Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

The paper introduces structural transparency, a framework for making visible the institutional and organizational decisions that shape AI alignment.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review

2026-08-03 03:18 UTC pith:7EEE77XI

load-bearing objection A well-built conceptual framework for macro-level transparency in AI alignment; the C2 hybrid/hijacked classifier is the main soft spot and needs empirical testing. the 3 major comments →

arxiv 2602.08246 v1 pith:7EEE77XI submitted 2026-02-09 cs.CY

Structural transparency of societal AI alignment through Institutional Logics

Atrisha Sarkar , Isam Faik This is my paper
classification cs.CY
keywords AI alignmentstructural transparencyinstitutional logicsAI governancesociotechnical harmsorganizational decisionsgenerative AItransparency
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper argues that transparency about AI alignment has focused on models, data, and procedures, while the institutional and organizational decisions that shape which values get encoded remain opaque. To close that gap, it introduces 'structural transparency': a systematic analysis of the institutional structures and organizational decisions behind the constitutive inputs of alignment, and of how deployed aligned systems reshape those structures. The framework is operationalized as five analytical components with step-by-step 'analyst recipes' that identify dominant and secondary institutional logics, distinguish genuine integration from symbolic invocation, trace disruptions to external fields, and map logics to sociotechnical harms. A sympathetic reader would take this as a serious proposal to extend transparency from the informational to the institutional level, giving accountability efforts a macro-level lens.

Core claim

On the paper's own terms, the central discovery is that the organizational decisions that constitute AI alignment—annotator selection, data sourcing, principle definitions, safety guardrails, deployment choices—are not neutral technical choices but expressions of institutional logics: socially constructed patterns of meaning and practice associated with orders such as the market, state, profession, corporation, community, religion, and family. By analyzing these logics through five components, the framework claims to make visible how alignment decisions are made, how those decisions disrupt or reinforce existing social orders, and how the structural risks of each logic translate into catalog

What carries the argument

The central object is 'structural transparency,' defined as the systematic analysis that makes visible the institutional structures and organizational decisions shaping the constitutive elements of alignment, and the ways aligned systems subsequently reshape those structures once deployed. The mechanism that carries the argument is the five-component analytical procedure (C1–C5): identifying primary and secondary institutional logics governing alignment decisions (C1); determining whether secondary logics are substantively integrated ('hybrid') or only rhetorically invoked ('hijacked') using indicators of temporal durability, organizational integration, and conflict resolution (C2); assessin

Load-bearing premise

The classification of a secondary logic as genuinely 'hybrid' versus merely 'hijacked' relies on inferring organizational intent from observable indicators, which the paper concedes is hard; if analysts cannot apply these indicators consistently to real organizations, the entire harm-mapping chain loses its basis.

What would settle it

Apply the C2 recipe to a documented case of ethics washing—an organization publicly committed to ethical alignment principles but with no evidence of conflict-resolution mechanisms or durable integration—and show that the recipe classifies it as a hybrid logic. Alternatively, have multiple trained analysts independently score the same organizational documents for the three indicators and show that inter-rater agreement is too low to produce stable classifications; either result would undermine the operational core of the framework.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Analysts can complement informational transparency (model, data, procedure) with a macro-level account of the institutional forces behind alignment choices.
  • Organizations' alignment decisions become comparable: the same alignment technique can be governed by different logics, yielding different structural risk profiles.
  • The framework gives a structured way to anticipate sociotechnical harms before deployment by tracing logics to harm categories.
  • The analytical approach can be adapted from organizational decisions to AI policy, broadening its governance relevance.
  • Public and internal documents, contracts, and communications can serve as evidence for institutional logics, making the analysis feasible without privileged access to model internals.

Where Pith is reading between the lines

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

  • If the framework is right, regulators could build an 'institutional audit' that screens companies' alignment documentation for hybrid vs. hijacked logics, though the paper itself stops short of proposing such an instrument.
  • The hybrid/hijacked distinction yields a testable prediction: organizations with true hybrid logics should show observable differences in decision outcomes—such as resource allocation, grievance handling, or conflict-resolution usage—compared with organizations that merely invoke secondary logics; a comparative case study could test this.
  • The paper's harm mapping (C5) could be operationalized into a risk registry for deployed systems by combining its logic-to-risk table with existing algorithmic harm taxonomies, a step the authors leave for future empirical work.
  • The framework may connect to a broader research program on 'ethics washing' by providing an operational definition of when ethics language is substantive versus symbolic, but that connection is the paper's implicit extension rather than an explicit claim.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a framework of 'structural transparency' for analyzing the institutional and organizational decisions that shape AI alignment, grounded in the theory of institutional logics. It introduces five analytical components (C1–C5) with accompanying 'analyst recipes': C1 identifies primary and secondary institutional logics in alignment decisions; C2 distinguishes 'hybrid' from 'hijacked' relations between logics; C3 assesses external disruptions of aligned systems on a target field (e.g., higher education); C4 examines how external pressures are internalized by the organization; and C5 maps institutional logics to structural risks and, via Shelby et al.'s harm taxonomy, to sociotechnical harms. The paper motivates the framework by arguing that existing informational, material, and procedural transparency approaches omit the macro-level institutional forces that shape alignment. A hypothetical LLM-assisted tutoring scenario illustrates the components. The authors claim that structural transparency complements informational transparency by making visible the institutional structures behind alignment decisions and their societal consequences.

Significance. The paper addresses a real gap: AI alignment transparency has largely focused on model internals, data, and procedures, while the organizational and institutional determinants of alignment choices remain underexamined. If the framework can be reliably operationalized, it offers a novel lens that connects macro-level institutional analysis to concrete sociotechnical harms, potentially enriching AI governance scholarship. Strengths include the explicit grounding in established institutional-logics theory, a transparent component-wise decomposition, and the honest labeling of the illustrative example as hypothetical. The paper also acknowledges key limitations, such as the difficulty of inferring intent from organizational actions. However, the framework's practical usefulness hinges on whether analysts can apply the recipes consistently, which is not empirically demonstrated.

major comments (3)
  1. [Section 4, C2] The recipe for distinguishing 'hybrid' from 'hijacked' logics is underdetermined. The paper concedes 'it is hard to infer intent from observed organizational actions,' yet the three proposed indicators—temporal durability, organizational integration, and conflict resolution—can be satisfied by strategic or ceremonial adoption. A long-term contract may reflect lock-in rather than genuine commitment; an oversight committee may lack authority; conflict-resolution processes can be performative. In the illustrative example (Section 5, C2), the professional/market relationship is classified as hybrid on the basis of contract duration and committee oversight, but these observations are equally consistent with a hijacked logic. Because C5's risk composition uses this classification to decide whether a secondary logic mitigates or exacerbates harms (Section 4, C5, step 4), misclassification mater
  2. [Section 4, C5, Table 2] The mapping from institutional logics to structural risk categories (Table 2) is asserted row-by-row with isolated citations (e.g., State→Surveillance [31]; Market→Market failure [58]; Profession→Technocratic gatekeeping [5]). No derivation is provided from the categorical elements of institutional logics (Fig. 2) or from a principled reading of Shelby et al.'s taxonomy. Several mappings are contestable: 'market failure' in economic theory does not directly correspond to the listed socioeconomic harms; 'religion→religious persecution' conflates religious institutions with actions by external actors. Since C5 instructs the analyst to map each logic in the logic set to baseline risks using Table 2, the output harm set is only as defensible as this mapping. The paper should justify the selection of risk categories, clarify the risk–harm distinction, and ideally validate the mapping against
  3. [Section 4, C3 and C5] The 'analyst recipes' for C3 (assessing segregation mechanisms) and C5 (risk composition) rely on qualitative judgments with no operational guidance comparable to the indicators in C2. In C3 Step 3, the analyst must determine whether a field has 'credible mechanisms' of segregation, but no criteria are provided. In C5 step 4, deployment conditions 'activate' or 'constrain' structural risks, but the paper does not specify how to weigh different information artefacts (contracts, policy documents, press releases). The central claim that structural transparency 'enables analysts' to produce dependable macro-level analyses remains unsupported; the illustrative example is author-constructed and cannot demonstrate reproducible application. The paper should either provide more explicit coding procedures for these judgments or explicitly acknowledge that the framework currently requires substanti
minor comments (5)
  1. [Abstract] Typo: 'existing approached based on informational transparency' should be 'existing approaches.'
  2. [Section 4, C1] Typo: 'aa priori conceptualisations' should be 'a priori conceptualisations.'
  3. [Figure 2 caption] The name is misspelled: 'Thornton and Ocassio 2012' should be 'Thornton and Ocasio 2012.'
  4. [Section 2] Use 'cf.' instead of 'c.f.' in two places.
  5. [Section 7] The Collingridge dilemma is attributed to [15, 61]; reference [61] is Ribeiro et al., not Collingridge. Please cite the original source or clarify.

Circularity Check

0 steps flagged

No significant circularity: the framework is a conceptual synthesis with an explicitly stated mapping table; C5 is an operationalization, not a derived prediction.

full rationale

The paper does not contain fitted parameters, mathematical derivations, or empirical predictions that could reduce to their inputs. Structural transparency is defined as a mode of analysis, and the five components C1–C5 are presented as explicit analyst recipes rather than as theorems or forecasts. The most plausibly circular-looking step is C5, where the logic set identified in C1–C4 is mapped through Table 2 to sociotechnical harms. However, this is transparently a mapping by construction: the paper states that Table 2 provides the risk/benefit categories the analyst 'can draw from' and that the output is a 'candidate set of harms that are structurally plausible given the institutional ordering of the system.' The illustrative example explicitly invokes this mapping ('The structural risk mapping (Table 2) identifies that market logic carries risks of market failure, which map to sociotechnical harms...'). This is an operationalization of a framework, not a hidden tautology presented as an empirical result. There are no load-bearing self-citations: the authors do not cite their own prior work as authority for any central premise. The operationalizability concern about C2's hybrid/hijacked distinction is a validity and reliability question, not a circularity one; the paper itself concedes that 'it is hard to infer intent from observed organizational actions,' which is an honest limitation rather than a circular step. The framework is therefore self-contained as a conceptual contribution, and no circularity score is warranted.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 0 invented entities

The framework introduces no free parameters or invented entities; it is a conceptual synthesis. Its analytical output, however, is heavily determined by the input categories (logics, harms taxonomy) and the asserted Table 2 mapping, which functions as a compact set of assumed relationships rather than an empirically derived result.

axioms (5)
  • domain assumption Institutional Logics theory (Thornton & Ocasio) provides a valid analytical lens for understanding organizational decisions in AI alignment.
    Section 3 adopts the theory wholesale as 'the underlying theoretical foundation' without independent justification. The framework's validity depends on this lens actually explaining alignment-relevant behavior.
  • domain assumption AI alignment can be exhaustively categorized into learning-from-feedback, assurance, and governance (from Ji et al. 2025).
    Table 1 builds on this categorization. If there are alignment-relevant organizational decisions outside these three categories, the framework's scope is incomplete.
  • domain assumption Values reproduced by generative AI are shaped primarily by macro/meso institutional structures rather than reducible to individuals.
    Section 1 critiques methodological individualism and posits that institutions shape values. This is a foundational premise, not empirically established within the paper.
  • ad hoc to paper The mapping in Table 2 from institutional logics to structural risks and sociotechnical harms is valid.
    Table 2 is constructed by selecting one or two citations per logic-row (e.g., surveillance for state, shareholder capitalism for corporation). No systematic derivation or empirical validation is provided; the table is asserted as a baseline for C5.
  • domain assumption The sociotechnical harms taxonomy of Shelby et al. (2023) is adopted as the definitive catalogue of harms.
    C5 relies on Shelby et al.'s taxonomy to translate structural risks into concrete harms. The completeness and cross-cultural validity of that taxonomy is not examined here.

reviewed 2026-08-03 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Structural transparency of societal AI alignment through Institutional Logics." pith.science (2026). https://pith.science/paper/7EEE77XI

@misc{pith2026260208246,
  author       = {Pith},
  title        = {Pith review of: Structural transparency of societal AI alignment through Institutional Logics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7EEE77XI}},
  note         = {Machine review of arXiv:2602.08246}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The field of AI alignment is increasingly concerned with the questions of how values are integrated into the design of generative AI systems and how their integration shapes the social consequences of AI. However, existing transparency frameworks focus on the informational aspects of AI models, data, and procedures, while the institutional and organizational forces that shape alignment decisions and their downstream effects remain underexamined in both research and practice. To address this gap, we develop a framework of \emph{structural transparency} for analyzing organizational and institutional decisions concerning AI alignment, drawing on the theoretical lens of Institutional Logics. We develop a categorization of organizational decisions that are present in the governance of AI alignment, and provide an explicit analytical approach to examining them. We operationalize the framework through five analytical components, each with an accompanying "analyst recipe" that collectively identify the primary institutional logics and their internal relationships, external disruptions to existing social orders, and finally, how the structural risks of each institutional logic are mapped to a catalogue of sociotechnical harms. The proposed concept of structural transparency enables analysts to complement existing approached based on informational transparency with macro-level analyses that capture the institutional dynamics and consequences of decisions regarding AI alignment.

Figures

Figures reproduced from arXiv: 2602.08246 by Atrisha Sarkar, Isam Faik.

Figure 1
Figure 1. Figure 1: Analytical components of Structural Transparency. Institutional logics inform the analytical components (C1–C5), which help [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: ’Ideal types’ of institutional orders (reproduced from Thornton and Ocassio 2012 [ [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. NodeSynth: Socially Aligned Synthetic Data for AI Evaluation

    cs.LG 2026-05 unverdicted novelty 6.0

    NodeSynth generates evidence-anchored synthetic queries that trigger up to five times higher failure rates in mainstream LLMs than human-authored benchmarks.

  2. NodeSynth: Socially Aligned Synthetic Data for AI Evaluation

    cs.LG 2026-05 unverdicted novelty 6.0

    NodeSynth creates evidence-based synthetic queries via a taxonomy generator to evaluate LLMs, revealing up to 5x higher failure rates than human benchmarks and gaps in guard models.

Reference graph

Works this paper leans on

81 extracted references · 7 linked inside Pith · cited by 1 Pith paper

  1. [1]

    Gloria Andrada, Robert W Clowes, and Paul R Smart. 2023. Varieties of transparency: Exploring agency within AI systems.AI & society38, 4 (2023), 1321–1331

    2023

  2. [2]

    Anthropic. 2025. Understanding and Addressing AI Harms — anthropic.com. https://www.anthropic.com/news/our-approach-to-understanding- and-addressing-ai-harms. [Accessed 31-12-2025]

    2025

  3. [3]

    Aparna Balagopalan, David Madras, David H Yang, Dylan Hadfield-Menell, Gillian K Hadfield, and Marzyeh Ghassemi. 2023. Judging facts, judging norms: Training machine learning models to judge humans requires a modified approach to labeling data.Science advances9, 19 (2023), eabq0701

    2023

  4. [4]

    Stevie Bergman, Nahema Marchal, John Mellor, Shakir Mohamed, Iason Gabriel, and William Isaac. 2024. STELA: a community-centred approach to norm elicitation for AI alignment.Scientific Reports14, 1 (2024), 6616

    2024

  5. [5]

    2020.The technocratic challenge to democracy

    Eri Bertsou and Daniele Caramani. 2020.The technocratic challenge to democracy. Routledge London

    2020

  6. [6]

    Elettra Bietti. 2020. From ethics washing to ethics bashing: a view on tech ethics from within moral philosophy. InProceedings of the 2020 conference on fairness, accountability, and transparency. 210–219

    2020

  7. [7]

    2011.The open innovation marketplace: creating value in the challenge driven enterprise

    Alpheus Bingham and Dwayne Spradlin. 2011.The open innovation marketplace: creating value in the challenge driven enterprise. FT press

    2011

  8. [8]

    Abeba Birhane, William Isaac, Vinodkumar Prabhakaran, Mark Diaz, Madeleine Clare Elish, Iason Gabriel, and Shakir Mohamed. 2022. Power to the people? Opportunities and challenges for participatory AI. InProceedings of the 2nd ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization. 1–8

    2022

  9. [9]

    Stefan Buijsman. 2024. Transparency for AI systems: a value-based approach.Ethics and Information Technology26, 2 (2024), 34

    2024

  10. [10]

    Yuzhuo Cai and Nicola Mountford. 2022. Institutional logics analysis in higher education research.Studies in Higher Education47, 8 (2022), 1627–1651

    2022

  11. [11]

    Jidong Chen and Yiqing Xu. 2017. Information manipulation and reform in authoritarian regimes.Political Science Research and Methods5, 1 (2017), 163–178

    2017

  12. [12]

    Yuchen Chen, Silvia Lindtner, and Yuling Sun. 2025. The Moral Economy of AI. InProceedings of the sixth decennial Aarhus conference: Computing X Crisis. 117–126

    2025

  13. [13]

    Wolfie Christl, Katharina Kopp, and Patrick Urs Riechert. 2017. Corporate surveillance in everyday life.Cracked Labs6, 1 (2017), 1

    2017

  14. [14]

    Jack Clark and Gillian K Hadfield. 2019. Regulatory markets for AI safety.arXiv preprint arXiv:2001.00078(2019)

    Pith/arXiv arXiv 2019

  15. [15]

    Collingridge

    D. Collingridge. 1980.The Social Control of Technology. St. Martin’s Press. https://books.google.ca/books?id=hCSdAQAACAAJ

    1980

  16. [16]

    Vincent Conitzer, Rachel Freedman, Jobst Heitzig, Wesley H Holliday, Bob M Jacobs, Nathan Lambert, Milan Mossé, Eric Pacuit, Stuart Russell, Hailey Schoelkopf, et al. 2024. Social choice for ai alignment: Dealing with diverse human feedback.CoRR(2024)

    2024

  17. [17]

    Fernando Delgado, Stephen Yang, Michael Madaio, and Qian Yang. 2023. The participatory turn in ai design: Theoretical foundations and the current state of practice. InProceedings of the 3rd ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization. 1–23

    2023

  18. [18]

    Wesley Hanwen Deng, Nur Yildirim, Monica Chang, Motahhare Eslami, Kenneth Holstein, and Michael Madaio. 2023. Investigating practices and opportunities for cross-functional collaboration around AI fairness in industry practice. InProceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency. 705–716

    2023

  19. [19]

    Nicholas Diakopoulos. 2020. 197 Transparency. InThe Oxford Handbook of Ethics of AI. Oxford University Press, 197–213

    2020

  20. [20]

    Ewald Engelen. 2002. Corporate governance, property and democracy: a conceptual critique of shareholder ideology.Economy and Society31, 3 (2002), 391–413

    2002

  21. [21]

    Robin Fincham and Tom Forbes. 2015. Three’s a crowd: The role of inter-logic relationships in highly complex institutional fields.British Journal of Management26, 4 (2015), 657–670

    2015

  22. [22]

    Roger Friedland. 1991. Bringing society back in: Symbols, practices, and institutional contradictions.The new institutionalism in organizational analysis(1991), 232–263

    1991

  23. [23]

    Batya Friedman, Peter H Kahn Jr, Alan Borning, and Alina Huldtgren. 2013. Value sensitive design and information systems. InEarly engagement and new technologies: Opening up the laboratory. Springer, 55–95

    2013

  24. [24]

    Iason Gabriel. 2020. Artificial intelligence, values, and alignment.Minds and machines30, 3 (2020), 411–437

    2020

  25. [25]

    Iason Gabriel and Vafa Ghazavi. 2022. The challenge of value alignment.The Oxford handbook of digital ethics(2022), 336–355

    2022

  26. [26]

    Andreas Georgiou and Daniel Arenas. 2023. Community in organizational research: A review and an institutional logics perspective.Organization Theory4, 1 (2023), 26317877231153189

    2023

  27. [27]

    Giuseppe Grossi, Dorota Dobija, and Wojciech Strzelczyk. 2020. The impact of competing institutional pressures and logics on the use of performance measurement in hybrid universities.Public performance & management review43, 4 (2020), 818–844

    2020

  28. [28]

    Sanford J Grossman and Joseph E Stiglitz. 1980. On the impossibility of informationally efficient markets.The American economic review70, 3 (1980), 393–408. Manuscript submitted to ACM Structural transparency of societal AI alignment through Institutional Logics. 17

    1980

  29. [29]

    2009.Ethical leadership in higher education: Evolution of institutional ethics logic

    William Roderick Hanson. 2009.Ethical leadership in higher education: Evolution of institutional ethics logic. Ph. D. Dissertation. Clemson University

    2009

  30. [30]

    Geoffrey M Hodgson. 2007. Meanings of methodological individualism.Journal of Economic Methodology14, 2 (2007), 211–226

    2007

  31. [31]

    Margaret Hu. 2017. From the national surveillance state to the cybersurveillance state.Annual Review of Law and Social Science13, 1 (2017), 161–180

    2017

  32. [32]

    Saffron Huang, Divya Siddarth, Liane Lovitt, Thomas I Liao, Esin Durmus, Alex Tamkin, and Deep Ganguli. 2024. Collective constitutional ai: Aligning a language model with public input. InProceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency. 1395–1417

    2024

  33. [33]

    Jianchao Ji, Yutong Chen, Mingyu Jin, Wujiang Xu, Wenyue Hua, and Yongfeng Zhang. 2024. Moralbench: Moral evaluation of llms.arXiv preprint arXiv:2406.04428(2024)

    Pith/arXiv arXiv 2024

  34. [34]

    Jiaming Ji, Tianyi Qiu, Boyuan Chen, Jiayi Zhou, Borong Zhang, Donghai Hong, Hantao Lou, Kaile Wang, Yawen Duan, Zhonghao He, et al. 2025. AI Alignment: A Contemporary Survey.Comput. Surveys(2025)

    2025

  35. [35]

    Kirsi-Mari Kallio, Tomi J Kallio, Giuseppe Grossi, and Janne Engblom. 2021. Institutional logic and scholars’ reactions to performance measurement in universities.Accounting, Auditing & Accountability Journal34, 9 (2021), 135–161

    2021

  36. [36]

    Timo Kaufmann, Paul Weng, Viktor Bengs, and Eyke Hüllermeier. 2024. A survey of reinforcement learning from human feedback. (2024)

    2024

  37. [37]

    Philipp Kellmeyer. 2024. Beyond participation: Towards a community-led approach to value alignment of AI in medicine. InDevelopments in neuroethics and bioethics. Vol. 7. Elsevier, 249–269

    2024

  38. [38]

    alignment

    Hannah Rose Kirk, Bertie Vidgen, Paul Röttger, and Scott A Hale. 2023. The empty signifier problem: Towards clearer paradigms for operationalising" alignment" in large language models.arXiv preprint arXiv:2310.02457(2023)

    Pith/arXiv arXiv 2023

  39. [39]

    2022.Aligned with whom? Direct and social goals for AI systems

    Anton Korinek and Avital Balwit. 2022.Aligned with whom? Direct and social goals for AI systems. Technical Report. National Bureau of Economic Research

    2022

  40. [40]

    Stefan Larsson and Fredrik Heintz. 2020. Transparency in artificial intelligence.Internet policy review9, 2 (2020), 1–16

    2020

  41. [41]

    John Law. 2016. STS as Method.The handbook of science and technology studies(2016), 31

    2016

  42. [42]

    Harrison Lee, Samrat Phatale, Hassan Mansoor, Kellie Ren Lu, Thomas Mesnard, Johan Ferret, Colton Bishop, Ethan Hall, Victor Carbune, and Abhinav Rastogi. 2023. Rlaif: Scaling reinforcement learning from human feedback with ai feedback. (2023)

    2023

  43. [43]

    Min-Dong Paul Lee and Michael Lounsbury. 2015. Filtering institutional logics: Community logic variation and differential responses to the institutional complexity of toxic waste.Organization Science26, 3 (2015), 847–866

    2015

  44. [44]

    Joel Z Leibo, Alexander Sasha Vezhnevets, Manfred Diaz, John P Agapiou, William A Cunningham, Peter Sunehag, Julia Haas, Raphael Koster, Edgar A Duéñez-Guzmán, William S Isaac, et al. 2024. A theory of appropriateness with applications to generative artificial intelligence.arXiv preprint arXiv:2412.19010(2024)

    Pith/arXiv arXiv 2024

  45. [45]

    Joon Soo Lim, Chunsik Lee, Donghee Shin, Junga Kim, and Jun Zhang. 2025. Perceived Stakeholder Engagement in Corporate Data Responsibility (CDR) Communication and Its Relationship with Trust in Generative AI Systems: The Mediating Role of Algorithmic and Institutional Responsibility. Journal of Public Relations Research(2025), 1–23

    2025

  46. [46]

    Ruibo Liu, Ruixin Yang, Chenyan Jia, Ge Zhang, Denny Zhou, Andrew M Dai, Diyi Yang, and Soroush Vosoughi. 2023. Training socially aligned language models in simulated human society.arXiv preprint arXiv:2305.169602 (2023)

    Pith/arXiv arXiv 2023

  47. [47]

    Brady Lund, Zeynep Orhan, Nishith Reddy Mannuru, Ravi Varma Kumar Bevara, Brett Porter, Meka Kasi Vinaih, and Padmapadanand Bhaskara

  48. [48]

    Dag Øivind Madsen and Kåre Slåtten. 2025. Podcasting Management: How Audio Platforms Are Shaping Business Ideas.Administrative Sciences15, 9 (2025), 342

    2025

  49. [49]

    Toni Makkai and John Braithwaite. 1992. In and out of the revolving door: Making sense of regulatory capture.Journal of public policy12, 1 (1992), 61–78

    1992

  50. [50]

    Kevin R McKee, Ian Gemp, Brian McWilliams, Edgar A Duéñez-Guzmán, Edward Hughes, and Joel Z Leibo. 2020. Social diversity and social preferences in mixed-motive reinforcement learning.arXiv preprint arXiv:2002.02325(2020)

    Pith/arXiv arXiv 2020

  51. [51]

    Chad Michael McPherson and Michael Sauder. 2013. Logics in action: Managing institutional complexity in a drug court.Administrative science quarterly58, 2 (2013), 165–196

    2013

  52. [52]

    Dang Minh, H Xiang Wang, Y Fen Li, and Tan N Nguyen. 2022. Explainable artificial intelligence: a comprehensive review.Artificial Intelligence Review55, 5 (2022), 3503–3568

    2022

  53. [53]

    Margaret Mitchell, Simone Wu, Andrew Zaldivar, Parker Barnes, Lucy Vasserman, Ben Hutchinson, Elena Spitzer, Inioluwa Deborah Raji, and Timnit Gebru. 2019. Model cards for model reporting. InProceedings of the conference on fairness, accountability, and transparency. 220–229

    2019

  54. [54]

    C Thi Nguyen. 2020. Echo chambers and epistemic bubbles.Episteme17, 2 (2020), 141–161

    2020

  55. [55]

    Rajvardhan Patil, Sorio Boit, Venkat Gudivada, and Jagadeesh Nandigam. 2023. A survey of text representation and embedding techniques in nlp. IEEe Access11 (2023), 36120–36146

    2023

  56. [56]

    Vasile-Daniel Păvăloaia and Sabina-Cristiana Necula. 2023. Artificial intelligence as a disruptive technology—a systematic literature review. Electronics12, 5 (2023), 1102

    2023

  57. [57]

    Andi Peng, Besmira Nushi, Emre Kiciman, Kori Inkpen, and Ece Kamar. 2022. Investigations of performance and bias in human-AI teamwork in hiring. InProceedings of the AAAI conference on artificial intelligence, Vol. 36. 12089–12097

    2022

  58. [58]

    Alan Randall. 1983. The problem of market failure.Natural Resources Journal23, 1 (1983), 131–148

    1983

  59. [59]

    Trish Reay and Candace Jones. 2016. Qualitatively capturing institutional logics.Strategic organization14, 4 (2016), 441–454. Manuscript submitted to ACM 18 Atrisha Sarkar and Isam Faik

    2016

  60. [60]

    Ellen Reeves, Jasmine McGowan, and Ben Scott. 2025. ‘It was dangerous, corrosive and cruel but not illegal’: Legal help-seeking behaviours amongst LGBTQA+ domestic and family violence victim-survivors experiencing coercive control in Australia.Journal of Family Violence40, 1 (2025), 27–38

    2025

  61. [61]

    Barbara Ribeiro, Lars Bengtsson, Paul Benneworth, Susanne Bührer, Elena Castro-Martínez, Meiken Hansen, Katharina Jarmai, Ralf Lindner, Julia Olmos-Peñuela, Cordula Ott, et al. 2018. Introducing the dilemma of societal alignment for inclusive and responsible research and innovation. Journal of responsible innovation5, 3 (2018), 316–331

    2018

  62. [62]

    Mohammad Rashidujjaman Rifat, Dipto Das, Arpon Poddar, Mahiratul Jannat, Robert Soden, Bryan Semaan, and Syed Ishtiaque Ahmed. 2024. The Politics of Fear and the Experience of Bangladeshi Religious Minority Communities Using Social Media Platforms.Proceedings of the ACM on Human-Computer Interaction8, CSCW2 (2024), 1–32

    2024

  63. [63]

    Bruce H Rowlands. 2005. Grounded in practice: Using interpretive research to build theory.Electronic Journal of Business Research Methods3, 1 (2005), pp81–92

    2005

  64. [64]

    Catharina Rudschies, Ingrid Schneider, and Judith Simon. 2020. Value pluralism in the AI ethics debate–Different actors, different priorities.The International Review of Information Ethics29 (2020)

    2020

  65. [65]

    Nino Scherrer, Claudia Shi, Amir Feder, and David Blei. 2023. Evaluating the moral beliefs encoded in llms.Advances in Neural Information Processing Systems36 (2023), 51778–51809

    2023

  66. [66]

    Renee Shelby, Shalaleh Rismani, Kathryn Henne, AJung Moon, Negar Rostamzadeh, Paul Nicholas, N’Mah Yilla-Akbari, Jess Gallegos, Andrew Smart, Emilio Garcia, et al. 2023. Sociotechnical harms of algorithmic systems: Scoping a taxonomy for harm reduction. InProceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society. 723–741

    2023

  67. [67]

    Hua Shen, Tiffany Knearem, Reshmi Ghosh, Michael Xieyang Liu, Andrés Monroy-Hernández, Tongshuang Wu, Diyi Yang, Yun Huang, Tanushree Mitra, Yang Li, et al. 2025. Bidirectional Human-AI Alignment: Emerging Challenges and Opportunities. InProceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems. 1–6

    2025

  68. [68]

    Chris Skelcher and Steven Rathgeb Smith. 2015. Theorizing hybridity: Institutional logics, complex organizations, and actor identities: The case of nonprofits.Public administration93, 2 (2015), 433–448

    2015

  69. [69]

    Karolina Stańczak, Nicholas Meade, Mehar Bhatia, Hattie Zhou, Konstantin Böttinger, Jeremy Barnes, Jason Stanley, Jessica Montgomery, Richard Zemel, Nicolas Papernot, et al. 2025. Societal alignment frameworks can improve llm alignment.arXiv preprint arXiv:2503.00069(2025)

    Pith/arXiv arXiv 2025

  70. [70]

    David Szakonyi. 2019. Princelings in the private sector: The value of nepotism.Quarterly Journal of Political Science14, 4 (2019), 349–381

    2019

  71. [71]

    Justin Tan and Liang Wang. 2011. MNC strategic responses to ethical pressure: An institutional logic perspective.Journal of Business Ethics98, 3 (2011), 373–390

    2011

  72. [72]

    Yan Tao, Olga Viberg, Ryan S Baker, and René F Kizilcec. 2024. Cultural bias and cultural alignment of large language models.PNAS nexus3, 9 (2024), pgae346

    2024

  73. [73]

    Patricia H Thornton and William Ocasio. 2008. Institutional logics.The Sage handbook of organizational institutionalism840, 2008 (2008), 99–128

    2008

  74. [74]

    2012.The institutional logics perspective: A new approach to culture, structure, and process

    Patricia H Thornton, William Ocasio, and Michael Lounsbury. 2012.The institutional logics perspective: A new approach to culture, structure, and process. Oxford University Press

    2012

  75. [75]

    Sunil Thulasidasan, Sushil Thapa, Sayera Dhaubhadel, Gopinath Chennupati, Tanmoy Bhattacharya, and Jeff Bilmes. 2021. An effective baseline for robustness to distributional shift. In2021 20th IEEE International Conference on Machine Learning and Applications (ICMLA). IEEE, 278–285

    2021

  76. [76]

    H Akin Ünver. 2024. Artificial intelligence (AI) and human rights: Using AI as a weapon of repression and its impact on human rights.Report for Directorate-General for External Policies, European Parliament. https://www. europarl. europa. eu/Reg-Data/etudes/IDAN/2024/754450/EXPO_IDA (2024) 754450_EN. pdf(2024)

    2024

  77. [77]

    Ibo Van de Poel. 2020. Embedding values in artificial intelligence (AI) systems.Minds and machines30, 3 (2020), 385–409

    2020

  78. [78]

    Ian Vickers, Fergus Lyon, Leandro Sepulveda, and Caitlin McMullin. 2017. Public service innovation and multiple institutional logics: The case of hybrid social enterprise providers of health and wellbeing.Research Policy46, 10 (2017), 1755–1768

    2017

  79. [79]

    Max Weber. 1995. Religious rejections of the world and their directions.The sociology of religion; 1(1995), 80–116

    1995

  80. [80]

    Yifu Yuan, HAO Jianye, Yi Ma, Zibin Dong, Hebin Liang, Jinyi Liu, Zhixin Feng, Kai Zhao, and YAN ZHENG. [n. d.]. Uni-RLHF: Universal Platform and Benchmark Suite for Reinforcement Learning with Diverse Human Feedback. InThe Twelfth International Conference on Learning Representations. Manuscript submitted to ACM

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 3, 2026.