pith. sign in

arxiv: 2606.31755 · v1 · pith:REVMCGY3new · submitted 2026-06-30 · 💻 cs.CY · cs.AI

A Technical Typology of AI Systems in Public Administration

Jonathan Rystr{\o}m , Chris Schmitz , Nathan Davies , Gerhard Hammerschmid , Albert Meijer , Chris Russell This is my paper

Pith reviewed 2026-07-01 02:40 UTC · model grok-4.3

classification 💻 cs.CY cs.AI
keywords AI typologypublic administrationtechnical precisionaccountabilitynon-discriminationAI systems classificationpublic valuesresearch methodology
0
0 comments X

The pith

A five-category typology shows that most public administration research on AI leaves the system underspecified or mismatched to its conclusions.

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

The paper contends that collapsing all AI into one category hides technical differences that shape how systems affect accountability, procedural justice, and non-discrimination in government. It defines five distinct system types and shows they carry separate implications for these public values. Coding 91 recent highly-cited papers reveals that 55 percent leave the studied system underspecified, 31 percent motivate the work with a different system than the one examined, and 41 percent draw conclusions broader than the system supports. The authors supply a short diagnostic guide so future work can locate any studied system inside the typology using only public information.

Core claim

The paper introduces a typology of five AI system categories—hand-coded, glass-box, black-box, general-purpose, and agentic—grouped by their distinct effects on public values, then demonstrates through systematic coding that recent public administration literature routinely fails to specify which category is under study, frequently mismatches the motivating example with the actual system analyzed, and overgeneralizes findings beyond what the examined system can support.

What carries the argument

The five-category typology of AI systems (hand-coded, glass-box, black-box, general-purpose, agentic) that groups types by their differing implications for public values such as accountability and non-discrimination.

If this is right

  • Researchers must supply enough technical detail to place the studied system inside one of the five categories.
  • Studies should avoid motivating the work with one system type while examining a different type.
  • Conclusions must stay within the scope supported by the specific system analyzed.
  • A short list of diagnostic questions answerable from public sources allows non-specialists to classify systems consistently.

Where Pith is reading between the lines

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

  • The typology could be used to design differentiated oversight rules that match regulatory intensity to system type rather than treating all AI alike.
  • Existing meta-reviews of AI in government may require re-examination once papers are reclassified by system category.
  • The same diagnostic questions could be applied to AI studies in adjacent domains such as healthcare or criminal justice to test whether similar imprecision appears.
  • Longitudinal tracking of new papers after the typology is published would show whether specification rates improve.

Load-bearing premise

The five categories are sufficiently distinct in their implications for public values such as accountability and non-discrimination, and the coding of the 91 papers accurately reflects the technical details present in those papers.

What would settle it

An independent re-coding of the same 91 papers by multiple coders that produces materially lower rates of underspecification, motivation-study mismatch, or overgeneralization.

Figures

Figures reproduced from arXiv: 2606.31755 by Albert Meijer, Chris Russell, Chris Schmitz, Gerhard Hammerschmid, Jonathan Rystr{\o}m, Nathan Davies.

Figure 1
Figure 1. Figure 1: Conceptual diagram of three approaches to classifying AI systems. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of Technical Typology of AI Systems. The five system classes (left) are separated by two kinds of relationships. Orange diamonds mark thresholds – qualitative distinctions; ⊂ markers mark subsets, where a system is a special case of the one above. Diagnostic questions distinguish each class from the class above, and can therefore be used as a specification tool, as described in §6.2. Affordance: e… view at source ↗
Figure 3
Figure 3. Figure 3: Underspecification. 55% of papers provide insufficient information to determine which system is empirically studied. The most commonly studied system is black-box systems, with agentic systems completely unstudied. novel ones. Overturned flags lower the reported rate, while un￾counted misses can only raise the true rate (Begg and Greenes, 1983). The reported figures are therefore a conservative esti￾mate o… view at source ↗
Figure 5
Figure 5. Figure 5: Trends in rates. We find no significant changes in any specification category over time. Underspecified Hand-coded Glass-box Black-box General Purpose Generic Total Empirical AI System Classification Total Participation Procedural justice Human rights Quality of governance Responsibility Public Sector Value 44% (n=50) (n=2) (n=4) 26% (n=19) 18% (n=11) 58% (n=12) 41% (n=91) 12% (n=16) (n=2) (n=3) 25% (n=8) … view at source ↗
Figure 6
Figure 6. Figure 6: Overgeneralisation. Heatmap between overgeneralisation for sys￾tem type (X-axis) and public value (Y-axis). Outer cells indicate marginals. In total, 41% of papers overgeneralise. We therefore make two contributions with the aim of improv￾ing the technical precision of future work on AI in public ad￾ministration. First, in §6.1, we highlight three common types of pitfall we find in our analysis – both to i… view at source ↗
read the original abstract

Research on artificial intelligence (AI) in the public sector often treats "AI" as a single category, neglecting technical distinctions between different AI systems. But these distinctions affect how different systems impact core public values like accountability, procedural justice, and non-discrimination. This paper argues that public administration research would benefit from more technical precision on "AI" and makes three contributions to this end. First, we introduce a typology of five categories of AI systems: hand-coded, glass-box, black-box, general-purpose, and agentic systems. We calibrate the typology to public administration by grouping system types by their distinct implications for public values. Second, we evaluate technical precision in recent public administration research about AI by coding 91 highly-cited papers (2019-2025) using our typology. We find widespread imprecision: most papers (55\%) leave the studied system underspecified, 31\% motivate their work with a different system than they study, and 41\% make more general conclusions than the studied system supports. Finally, we give practical recommendations for future research. We highlight common pitfalls to avoid, and suggest that researchers should, at a minimum, provide enough technical detail to locate the studied system in our typology. To this end, we provide a practical guide -- a short set of diagnostic questions answerable from public information and without specialist technical knowledge.

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

3 major / 2 minor

Summary. The paper introduces a five-category typology of AI systems (hand-coded, glass-box, black-box, general-purpose, and agentic) calibrated to implications for public values such as accountability and non-discrimination. It then codes 91 highly-cited public administration papers (2019-2025) and reports that 55% leave the studied system underspecified, 31% motivate their work with a different system than studied, and 41% draw more general conclusions than the studied system supports. The paper concludes with practical recommendations and a short diagnostic guide using publicly available information.

Significance. If the coding results are reliable, the work usefully documents a pattern of technical imprecision in the public administration AI literature and supplies a usable typology plus diagnostic questions that could raise standards for future papers. The provision of a practical, non-specialist guide is a concrete strength.

major comments (3)
  1. [Empirical evaluation / coding exercise] Empirical evaluation section (abstract and methods description): the headline percentages (55% underspecified, 31% mismatched motivation, 41% overgeneralized conclusions) rest on a hand-coding exercise of 91 papers, yet the manuscript reports neither inter-coder agreement statistics nor a public coding table or decision log for borderline cases. Without these, the stability of the reported frequencies cannot be assessed.
  2. [Methods / sample construction] Paper selection paragraph: the criteria used to identify the '91 highly-cited papers (2019-2025)' are not stated (e.g., database, citation threshold, search terms, exclusion rules). This omission prevents evaluation of possible selection bias that could affect the reported proportions.
  3. [Typology introduction] Typology calibration section: the claim that the five categories have 'distinct implications for public values' is asserted but not supported by concrete, paper-specific examples showing how each category differentially affects accountability or non-discrimination; the diagnostic questions are useful but do not substitute for this grounding.
minor comments (2)
  1. [Typology section] The abstract states the three contributions clearly, but the manuscript would benefit from an explicit table mapping each of the five categories to the specific public-value implications discussed.
  2. [Abstract] Minor: the phrase 'without specialist technical knowledge' in the abstract could be clarified by noting the exact public information sources the diagnostic questions rely on.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback, which highlights opportunities to improve transparency and grounding in the manuscript. We address each major comment below and will make revisions to strengthen the empirical and methodological sections.

read point-by-point responses

  1. Referee: Empirical evaluation section (abstract and methods description): the headline percentages (55% underspecified, 31% mismatched motivation, 41% overgeneralized conclusions) rest on a hand-coding exercise of 91 papers, yet the manuscript reports neither inter-coder agreement statistics nor a public coding table or decision log for borderline cases. Without these, the stability of the reported frequencies cannot be assessed.

    Authors: We agree that inter-coder agreement statistics and access to the coding decisions would strengthen the empirical claims. The coding followed an iteratively developed codebook by the author team. In revision we will add a methods subsection reporting inter-coder reliability (e.g., Cohen’s kappa) and include a supplementary coding table with decision notes for borderline cases, allowing readers to evaluate frequency stability. revision: yes

  2. Referee: Paper selection paragraph: the criteria used to identify the '91 highly-cited papers (2019-2025)' are not stated (e.g., database, citation threshold, search terms, exclusion rules). This omission prevents evaluation of possible selection bias that could affect the reported proportions.

    Authors: The selection criteria are only summarized at a high level in the current text. We will expand the methods section with the precise protocol: database(s) queried, citation threshold definition, exact search terms, date range application, and all exclusion rules. This addition will permit direct assessment of selection bias. revision: yes

  3. Referee: Typology introduction section: the claim that the five categories have 'distinct implications for public values' is asserted but not supported by concrete, paper-specific examples showing how each category differentially affects accountability or non-discrimination; the diagnostic questions are useful but do not substitute for this grounding.

    Authors: The typology section already groups categories according to their implications for accountability and non-discrimination, drawing on prior AI-ethics literature. To address the request for concrete grounding we will insert brief, paper-specific illustrations (drawn from the coded corpus or canonical cases) showing differential effects for each of the five types. The diagnostic questions will remain as a practical supplement. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical coding of external literature

full rationale

The paper defines a new five-category typology and applies it to code 91 external papers, yielding the reported percentages as direct observational results. No equations, fitted parameters, or self-referential derivations exist; the central claims are measurements obtained by applying the authors' diagnostic questions to the sampled literature rather than any reduction of outputs to the typology inputs by construction. Self-citations, if present, are not load-bearing for the frequency statistics.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The typology introduces five new categories whose boundaries and value implications are asserted rather than derived from prior data. The coding exercise assumes that public information in the 91 papers is sufficient to assign each paper to a category.

axioms (1)
  • domain assumption The five system types produce meaningfully different effects on accountability, procedural justice, and non-discrimination.
    This premise is required to justify grouping systems by public-value implications; it is stated in the abstract but not derived.
invented entities (1)
  • hand-coded, glass-box, black-box, general-purpose, and agentic system categories no independent evidence
    purpose: To classify AI systems according to inspectability and autonomy for public administration analysis.
    These are newly defined groupings introduced by the paper; no independent empirical test of the categories is described in the abstract.

pith-pipeline@v0.9.1-grok · 5784 in / 1376 out tokens · 31412 ms · 2026-07-01T02:40:25.797730+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

257 extracted references · 186 canonical work pages · 4 internal anchors

  1. [14]

    , author Gebru, T

    author Bender, E.M. , author Gebru, T. , author McMillan-Major , A. , author Shmitchell, S. , year 2021 . title On the dangers of stochastic parrots: Can language models Be too big? , in: booktitle Proceedings of the 2021 ACM Conference on Fairness , Accountability , and Transparency , pp. pages 610--623

    2021

  2. [16]

    , author Hudson, D.A

    author Bommasani, R. , author Hudson, D.A. , author Adeli, E. , author Altman, R. , author Arora, S. , author von Arx , S. , author Bernstein, M.S. , author Bohg, J. , author Bosselut, A. , author Brunskill, E. , year 2021 . title On the opportunities and risks of foundation models

    2021

  3. [22]

    , year 2023

    author Brown, I. , year 2023 . title Allocating Accountability in AI Supply Chains . type Technical Report . Ada Lovelace Institute. https://www.adalovelaceinstitute.org/resource/ai-supply-chains/

    2023

  4. [23]

    , author Mann, B

    author Brown, T.B. , author Mann, B. , author Ryder, N. , author Subbiah, M. , author Kaplan, J. , author Dhariwal, P. , author Neelakantan, A. , author Shyam, P. , author Sastry, G. , , author Amodei, D. , year 2020 . title Language models are few-shot learners , in: booktitle Proceedings of the 34th International Conference on Neural Information Process...

    work page doi:10.5555/3495724.3495883 2020

  5. [31]

    , year 2008

    author Citron, D.K. , year 2008 . title Technological due process . journal Washington University Law Review volume 85 , pages 1249--1313

    2008

  6. [33]

    , author Gaebler, J.D

    author Corbett-Davies , S. , author Gaebler, J.D. , author Nilforoshan, H. , author Shroff, R. , author Goel, S. , year 2023 . title The measure and mismeasure of fairness . journal Journal of Machine Learning Research volume 24

    2023

  7. [35]

    , year 2018

    author Dastin, J. , year 2018 . title Insight - amazon scraps secret AI recruiting tool that showed bias against women . journal Reuters https://www.reuters.com/article/world/insight-amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK0AG/

    2018

  8. [42]

    , author Klein, L.F

    author D'ignazio, C. , author Klein, L.F. , year 2023 . title Data Feminism . publisher MIT press . https://books.google.com/books?hl=en&lr=&id=rHOdEAAAQBAJ&oi=fnd&pg=PR9&dq=data+feminism+klein&ots=mXwrpceedi&sig=MI7_gJ8l1kdTlo-GEUXh2kRrskk

    2023

  9. [50]

    , year 2015

    author Fox, J. , year 2015 . title Applied Regression Analysis and Generalized Linear Models . publisher Sage Publications

    2015

  10. [53]

    , author Leyer, M

    author Gesk, T.S. , author Leyer, M. , year 2022 . title Artificial intelligence in public services: When and why citizens accept its usage . journal Government Information Quarterly volume 39 , pages 101704 . https://www.sciencedirect.com/science/article/pii/S0740624X22000375, :10.1016/j.giq.2022.101704

    work page doi:10.1016/j.giq.2022.101704 2022

  11. [54]

    , year 1979

    author Gibson, J.J. , year 1979 . title The Ecological Approach to Visual Perception . publisher Houghton Mifflin Comp , address Boston, Mass

    1979

  12. [55]

    , author Bengio, Y

    author Goodfellow, I. , author Bengio, Y. , author Courville, A. , author Bengio, Y. , year 2016 . title Deep Learning . volume volume 1 . publisher MIT press Cambridge

    2016

  13. [56]

    , year 2022

    author Green, B. , year 2022 . title The flaws of policies requiring human oversight of government algorithms . journal Computer Law & Security Review volume 45 , pages 105681 . https://linkinghub.elsevier.com/retrieve/pii/S0267364922000292, :10.1016/j.clsr.2022.105681

    work page doi:10.1016/j.clsr.2022.105681 2022

  14. [57]

    , author Chen, Y

    author Green, B. , author Chen, Y. , year 2021 . title Algorithmic Risk Assessments Can Alter Human Decision-Making Processes in High-Stakes Government Contexts . journal Proceedings of the ACM on Human-Computer Interaction volume 5 , pages 1--33 . https://dl.acm.org/doi/10.1145/3479562, :10.1145/3479562

    work page doi:10.1145/3479562 2021

  15. [58]

    , author Meijer, A

    author Grimmelikhuijsen, S. , author Meijer, A. , year 2022 . title Legitimacy of algorithmic decision-making: Six threats and the need for a calibrated institutional response . journal Perspectives on Public Management and Governance volume 5 , pages 232--242 . https://doi.org/10.1093/ppmgov/gvac008, :10.1093/ppmgov/gvac008

    work page doi:10.1093/ppmgov/gvac008 2022

  16. [59]

    , author Haleem, N

    author Gstrein, O.J. , author Haleem, N. , author Zwitter, A. , year 2024 . title General-purpose AI regulation and the European union AI act . journal Internet Policy Review volume 13 . https://policyreview.info/articles/analysis/general-purpose-ai-regulation-and-ai-act, :10.14763/2024.3.1790

    work page doi:10.14763/2024.3.1790 2024

  17. [60]

    , author Namey, E

    author Guest, G. , author Namey, E. , author Chen, M. , year 2020 . title A simple method to assess and report thematic saturation in qualitative research . journal PLOS ONE volume 15 , pages e0232076 . https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0232076, :10.1371/journal.pone.0232076

    work page doi:10.1371/journal.pone.0232076 2020

  18. [61]

    , author Sage, M

    author Hall, S.F. , author Sage, M. , author Scott, C.F. , author Joseph, K. , year 2024 . title A systematic review of sophisticated predictive and prescriptive analytics in child welfare: Accuracy, equity, and bias . journal Child and Adolescent Social Work Journal volume 41 , pages 831--847 . https://doi.org/10.1007/s10560-023-00931-2, :10.1007/s10560-...

    work page doi:10.1007/s10560-023-00931-2 2024

  19. [62]

    , author Bright, J

    author Hashem, Y. , author Bright, J. , author Chakraborty, S. , year 2025 . title Mapping the potential: Generative AI and public sector work https://apo.org.au/node/330966

    2025

  20. [63]

    , author Bailur, S

    author Heeks, R. , author Bailur, S. , year 2007 . title Analyzing e-government research: Perspectives, philosophies, theories, methods, and practice . journal Government Information Quarterly volume 24 , pages 243--265 . https://www.sciencedirect.com/science/article/pii/S0740624X06000943, :10.1016/j.giq.2006.06.005

    work page doi:10.1016/j.giq.2006.06.005 2007

  21. [64]

    , author Kilian, M

    author Ilves, L. , author Kilian, M. , author Parazzoli, S.M. , author Peixoto, T.C. , author Velsberg, O. , year 2025 . title The Agentic State: Rethinking Government for the Era of Agentic AI . type Technical Report . Global Government Technology Centre Berlin and The World Bank

    2025

  22. [65]

    , author Brous, P

    author Janssen, M. , author Brous, P. , author Estevez, E. , author Barbosa, L.S. , author Janowski, T. , year 2020 . title Data governance: Organizing data for trustworthy artificial intelligence . journal Government Information Quarterly volume 37 , pages 101493 . https://www.sciencedirect.com/science/article/pii/S0740624X20302719, :10.1016/j.giq.2020.101493

    work page doi:10.1016/j.giq.2020.101493 2020

  23. [66]

    , author Zhang, H

    author Jo, J. , author Zhang, H. , author Cai, J. , author Goyal, N. , year 2025 . title AI trust reshaping administrative burdens: Understanding trust-burden dynamics in LLM-assisted benefits systems , in: booktitle Proceedings of the 2025 ACM Conference on Fairness , Accountability , and Transparency , publisher Association for Computing Machinery , add...

    work page doi:10.1145/3715275.3732077 2025

  24. [67]

    , author Meng, Q

    author Ju, J. , author Meng, Q. , author Sun, F. , author Liu, L. , author Singh, S. , year 2023 . title Citizen preferences and government chatbot social characteristics: Evidence from a discrete choice experiment . journal Government Information Quarterly volume 40 , pages 101785 . https://www.sciencedirect.com/science/article/pii/S0740624X22001216, :10...

    work page doi:10.1016/j.giq.2022.101785 2023

  25. [68]

    , author Gabriel, I

    author Kasirzadeh, A. , author Gabriel, I. , year 2025 . title Characterizing AI Agents for Alignment and Governance . http://arxiv.org/abs/2504.21848, :10.48550/arXiv.2504.21848, http://arxiv.org/abs/2504.21848 arXiv:2504.21848

    work page doi:10.48550/arxiv.2504.21848 2025

  26. [69]

    , year 2024

    author Keppeler, F. , year 2024 . title No thanks, dear AI ! Understanding the effects of disclosure and deployment of artificial intelligence in public sector recruitment . journal Journal of Public Administration Research and Theory volume 34 , pages 39--52 . https://academic.oup.com/jpart/article/34/1/39/7174960, :10.1093/jopart/muad009

    work page doi:10.1093/jopart/muad009 2024

  27. [70]

    , author Borchert, J

    author Keppeler, F. , author Borchert, J. , author Pedersen, M.J. , author Lehmann Nielsen, V. , year 2025 . title How ensembling AI and public managers improves decision-making . journal Journal of Public Administration Research and Theory volume 35 , pages 261--276 . https://academic.oup.com/jpart/article/35/3/261/8116003, :10.1093/jopart/muaf009

    work page doi:10.1093/jopart/muaf009 2025

  28. [71]

    , author Shoaib, A

    author Khan, M.S. , author Shoaib, A. , author Arledge, E. , year 2024 . title How to promote AI in the US federal government: Insights from policy process frameworks . journal Government Information Quarterly volume 41 , pages 101908 . https://www.sciencedirect.com/science/article/pii/S0740624X23001089, :10.1016/j.giq.2023.101908

    work page doi:10.1016/j.giq.2023.101908 2024

  29. [72]

    , author Wenzelburger, G

    author K \"o nig, P.D. , author Wenzelburger, G. , year 2020 . title Opportunity for renewal or disruptive force? How artificial intelligence alters democratic politics . journal Government Information Quarterly volume 37 , pages 101489 . https://www.sciencedirect.com/science/article/pii/S0740624X1930245X, :10.1016/j.giq.2020.101489

    work page doi:10.1016/j.giq.2020.101489 2020

  30. [73]

    , year 2019

    author Krippendorff, K. , year 2019 . title Content Analysis: An Introduction to Its Methodology . publisher SAGE Publications, Inc. https://methods.sagepub.com/book/mono/content-analysis-4e/toc, :10.4135/9781071878781

    work page doi:10.4135/9781071878781 2019

  31. [74]

    , author Sutskever, I

    author Krizhevsky, A. , author Sutskever, I. , author Hinton, G.E. , year 2012 . title ImageNet classification with deep convolutional neural networks , in: booktitle Advances in Neural Information Processing Systems , publisher Curran Associates, Inc. https://papers.nips.cc/paper_files/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html

    2012

  32. [75]

    , author Huey, J

    author Kroll, J.A. , author Huey, J. , author Barocas, S. , author Felten, E.W. , author Reidenberg, J.R. , author Robinson, D.G. , author Yu, H. , year 2017 . title Accountable algorithms . journal University of Pennsylvania Law Review volume 165 , pages 633

    2017

  33. [76]

    , author Wachter, S

    author Laux, J. , author Wachter, S. , author Mittelstadt, B. , year 2024 . title Trustworthy artificial intelligence and the european union AI act: On the conflation of trustworthiness and acceptability of risk . journal Regulation & Governance volume 18 , pages 3--32 . https://onlinelibrary.wiley.com/doi/abs/10.1111/rego.12512, :10.1111/rego.12512

    work page doi:10.1111/rego.12512 2024

  34. [77]

    , author Cui, I

    author Lawrence, C. , author Cui, I. , author Ho, D. , year 2023 . title The bureaucratic challenge to AI governance: An empirical assessment of implementation at U . S . federal agencies , in: booktitle Proceedings of the 2023 AAAI / ACM Conference on AI , Ethics , and Society , publisher Association for Computing Machinery , address New York, NY, USA . ...

    work page doi:10.1145/3600211.3604701 2023

  35. [78]

    Oxford University Press, New York, doi:10.1093/oso/9780198537793.001.0001

    author Lazar, S. , year 2024 . title Legitimacy, Authority , and Democratic Duties of Explanation , in: editor Sobel, D. , editor Wall, S. (Eds.), booktitle Oxford Studies in Political Philosophy Volume 10 . edition 1 ed.. publisher Oxford University Press, Oxford , pp. pages 28--56 . https://academic.oup.com/book/56337/chapter/445461225, :10.1093/oso/978...

    work page doi:10.1093/oso/9780198909460.003.0002 2024

  36. [79]

    , year 2024

    author Lecher, C. , year 2024 . title NYC 's AI Chatbot Tells Businesses to Break the Law . journal The Markup https://themarkup.org/artificial-intelligence/2024/03/29/nycs-ai-chatbot-tells-businesses-to-break-the-law

    2024

  37. [80]

    , year 2011

    author Leonardi, P.M. , year 2011 . title When Flexible Routines Meet Flexible Technologies : Affordance , Constraint , and the Imbrication of Human and Material Agencies1 . journal MIS Quarterly volume 35 , pages 147--167 . https://doi.org/10.2307/23043493, :10.2307/23043493

    work page doi:10.2307/23043493 2011

  38. [81]

    , author Delen, D

    author Lima, M.S.M. , author Delen, D. , year 2020 . title Predicting and explaining corruption across countries: A machine learning approach . journal Government Information Quarterly volume 37 , pages 101407 . https://www.sciencedirect.com/science/article/pii/S0740624X19302473, :10.1016/j.giq.2019.101407

    work page doi:10.1016/j.giq.2019.101407 2020

  39. [82]

    , year 2018

    author Lipton, Z.C. , year 2018 . title The mythos of model interpretability . journal Communications of The Acm volume 61 , pages 36--43 . https://doi.org/10.1145/3233231, :10.1145/3233231

    work page doi:10.1145/3233231 2018

  40. [83]

    , author Ashok, M

    author Madan, R. , author Ashok, M. , year 2023 . title AI adoption and diffusion in public administration: A systematic literature review and future research agenda . journal Government Information Quarterly volume 40 , pages 101774 . https://www.sciencedirect.com/science/article/pii/S0740624X22001101, :10.1016/j.giq.2022.101774

    work page doi:10.1016/j.giq.2022.101774 2023

  41. [84]

    , author Markus, M.L

    author Majchrzak, A. , author Markus, M.L. , year 2013 . title Technology Affordances and Constraints Theory (of MIS ) https://doi.org/10.4135/9781452276090.n282, :10.4135/9781452276090.n282

    work page doi:10.4135/9781452276090.n282 2013

  42. [85]

    , author Nili, A

    author Makasi, T. , author Nili, A. , author Desouza, K.C. , author Tate, M. , year 2022 . title A typology of chatbots in public service delivery . journal IEEE Software volume 39 , pages 58--66 . https://ieeexplore.ieee.org/document/9405373/, :10.1109/MS.2021.3073674

    work page doi:10.1109/ms.2021.3073674 2022

  43. [86]

    , author Duarte, A.V

    author Marques, J.D. , author Duarte, A.V. , author de Carvalho , A.M.M. , author Rocha, G. , author Martins, B. , author Oliveira, A.L. , year 2025 . title Leveraging LLMs to streamline the review of public funding applications , in: editor Potdar, S. , editor Rojas-Barahona , L. , editor Montella, S. (Eds.), booktitle Proceedings of the 2025 Conference ...

    work page doi:10.18653/v1/2025.emnlp-industry.143 2025

  44. [87]

    , year 2004

    author Matthias, A. , year 2004 . title The responsibility gap: Ascribing responsibility for the actions of learning automata . journal Ethics and Information Technology volume 6 , pages 175--183 . https://doi.org/10.1007/s10676-004-3422-1, :10.1007/s10676-004-3422-1

    work page doi:10.1007/s10676-004-3422-1 2004

  45. [88]

    , author Kang, J.S

    author Mayne, H. , author Kang, J.S. , author Gould, D. , author Ramchandran, K. , author Mahdi, A. , author Siegel, N.Y. , year 2026 . title A positive case for faithfulness: LLM self-explanations help predict model behavior . http://arxiv.org/abs/2602.02639, :10.48550/arXiv.2602.02639, http://arxiv.org/abs/2602.02639 arXiv:2602.02639

    work page doi:10.48550/arxiv.2602.02639 2026

  46. [89]

    , year 2015

    author Mayring, P. , year 2015 . title Qualitative content analysis: Theoretical background and procedures , in: editor Bikner-Ahsbahs , A. , editor Knipping, C. , editor Presmeg, N. (Eds.), booktitle Approaches to Qualitative Research in Mathematics Education: Examples of Methodology and Methods . publisher Springer Netherlands , address Dordrecht , pp. ...

    work page doi:10.1007/978-94-017-9181-6_13 2015

  47. [90]

    , author Healy, J

    author McInnes, L. , author Healy, J. , author Melville, J. , year 2020 . title UMAP : Uniform manifold approximation and projection for dimension reduction . http://arxiv.org/abs/1802.03426, http://arxiv.org/abs/1802.03426 arXiv:1802.03426

    Pith/arXiv arXiv 2020

  48. [91]

    , author Lorenz, L

    author Meijer, A. , author Lorenz, L. , author Wessels, M. , year 2021 . title Algorithmization of bureaucratic organizations: Using a practice lens to study how context shapes predictive policing systems . journal Public Administration Review volume 81 , pages 837--846 . https://onlinelibrary.wiley.com/doi/abs/10.1111/puar.13391, :10.1111/puar.13391

    work page doi:10.1111/puar.13391 2021

  49. [92]

    , author Huberman, A.M

    author Miles, M.B. , author Huberman, A.M. , author Saldana, J. , year 2014 . title Qualitative Data Analysis: A Methods Sourcebook . publisher SAGE Publications, Inc , address Los Angeles London New Delhi Singapore Washington DC

    2014

  50. [93]

    , year 2013

    author Mitchell, T.M. , year 2013 . title Machine Learning . McGraw-Hill Series in Computer Science . edition nachdr. ed., publisher McGraw-Hill , address New York

    2013

  51. [94]

    , author Russell, C

    author Mittelstadt, B. , author Russell, C. , author Wachter, S. , year 2019 . title Explaining explanations in AI , in: booktitle Proceedings of the Conference on Fairness , Accountability , and Transparency , publisher Association for Computing Machinery , address New York, NY, USA . pp. pages 279--288 . https://doi.org/10.1145/3287560.3287574, :10.1145...

    work page doi:10.1145/3287560.3287574 2019

  52. [95]

    , author Schroeder, R

    author M \"o kander, J. , author Schroeder, R. , year 2024 . title Artificial intelligence, rationalization, and the limits of control in the public sector: The case of tax policy optimization . journal Social Science Computer Review volume 42 , pages 1359--1378 . https://doi.org/10.1177/08944393241235175, :10.1177/08944393241235175

    work page doi:10.1177/08944393241235175 2024

  53. [96]

    , author Schuett, J

    author M \"o kander, J. , author Schuett, J. , author Kirk, H.R. , author Floridi, L. , year 2024 . title Auditing large language models: A three-layered approach . journal AI and Ethics volume 4 , pages 1085--1115 . https://doi.org/10.1007/s43681-023-00289-2, :10.1007/s43681-023-00289-2

    work page doi:10.1007/s43681-023-00289-2 2024

  54. [97]

    , author Chung, P

    author Mowbray, A. , author Chung, P. , author Greenleaf, G. , year 2023 . title Explainable AI ( XAI ) in Rules as Code ( RaC ): The DataLex approach . journal Computer Law & Security Review volume 48 , pages 105771 . https://linkinghub.elsevier.com/retrieve/pii/S0267364922001145, :10.1016/j.clsr.2022.105771

    work page doi:10.1016/j.clsr.2022.105771 2023

  55. [98]

    , author Guirguis, K

    author Neumann, O. , author Guirguis, K. , author Steiner, R. , year 2024 . title Exploring artificial intelligence adoption in public organizations: A comparative case study . journal Public Management Review volume 26 , pages 114--141 . https://www.tandfonline.com/doi/full/10.1080/14719037.2022.2048685, :10.1080/14719037.2022.2048685

    work page doi:10.1080/14719037.2022.2048685 2024

  56. [99]

    , year 2025

    author Nguyen-Trung , K. , year 2025 . title ChatGPT in thematic analysis: Can AI become a research assistant in qualitative research? journal Quality & Quantity volume 59 , pages 4945--4978 . https://doi.org/10.1007/s11135-025-02165-z, :10.1007/s11135-025-02165-z

    work page doi:10.1007/s11135-025-02165-z 2025

  57. [100]

    , year 1996

    author Nissenbaum, H. , year 1996 . title Accountability in a computerized society . journal Science and Engineering Ethics volume 2 , pages 25--42 . https://doi.org/10.1007/BF02639315, :10.1007/BF02639315

    work page doi:10.1007/bf02639315 1996

  58. [101]

    , author Medaglia, R

    author van Noordt, C. , author Medaglia, R. , author Tangi, L. , year 2025 . title Policy initiatives for artificial intelligence-enabled government: An analysis of national strategies in Europe . journal Public Policy and Administration volume 40 , pages 215--253 . https://research.cbs.dk/en/publications/policy-initiatives-for-artificial-intelligence-ena...

    work page doi:10.1177/09520767231198411 2025

  59. [102]

    , author Joffe, H

    author O'Connor, C. , author Joffe, H. , year 2020 . title Intercoder reliability in qualitative research: Debates and practical guidelines . journal International Journal of Qualitative Methods volume 19 , pages 1609406919899220 . https://journals.sagepub.com/doi/10.1177/1609406919899220, :10.1177/1609406919899220

    work page doi:10.1177/1609406919899220 2020

  60. [103]

    title OECD Framework for the Classification of AI Systems

    author OECD , year 2022 . title OECD Framework for the Classification of AI Systems . type OECD Digital Economy Papers number 323 . OECD. https://www.oecd.org/en/publications/oecd-framework-for-the-classification-of-ai-systems_cb6d9eca-en.html, :10.1787/cb6d9eca-en

    work page doi:10.1787/cb6d9eca-en 2022

  61. [104]

    title ChatGPT : Optimizing language models for dialogue

    author OpenAI , year 2022 . title ChatGPT : Optimizing language models for dialogue . https://openai.com/blog/chatgpt/

    2022

  62. [105]

    , author McKenzie, J.E

    author Page, M.J. , author McKenzie, J.E. , author Bossuyt, P.M. , author Boutron, I. , author Hoffmann, T.C. , author Mulrow, C.D. , author Shamseer, L. , author Tetzlaff, J.M. , author Akl, E.A. , , author Moher, D. , year 2021 . title The PRISMA 2020 statement: An updated guideline for reporting systematic reviews . journal BMJ , pages n71 https://www....

    work page doi:10.1136/bmj.n71 2021

  63. [106]

    , author Widlak, A.C

    author Peeters, R. , author Widlak, A.C. , year 2023 . title Administrative exclusion in the infrastructure-level bureaucracy: The case of the dutch daycare benefit scandal . journal Public Administration Review volume 83 , pages 863--877 . https://onlinelibrary.wiley.com/doi/10.1111/puar.13615, :10.1111/puar.13615

    work page doi:10.1111/puar.13615 2023

  64. [107]

    , author Calinescu, R

    author Porter, Z. , author Calinescu, R. , author Lim, E. , author Hodge, V. , author Ryan, P. , author Burton, S. , author Habli, I. , author Lawton, T. , author McDermid, J. , , author Zou, J. , year 2025 . title INSYTE : A Classification Framework for Traditional to Agentic AI Systems . journal ACM Transactions on Autonomous and Adaptive Systems volume...

    work page doi:10.1145/3760424 2025

  65. [108]

    OpenAlex: A fully-open index of scholarly works, authors, venues, institutions, and concepts

    author Priem, J. , author Piwowar, H. , author Orr, R. , year 2022 . title OpenAlex : A fully-open index of scholarly works, authors, venues, institutions, and concepts . http://arxiv.org/abs/2205.01833, :10.48550/arXiv.2205.01833, http://arxiv.org/abs/2205.01833 arXiv:2205.01833

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2205.01833 2022

  66. [109]

    , author He, Z

    author Qiu, T. , author He, Z. , author Chugh, T. , author Kleiman-Weiner , M. , year 2025 . title The lock-in hypothesis: Stagnation by algorithm , in: booktitle Forty-Second International Conference on Machine Learning . https://openreview.net/forum?id=mE1M626qOo

    2025

  67. [110]

    , author Barocas, S

    author Raghavan, M. , author Barocas, S. , author Kleinberg, J. , author Levy, K. , year 2020 . title Mitigating bias in algorithmic hiring: Evaluating claims and practices , in: booktitle Proceedings of the 2020 Conference on Fairness , Accountability , and Transparency , publisher Association for Computing Machinery , address New York, NY, USA . pp. pag...

    work page doi:10.1145/3351095.3372828 2020

  68. [111]

    , author Frtunikj, J

    author Rao, Q. , author Frtunikj, J. , year 2018 . title Deep learning for self-driving cars: Chances and challenges , in: booktitle Proceedings of the 1st International Workshop on Software Engineering for AI in Autonomous Systems , publisher Association for Computing Machinery , address New York, NY, USA . pp. pages 35--38 . https://dl.acm.org/doi/10.11...

    work page doi:10.1145/3194085.3194087 2018

  69. [112]

    , year 2020

    author Rezende, I.N. , year 2020 . title Facial recognition in police hands: Assessing the `clearview case' from a European perspective . journal New Journal of European Criminal Law volume 11 , pages 375--389 . https://doi.org/10.1177/2032284420948161, :10.1177/2032284420948161

    work page doi:10.1177/2032284420948161 2020

  70. [113]

    , year 2026

    author Robinson, N. , year 2026 . title Open to open-source AI ? Navigating AI model choice in public sector agencies . journal Government Information Quarterly volume 43 , pages 102133 . https://www.sciencedirect.com/science/article/pii/S0740624X26000304, :10.1016/j.giq.2026.102133

    work page doi:10.1016/j.giq.2026.102133 2026

  71. [114]

    , author Hansen, M.B

    author Roehl, U.B.U. , author Hansen, M.B. , year 2024 . title Automated, administrative decision-making and good governance: Synergies, trade-offs, and limits . journal Public Administration Review volume 84 , pages 1184--1199 . https://onlinelibrary.wiley.com/doi/abs/10.1111/puar.13799, :10.1111/puar.13799

    work page doi:10.1111/puar.13799 2024

  72. [115]

    , year 2019

    author Rudin, C. , year 2019 . title Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead . journal Nature Machine Intelligence volume 1 , pages 206--215 . https://www.nature.com/articles/s42256-019-0048-x, :10.1038/s42256-019-0048-x

    work page doi:10.1038/s42256-019-0048-x 2019

  73. [116]

    , author Fu, Z

    author Rystr m, J. , author Fu, Z. , author Russell, C. , year 2026 a. title OxEnsemble : Fair ensembles for low-data classification , in: booktitle Medical Imaging with Deep Learning , publisher PMLR . https://openreview.net/forum?id=DuRUqZgwk8

    2026

  74. [117]

    , author Schmitz, C

    author Rystr m, J. , author Schmitz, C. , author Korgul, K. , author Batzner, J. , author Russell, C. , year 2026 b. title Agent benchmarks fail public sector requirements , in: booktitle IASEAI 2026 , publisher arXiv . http://arxiv.org/abs/2601.20617, :10.48550/arXiv.2601.20617, http://arxiv.org/abs/2601.20617 arXiv:2601.20617

    work page doi:10.48550/arxiv.2601.20617 2026

  75. [118]

    , year 2025

    author Salda \ n a, J. , year 2025 . title The Coding Manual for Qualitative Researchers . edition 5e ed., publisher Sage , address London Thousand Oaks, California

    2025

  76. [119]

    , author Horta Ribeiro, M

    author Salvi, F. , author Horta Ribeiro, M. , author Gallotti, R. , author West, R. , year 2025 . title On the conversational persuasiveness of GPT-4 . journal Nature Human Behaviour volume 9 , pages 1645--1653 . https://www.nature.com/articles/s41562-025-02194-6, :10.1038/s41562-025-02194-6

    work page doi:10.1038/s41562-025-02194-6 2025

  77. [120]

    , author Hamilton, K

    author Sandvig, C. , author Hamilton, K. , author Karahalios, K. , author Langbort, C. , year 2014 . title Auditing algorithms: Research methods for detecting discrimination on internet platforms . journal Data and Discrimination: Converting Critical Concerns into Productive Inquiry volume 22 , pages 4349--4357

    2014

  78. [121]

    , author Zhu, A

    author Sansone, D. , author Zhu, A. , year 2023 . title Using machine learning to create an early warning system for welfare recipients* . journal Oxford Bulletin of Economics and Statistics volume 85 , pages 959--992 . https://onlinelibrary.wiley.com/doi/10.1111/obes.12550, :10.1111/obes.12550

    work page doi:10.1111/obes.12550 2023

  79. [122]

    , author Bryson, J

    author Schmitz, C. , author Bryson, J. , year 2025 . title A moral agency framework for legitimate integration of AI in bureaucracies (extended abstract) . journal Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society volume 8 , pages 2292--2293 . https://ojs.aaai.org/index.php/AIES/article/view/36714, :10.1609/aies.v8i3.36714

    work page doi:10.1609/aies.v8i3.36714 2025

  80. [123]

    , author Rystr m, J

    author Schmitz, C. , author Rystr m, J. , author Batzner, J. , year 2025 . title Oversight structures for agentic AI in public-sector organizations , in: editor Kamalloo, E. , editor Gontier, N. , editor Lu, X.H. , editor Dziri, N. , editor Murty, S. , editor Lacoste, A. (Eds.), booktitle Proceedings of the 1st Workshop for Research on Agent Language Mode...

    2025

Showing first 80 references.