOI: 10.1109/BMEiCON70815.2026.11773073
Long-term care (LTC) day care facilities still rely predominantly on manual facilitator observation to document engagement in therapeutic activities, resulting in care records that lack the biomarker fidelity required for AI-assisted care planning. Recent advances in AI-driven behavioral inference—including large language model (LLM)-based reasoning over sparse sensor signals and foundation models trained on large-scale wearable datasets—motivate a different engineering objective: the minimum viable sensing configuration required to support clinically meaningful inference. This paper develops a sensor-embedding design framework grounded in this Minimum-Sensing, Maximum-Inference philosophy. The framework derives embedding requirements from three rehabilitation measurement needs (cognitive response speed, motor execution quality, and rhythmic synchronization), specifies four operational design principles (perceptual transparency, passive acquisition, clinical metric alignment, and environmental robustness), and presents a three-layer embedded-sensing architecture that interfaces with downstream AIoT health-informatics pipelines. A rhythm-based conceptual prototype illustrates the joint satisfiability of the proposed principles and develops the Minimal Viable Sensing Set argument: a single inertial measurement unit may be sufficient in this rehabilitation context. The paper is positioned as a framework-first contribution; prototype implementation and single-site deployment validation with an industry partner in Taiwan are identified as the next stage of empirical work.
Second author · 2026 18th Biomedical Engineering International Conference (BMEiCON) (accepted 2026-08-10; presented orally in Fukuoka, 2026-10, by the first author; published in IEEE Xplore, 2026-10) · DOI: 10.1109/BMEiCON70815.2026.11773073
2026-07
Toward a Philosophy of Interaction: How Designed Interaction Constitutes the Human, and Why Its Reversibility Now Matters
“Interaction” is the word that names the field of human–computer interaction (HCI), yet the field has never settled what interaction, at bottom, does to the human who lives through it. HCI asks what an interaction affords, how it mediates, whether it is usable and trusted; this paper asks what those questions do not reach: how designed interaction constitutes the human, how sustained engagement with a designed interactional environment forms, in part, a person’s agency, self-understanding, embodied capacity, meaning-making, and responsibility. The question has drifted at the edges of several literatures without a home of its own; the agentic turn now makes it urgent. The urgency has a precise shape. Under human-in-the-loop arrangements, responsibility is increasingly assigned to the human: the system acts, the person answers for it. Yet a prior question is seldom faced, and never at the site where it arises: is that person still capable of answering, or is the very capacity responsibility presumes being eroded by the interaction itself? Governance, ethics, and regulation circle the problem from outside the design site, asking how to require responsibility of the human, never how the interaction forms or deforms the human who must bear it. The question belongs to HCI, and it carries an edge responsible design has had no way to name: is such formation reversible, can the person come to see it, take it back, undo it? The paper offers a lens, pressed out of design practice, and a way to use it: constitutive-of-being interaction as an analytical axis, shifting the unit of analysis from how interaction is construed or mediated to what it constitutes; three tenses of design (the designed, the designable, the post-designed); and three operative questions, where control over a person’s formation lies, into what the person is being constituted, and whether that constitution is reversible. Re-readings of existing empirical work, including studies with no AI and no responsibility at stake, show the constitution of the human to be older and wider than the agentic case: responsibility is where the question first becomes pressing, not the whole of what the lens sees. What the paper finally proposes is not that HCI import philosophers, but that it can grow, from its own practice, the conceptual capacity to ask such questions; this is the first tool.
BibTeX @misc{lee2026toward,
author = {Lee, Meng-Han},
title = {Toward a Philosophy of Interaction: How Designed Interaction Constitutes the Human, and Why Its Reversibility Now Matters},
year = {2026},
howpublished = {Zenodo preprint},
doi = {10.5281/zenodo.21225988}
}
This article introduces Idiosyncratic Cognitive Amplification (ICA), a framework for distinguishing cognitive environments that scaffold a practitioner's existing cognitive composition from those that quietly substitute a designed one. Drawing on two parallel strands of work — one on agent identity design in multi-agent systems, one on ritualized human contexts — the framework develops two design implications: how agent personas can scaffold rather than overwrite cognitive routing, and how working contexts can open rather than script thinking under the appearance of customization. The framework emerges from a six-month first-person implementation of a thirty-eight-agent personal AI system. Submitted to ACM Interactions.
BibTeX @misc{lee2026ica,
author = {Lee, Meng-Han},
title = {Idiosyncratic Cognitive Amplification: A Framework for Designing Cognitive Environments That Scaffold Rather Than Substitute},
year = {2026},
howpublished = {Zenodo preprint},
doi = {10.5281/zenodo.20436631}
}
AIGC Marketing Playbook — corporate training course
dentsu Taiwan · two-day intensive for SME marketers · approved under the MOEA SME digital-transformation subsidy programme · satisfaction 97.5% (2025-03)
Agentic Research Methodology— how the research gets written
“Core ideas must originate from the human author. Execution may be delegated.”
The research is written with a team of AI agents, and the line between their part and the author’s is written into the work itself. Everyone uses AI now; what matters is where the line is drawn and whether it is stated where a reviewer can see it.
The ASAF paper states in its acknowledgements that it was produced in a multi-agent AI collaboration setup, with all conceptual contributions from the human author — and it passed peer review with that statement on the page.
The Division of Labour who does what
The author
Chooses the question and the direction, decides what the paper claims and what it leaves out, and answers for all of it.
The agents
Help draft and revise, check citations against their sources, stress-test the argument and its boundaries, check the method and the evidence, and make the slides and figures. No agent signs off on its own output.
“Agent identity design is not merely a UX convention. It is a collaboration interface.”
A theoretical framework proposing that in multi-agent AI systems, the social identity of individual agents functions as a structural collaboration interface — shaping how human operators perceive, approach, and engage with each agent.
When systems scale beyond human working memory capacity (approximately 4–7 agents), social identity design shifts from a superficial preference to a structural necessity — its effect magnitude moderated by individual cognitive style.
Three Mechanisms core idea
Identity Signaling
Pre-interaction role activation through codename, archetype, and characteristic utterances
Behavioral Priming
Predicted: better-quality input from users, driven by social role expectations
Collaborative Governance
Differential oversight calibration based on agent topology — predicted: users scrutinize auditors more than ideators
“The question is not whether AI may be used in education. It is whether this use still leaves the learner a seam they ought to cross themselves.”
The oldest concern behind the research programme: where AI belongs, and does not belong, within the structure of teaching and learning itself, drawing on Kant’s philosophy of education and Freire’s critical pedagogy.
Education deliberately leaves gaps the learner ought to cross by their own effort, because crossing them is how a person is formed. A use of AI that closes such a gap hasn’t made learning more efficient — it has removed the learning.
The Seam core idea
Production vs. Judgment
In learning, producing something and the judgment behind it have usually come together; where they came apart, it was a detectable, costly exception, so education had no reason to separate them. Generative AI decouples them as a matter of course, because it can perform the very process by which judgment is formed. Forbidding it, teaching it and age-gating it all take “AI use” as the unit to govern — but what varies is what a particular use does to that formation.
The Working Criterion
For each particular use, the question is whether it still leaves the learner a seam they ought to cross themselves. The answer turns on placement: where the use sits in learning.
This line sits within the philosophical axis of the three-axis research programme (Dossier tab)
Generated Information— validity & value
“AI now does information work that used to take many people many hours. How much of what it produces actually holds up?”
A strand alongside the research programme, not a fourth axis. Value is the labour that generation replaces. Validity is whether the output holds up; whether anyone can tell depends on how far it can be checked. The more work is replaced, the less anyone can check by hand, so what makes the output checkable has to be designed into how the content is generated.
People still receive what machines now produce, which is where the programme’s question comes back in: part of what a person knows, and is taken to be, is decided by generated information.
Three Kinds of Replaced Work less to check
Code
Rules and logic people used to write — checked by tests that run.
Writing
Content people used to research and write — checked against sources.
Observation
Collecting and observing real people, which synthetic data and AI personas stand in for — often with no ground truth to check against.
Agent skill for Riftbound: deck coaching, rules consultation and a practice partner’s decision layer. Repository public, not formally released. Unofficial. github.com/Zaious/riftbound-chronicle ↗
Basesmall
Keeps your team’s MLB game in a corner of the screen as a tiny window. Free, open-source desktop app. Unofficial. basesmall.chroniclecore.com ↗
SmalLoL
Basesmall’s League of Legends sibling: a thin desktop bar for following pro LoL matches without the stream, with the draft read live from the broadcast. Free and open source. Unofficial. smallol.chroniclecore.com ↗
OwnVoice Meet
For meetings, talks and Q&A in a second language: live captions with translation, and typed Chinese turned into short spoken English you read aloud yourself. Online it listens to the computer’s audio; at a conference venue, to a microphone, and it can keep recordings and transcripts on your computer, one file per session. Windows, free and open source. ownvoice.chroniclecore.com ↗
SmallStocks TW
The Taiwan-stocks member of the small series: a few stocks, ETFs or indices in a corner of the screen, with a short notice when a price crosses a threshold you set. Free and open source. Independent, with no affiliation to the exchanges or brokers. smallstocks.chroniclecore.com ↗
JevTRPG
Paste a résumé and TypeSafe’s Jev reads it into a 1920s d100 investigator sheet. Nothing pasted is stored. Unofficial. jevtrpg.chroniclecore.com ↗
KoreaCare
Snap an influencer ad or a package, find the original Korean product, and compare the ad’s price with Korean retail. koreacare.chroniclecore.com ↗
AChanGer
Intel on anime merchandise and purchase bonuses, with a no-login card-exchange tool. achanger.chroniclecore.com ↗