pith. sign in

arxiv: 2607.01776 · v1 · pith:YJZQZN2Anew · submitted 2026-07-02 · 💻 cs.CY · cs.AI

AI Virtue: What is "Good" Knowledge in the Age of Artificial Intelligence?

Alan Liu This is my paper

Pith reviewed 2026-07-03 06:18 UTC · model grok-4.3

classification 💻 cs.CY cs.AI
keywords epistemic virtuesAI knowledgedigital humanitiesgenerativitycreativityknowledge evaluationcultural AIAI ethics
0
0 comments X

The pith

Mapping epistemic virtues in 553 AI articles points toward a generativity-based framework for judging good knowledge.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper applies digital humanities methods to map epistemic virtues such as true, accurate, and creative across a corpus of 553 journal articles on AI published in 2024. It examines creativity in particular as a case study within this discourse of value. From the resulting patterns, the work considers how an evaluative framework for the knowledge-worth of AI might be constructed. The proposed framework would move away from values shaped by pre-AI knowledge work agents and structures and toward an openness to the values of generativity. An online digital kit is provided to explore the data models derived from the corpus.

Core claim

Through digital humanities mapping of epistemic virtues in the 2024 AI article corpus, with special attention to creativity, the discourse of value can serve as the basis for developing a framework to evaluate the knowledge-worth of AI that is less bound to pre-AI knowledge work paradigms and more attuned to the future values of generativity.

What carries the argument

Digital humanities mapping of epistemic virtues across the 553-article corpus, using creativity as the focal example, to ground consideration of a new evaluative framework.

If this is right

  • AI knowledge could be judged primarily by its generative capacity rather than by alignment with traditional accuracy or truth standards alone.
  • Individual human knowledge workers would lose their central position as the reference point for evaluating knowledge quality.
  • Creativity in AI contexts would be reinterpreted as a property of generative processes rather than solely of human insight.
  • Digital tools for corpus analysis would become standard for refining value frameworks as AI discourse evolves.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same mapping technique could be applied to other domains such as biotechnology or climate modeling to surface shifts in what counts as valuable knowledge.
  • Institutions might adopt generativity-oriented criteria when setting guidelines for AI use in research or education.
  • Empirical tests could check whether AI outputs rated high in generativity produce greater downstream societal effects than those rated by older virtues.

Load-bearing premise

The chosen corpus of 553 articles together with the digital humanities mapping of their epistemic virtues supplies an adequate foundation for building an evaluative framework that departs from pre-AI values.

What would settle it

Evidence that the virtues appearing in the 2024 AI articles are essentially the same as those in pre-AI discussions of knowledge production would show that the mapping does not enable a meaningfully new framework.

Figures

Figures reproduced from arXiv: 2607.01776 by Alan Liu.

Figure 4
Figure 4. Figure 4: Hypothetical groupings of epistemic values that may represent epistemic norms in discourse on AI. Thus consider how the kit can be used to explore my specific case study in this essay: creativity as a value. The following are three observations made with the kit, each leading to an intriguing research question. (1) Observing the words creative, creativity, and creatively, 31 we see that there is a notable … view at source ↗
Figure 5
Figure 5. Figure 5: Word-net graph of epistemic values and word-embedding neighbors for my WoS AHCI corpus. Terms in the “generative” family are close to “capable” and “ability” in the underlying embedding space. Seen one way, this registers a semantic difference from creativity. Creativity is genius or insight, we think, while generativity is just being capable in a workaday way. But seen another way, creativity, whose word … view at source ↗
read the original abstract

In the age of AI, what will be good knowledge? This article, which is accepted and forthcoming in a special issue of Modern Fiction Studies on "Cultural AI" in 2027, applies digital humanities methods to map epistemic virtues (like "true," "accurate," "creative") used in a corpus of 553 journal articles on AI published in 2024. "Creativity" comes in for special attention as an example. Exploring this discourse of value, the article considers how a framework might be developed for evaluating the knowledge-worth of AI -- one less locked into values formed around pre-AI "knowledge work" agents or structures, and more open to the future values of "generativity." The essay is supported by an online digital kit for exploring data models of the corpus of articles on AI it studies.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper applies digital humanities methods to map epistemic virtues (e.g., 'true,' 'accurate,' 'creative') across a corpus of 553 journal articles on AI published in 2024, with focused discussion of 'creativity.' It then considers how such mappings could support development of an evaluative framework for 'good' knowledge in AI—one oriented toward future 'generativity' values rather than pre-AI knowledge-work norms—and supplies an accompanying online digital kit for data exploration.

Significance. If the virtue mapping proves robust and the generativity framework is explicitly derived with clear, falsifiable criteria, the work could contribute to AI ethics and epistemology by providing an empirical basis for shifting evaluative standards. The digital kit is a positive feature that supports reproducibility and community extension of the analysis.

major comments (2)
  1. [Methods / Corpus description] The manuscript supplies no details on corpus construction (selection criteria for the 553 articles), coding procedures for epistemic virtues, validation steps, or inter-coder reliability. This absence directly undermines the reliability of the mapped virtue distributions that are intended to ground the proposed framework.
  2. [Discussion / Framework proposal] No control corpus from a pre-AI period is analyzed, and no explicit construction or re-weighting procedure is given to show how the observed 2024 virtues would be transformed into a 'generativity' criterion. The central claim that the mapping enables a framework departing from pre-AI values therefore rests on an unsupported interpretive move rather than demonstrated evidence.
minor comments (1)
  1. The abstract states that the essay is supported by an online digital kit, but the main text should include a dedicated subsection describing the kit's contents, data models, and how users can reproduce or extend the virtue mappings.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive report, which identifies key areas where greater transparency and explicitness would strengthen the manuscript. We address each major comment below, indicating planned revisions where appropriate.

read point-by-point responses

  1. Referee: [Methods / Corpus description] The manuscript supplies no details on corpus construction (selection criteria for the 553 articles), coding procedures for epistemic virtues, validation steps, or inter-coder reliability. This absence directly undermines the reliability of the mapped virtue distributions that are intended to ground the proposed framework.

    Authors: We agree that the current manuscript provides insufficient methodological detail on corpus construction and coding. Although the piece was written for a humanities journal special issue that favors interpretive narrative, the referee is correct that this limits assessment of the virtue mappings. In revision we will add a dedicated Methods subsection specifying the databases and search strings used to assemble the 553 articles, the procedure for extracting epistemic-virtue terms, the coding protocol (including whether it was manual or tool-assisted), and any validation or reliability checks performed. If single-coder coding was used, we will state the rationale and any supplementary checks. This addition will be made without altering the paper’s core argument. revision: yes

  2. Referee: [Discussion / Framework proposal] No control corpus from a pre-AI period is analyzed, and no explicit construction or re-weighting procedure is given to show how the observed 2024 virtues would be transformed into a 'generativity' criterion. The central claim that the mapping enables a framework departing from pre-AI values therefore rests on an unsupported interpretive move rather than demonstrated evidence.

    Authors: The manuscript does not present a comparative empirical study with a pre-AI control corpus, nor does it supply a formal re-weighting algorithm; the generativity framework is offered as a conceptual proposal derived from the observed 2024 distributions rather than a fully operationalized transformation. We therefore accept the referee’s observation that the interpretive step is not demonstrated with quantitative evidence. In revision we will (a) explicitly acknowledge the absence of a control corpus as a limitation and (b) add a short subsection that articulates the proposed generativity criteria in more falsifiable terms (e.g., how virtues such as creativity might be re-weighted toward future-oriented production). These clarifications will be added while preserving the paper’s primarily interpretive character. revision: partial

Circularity Check

0 steps flagged

No circularity: corpus mapping is descriptive; framework is proposed without reduction to inputs

full rationale

The paper applies digital humanities methods to map epistemic virtues (true, accurate, creative, etc.) across 553 2024 AI articles and then considers how a framework open to 'generativity' might be developed. This is an interpretive and exploratory exercise, not a derivation, prediction, or first-principles result that reduces to its own inputs by construction. No equations, fitted parameters renamed as predictions, self-citation load-bearing uniqueness theorems, or ansatzes smuggled via citation are present. The mapping step stands as independent description of the corpus; the framework suggestion is explicitly tentative and does not claim to be forced by the data or prior self-work. The analysis is self-contained against external benchmarks of corpus description.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical derivations, fitted parameters, or technical models are present in the abstract. The work relies on interpretive assumptions about what constitutes epistemic virtue and how corpus patterns translate into a new evaluative framework.

pith-pipeline@v0.9.1-grok · 5660 in / 1127 out tokens · 28988 ms · 2026-07-03T06:18:05.448159+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

21 extracted references · 17 canonical work pages · 1 internal anchor

  1. [1]

    Homogenization Effects of Large Language Models on Human Creative Ideation

    Anderson, Barrett R., et al. “Homogenization Effects of Large Language Models on Human Creative Ideation.” Creativity and Cognition, 2024, pp. 413–25, https://doi.org/10.1145/3635636.3656204. Anthony, Laurence. AntConc (Version 4.3.1). Waseda University, 29 July 2024, https://www.laurenceanthony.net/software/antconc/. ———. WordFamilyFinder, Ver. 1.0.1. n....

    work page doi:10.1145/3635636.3656204 2024

  2. [2]

    Attention (to Virtuosity) Is All You Need: Religious Studies Pedagogy and Generative AI

    Barlow, Jonathan, and Lynn Holt. “Attention (to Virtuosity) Is All You Need: Religious Studies Pedagogy and Generative AI.” Religions, vol. 15, no. 9.1059, 2024, https://doi.org/10.3390/rel15091059. Barnard, Joel. “What Are Word Embeddings?” IBM, n. d., https://www.ibm.com/think/topics/word-embeddings. Bauer, Laurie, and Paul Nation. “Word Families.” Inte...

    work page doi:10.3390/rel15091059 2024

  3. [4]

    On the Opportunities and Risks of Foundation Models

    arXiv.org, http://arxiv.org/abs/2108.07258. Bratton, Benjamin H. The Stack: On Software and Sovereignty. MIT Press,

    work page internal anchor Pith review Pith/arXiv arXiv

  4. [5]

    H2 2025 Innovation Economy Update

    Software Studies. Candy, Nick, and Vincent Harrison. “H2 2025 Innovation Economy Update.” J. P. Morgan, 28 Jan. 2026, https://www.jpmorgan.com/insights/markets-and-economy/economy/innovation- economy-outlook. Clanchy, M. T. From Memory to Written Record: England, 1066-1307. 3rd ed, Wiley,

    2025

  5. [6]

    Introduction: Epistemic Virtues in the Sciences and the Humanities

    www.wired.com, https://www.wired.com/story/tiktok-platforms-cory-doctorow/. Dongen, Jeroen van, and Herman Paul. “Introduction: Epistemic Virtues in the Sciences and the Humanities.” Epistemic Virtues in the Sciences and the Humanities, edited by Jeroen van Dongen 18 and Herman Paul, Springer International Publishing, 2017, pp. 1–10, https://doi.org/10.10...

    work page doi:10.1007/978-3-319-48893-6 2017

  6. [7]

    What Is an Author?

    Foucault, Michel. “What Is an Author?” Language Counter-Memory, Practice: Selected Essays and Interviews, edited by Donald F. Bouchard, translated by Donald F. Bouchard and Sherry Simon, Cornell University Press, 1977, pp. 113–38. Franceschelli, Giorgio, and Mirco Musolesi. “On the Creativity of Large Language Models.” AI & Society, vol. 40, Nov. 2024, pp...

    work page doi:10.1007/s00146-024-02127-3 1977

  7. [8]

    AI Is Changing the World: For Better or for Worse?

    Grewal, Dhruv, et al. “AI Is Changing the World: For Better or for Worse?” Journal of Macromarketing, vol. 44, no. 4, 2024, pp. 870–82, https://doi.org/10.1177/02761467241254450. Griffith, Erin. “A.I. Is Changing How Silicon Valley Builds Start-Ups.” New York Times, 20 Feb. 2025, https://www.nytimes.com/2025/02/20/technology/ai-silicon-valley-start-ups.ht...

    work page doi:10.1177/02761467241254450 2024

  8. [9]

    Using the Fractal Enterprise Model for Inter-Organizational Business Processes

    Henkel, Martin, et al. “Using the Fractal Enterprise Model for Inter-Organizational Business Processes.” Proceedings of the ER Forum and Poster & Demos Session 2019 on Publishing Papers with CEUR-WS Co-Located with 38th International Conference on Conceptual Modeling (ER 2019), Salvador, Brazil, November 4, 2019, edited by José Ignacio Panach et al., vol....

    2019

  9. [10]

    Networks of Power: Electrification in Western Society, 1880-1930

    Hughes, Thomas P. Networks of Power: Electrification in Western Society, 1880-1930. Johns Hopkins University Press,

    1930

  10. [11]

    Smaller Is Better in Silicon Valley’s ‘Tiny Team’ Moment

    Kelley, Lora. “Smaller Is Better in Silicon Valley’s ‘Tiny Team’ Moment.” New York Times, 28 Mar. 2026, https://www.nytimes.com/2026/03/28/business/silicon-valley-tiny-team-two- slice.html. 19 Kleinberg, Jon, and Manish Raghavan. “Algorithmic Monoculture and Social Welfare.” Proceedings of the National Academy of Sciences, vol. 118, no. 22, June 2021, p. ...

    work page doi:10.1073/pnas.2018340118 2026

  11. [12]

    Why Slop Matters

    16 Oct. 2025, JavaScript, HTML, Python. 2026, GitHub, https://github.com/scottkleinman/dfr-browser2. Kommers, Cody, et al. “Why Slop Matters.” ACM AI Letters, vol. 1, no. 1 Article 3, Mar. 2026, pp. 1–6, https://doi.org/10.1145/3786777. Latour, Bruno. “An Attempt at a ‘Compositionist Manifesto.’” New Literary History, vol. 41, no. 3, 2010, pp. 471–90. ———...

    work page doi:10.1145/3786777 2025

  12. [13]

    Artificial Intelligence Auxiliary Diagnosis and Treatment System for Breast Cancer in Developing Countries

    Li, Wenxiu, et al. “Artificial Intelligence Auxiliary Diagnosis and Treatment System for Breast Cancer in Developing Countries.” Journal of X-Ray Science and Technology, vol. 32, no. 2, Mar. 2024, pp. 395–413, https://doi.org/10.3233/XST-230194. Liu, Alan. “Kit for Exploring Articles on AI Published in 2024.” Alan Liu, 2025, https://alanyliu.org/citation/...

    work page doi:10.3233/xst-230194 2024

  13. [14]

    Thinking Destruction: Creativity, Rational Choice, Emergence, and Destruction Theory

    ———. “Thinking Destruction: Creativity, Rational Choice, Emergence, and Destruction Theory.” Occasion, vol. 1, 2009, https://shc.stanford.edu/arcade/publications/occasion/rational- choice-theory-and-humanities/thinking-destruction-creativity. ———. “What Is Good Writing in the Age of ChatGPT?” Alan Liu , 18 June 2023, https://liu.english.ucsb.edu/what-is-g...

    work page doi:10.6001/fil-soc.2024.35.2.8 2009

  14. [15]

    Creational and Conversational AI Affordances: How the New Breed of Chatbots Is Revolutionizing Knowledge Industries

    Project Gutenberg, https://www.gutenberg.org/ebooks/1200. Ramaul, Laavanya, et al. “Creational and Conversational AI Affordances: How the New Breed of Chatbots Is Revolutionizing Knowledge Industries.” Business Horizons, vol. 67, no. 5, 2024, pp. 615–27, https://doi.org/10.1016/j.bushor.2024.05.006. Rockmore, Dan. “What It’s Like to Brainstorm with a Bot....

    work page doi:10.1016/j.bushor.2024.05.006 2024

  15. [16]

    Agent Harness: The Infrastructure for Reliable AI

    www.newyorker.com, https://www.newyorker.com/culture/the-weekend-essay/what-its-like- to-brainstorm-with-a-bot. Salesforce, Inc. “Agent Harness: The Infrastructure for Reliable AI.” Agentforce, n. d., https://www.salesforce.com/agentforce/ai-agents/agent-harness/. Schindler, Samuel. “Theoretical Virtues: Do Scientists Think What Philosophers Think They Ou...

    work page doi:10.1017/psa.2021.40 2022

  16. [17]

    Who Are the Publics Engaging in AI?

    Sieber, Renée, et al. “Who Are the Publics Engaging in AI?” Public Understanding of Science, vol. 33, no. 5, July 2024, pp. 634–53, https://doi.org/10.1177/09636625231219853. Stiegler, Bernard. Technics and Time. Translated by Richard Beardsworth and George Colins, vol. 1, Stanford University Press,

    work page doi:10.1177/09636625231219853 2024

  17. [18]

    Digital Writing with AI Platforms: The Role of Fun with/in Generative AI

    Stornaiuolo, Amy, et al. “Digital Writing with AI Platforms: The Role of Fun with/in Generative AI.” English Teaching: Practice & Critique, vol. 23, no. 1, 2024, pp. 83–103, https://doi.org/10.1108/ETPC-08-2023-0103. Sui, Peiqi, Eamon Duede, Sophie Wu, et al. “Confabulation: The Surprising Value of Large Language Model Hallucinations.” arXiv:2406.04175, a...

    work page doi:10.1108/etpc-08-2023-0103 2024

  18. [19]

    Critical Confabulation: Can LLMs Hallucinate for Social Good?

    arXiv.org, https://doi.org/10.48550/arXiv.2406.04175. Sui, Peiqi, Eamon Duede, Hoyt Long, et al. “Critical Confabulation: Can LLMs Hallucinate for Social Good?” arXiv:2511.07722, arXiv, 11 Nov

    work page doi:10.48550/arxiv.2406.04175

  19. [20]

    The Agentic Organization: Contours of the Next Paradigm for the AI Era

    arXiv.org, https://doi.org/10.48550/arXiv.2511.07722. 21 Sukharevsky, Alexander, et al. “The Agentic Organization: Contours of the Next Paradigm for the AI Era.” McKinsey & Company, 26 Sept. 2025, https://www.mckinsey.com/capabilities/people-and-organizational-performance/our- insights/the-agentic-organization-contours-of-the-next-paradigm-for-the-ai-era....

    work page doi:10.48550/arxiv.2511.07722 2025

  20. [21]

    Advancing Human-Machine Teaming: Definitions, Challenges, Future Directions

    Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/archives/win2021/entries/epistemology-virtue/. University of California. Report of the Academic Senate Workgroup on Artificial Intelligence. 10 Dec. 2025, pp. 1–20, https://senate.universityofcalifornia.edu/_files/reports/council-chair-to-senate- divisions-senate-ai-workgroup-report.pdf. Verh...

    work page doi:10.3233/faia250624 2025

  21. [22]

    The Role of AI in Changing Company Structures and Dynamics

    Williamson, Dael, and Fabian Lanz. “The Role of AI in Changing Company Structures and Dynamics.” Databricks, 12 Nov. 2024, https://www.databricks.com/blog/role-ai-changing- company-structures-and-dynamics. Zhang, Yan, and Barbara M. Wildemuth. “Qualitative Analysis of Content.” Applications of Social Research Methods to Questions in Information and Librar...

    2024