pith. sign in

arxiv: 2605.20081 · v1 · pith:B23YY3LTnew · submitted 2026-05-19 · 💻 cs.CY

Bridging the Disciplinary Gap in Explainable AI: From Abstract Desiderata to Concrete Tasks

Hanwei Zhang , Jingwen Wang , Holger Hermanns This is my paper

Pith reviewed 2026-05-20 03:37 UTC · model grok-4.3

classification 💻 cs.CY
keywords Explainable AIDesiderataDependency structuresTaxonomyConcrete tasksInterdisciplinary XAIBenchmarking
0
0 comments X

The pith

XAI desiderata form dependency structures where higher goals rely on foundational properties, allowing translation into concrete benchmarkable tasks.

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

Researchers across disciplines often expect explainable AI to meet many abstract desiderata at once, yet these remain underspecified and lead to fragmented efforts. The paper establishes that many desiderata are interdependent rather than standalone, with higher-level aims such as trust and accountability depending on more basic properties like faithfulness and robustness. Some desiderata are multi-faceted and only make sense inside these dependency structures. To address the resulting disciplinary gaps, the work supplies a three-axis taxonomy and a three-step framework that scope subsets of dependencies into specific, solvable XAI tasks instead of broad unattainable targets.

Core claim

Many XAI desiderata are not independent but form dependency structures in which higher-level goals rely on foundational properties, and a three-axis taxonomy together with a three-step framework can systematically derive well-scoped, benchmarkable tasks from abstract desiderata by clarifying dependencies, scoping feasibility, and delimiting the design space.

What carries the argument

The three-axis taxonomy of target, functional role, and mode of justification, applied inside a three-step framework that identifies dependency structures among desiderata and converts selected subsets into concrete XAI tasks.

If this is right

  • Higher-level goals such as trust and accountability become achievable once foundational properties like faithfulness and robustness are secured first.
  • Complex research questions decompose into smaller benchmarkable units that can be addressed independently.
  • Clarification of desiderata and delimitation of the design space become repeatable steps rather than ad-hoc judgments.
  • Evaluation of XAI methods improves because each task is tied to a specific part of a dependency structure rather than an open-ended wish list.

Where Pith is reading between the lines

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

  • The framework might be tested on regulatory or deployment settings to check whether derived tasks align with legal accountability requirements.
  • Similar dependency mapping could be applied to neighboring areas such as AI fairness or safety where abstract goals also fragment across communities.
  • Empirical studies could measure whether teams using the taxonomy produce more consistent task definitions than teams without it.

Load-bearing premise

That the three-axis taxonomy and three-step framework can identify and scope dependency structures among XAI desiderata without omitting critical context-dependent factors or creating new incompatibilities across disciplines.

What would settle it

A demonstration that tasks produced by the framework fail to support the higher-level desiderata they are meant to enable, or that applying the framework still yields incompatible operationalizations when researchers from different fields interpret the same desideratum.

Figures

Figures reproduced from arXiv: 2605.20081 by Hanwei Zhang, Holger Hermanns, Jingwen Wang.

Figure 1
Figure 1. Figure 1: Dependency structure of desiderata for human oversight in risk prevention. [PITH_FULL_IMAGE:figures/full_fig_p011_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Dependency structure of desiderata for Legal Audit. [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
read the original abstract

Explainable AI (XAI) is often criticized for failing to satisfy broad desiderata (e.g., fairness, accountability) and for limited practical value to stakeholders. This challenge partly arises because researchers across disciplines prioritize different sets of desiderata that remain underspecified and context-dependent, yet expect XAI to satisfy them simultaneously, resulting in fragmented and sometimes incompatible operationalizations. We argue that many desiderata are not independent, but instead form dependency structures in which higher-level goals (\emph{e.g.}, trust, accountability) rely on more foundational properties (\emph{e.g.}, faithfulness, robustness). Some desiderata are multi-faceted and are best understood within these structures. In particular, instead of addressing all desiderata at once, we focus on subsets of dependency structures and translate them into concrete XAI tasks, thereby decomposing research questions into benchmarkable and solvable units. To this end, we propose a three-axis taxonomy (\emph{target}, \emph{functional role}, and \emph{mode of justification}) and a three-step framework for deriving well-scoped, benchmarkable XAI tasks. Our approach builds on a systematic literature review and conceptual analysis, and supports clarifying desiderata, identifying dependencies, scoping feasibility, and delimiting the design space to derive concrete XAI tasks from abstract desiderata. We illustrate its utility through two explanatory cases, showing how the taxonomy and framework guide systematic task design and evaluation in XAI. {\color{red}{This is a preprint of a paper that will appear in AISoLA 2026.}}

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

0 major / 2 minor

Summary. The manuscript claims that XAI desiderata are not independent but form dependency structures, with higher-level goals such as trust and accountability relying on foundational properties like faithfulness and robustness. It proposes a three-axis taxonomy (target, functional role, and mode of justification) together with a three-step framework to scope subsets of these structures and translate them into concrete, benchmarkable XAI tasks. The approach is derived from a systematic literature review and conceptual analysis, and is demonstrated through two illustrative cases showing how the taxonomy and framework guide task design and evaluation.

Significance. If the taxonomy and framework can be applied reliably, the work would help address fragmentation across disciplines in XAI by decomposing broad, context-dependent desiderata into focused, evaluable units. The systematic literature review and explicit illustrative cases constitute clear strengths that support the proposal as a clarifying tool rather than a universal solution.

minor comments (2)
  1. The abstract contains a colored note stating that this is a preprint for AISoLA 2026; this artifact should be removed for the final version.
  2. The distinction between the three-axis taxonomy and the three-step framework could be made sharper in the main text, for example by adding a summary table or diagram that maps each component to its role in identifying dependencies and scoping tasks.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive assessment of our work, the recognition of its strengths in the systematic literature review and illustrative cases, and the recommendation for minor revision. We appreciate the feedback highlighting the potential of the taxonomy and framework to address fragmentation in XAI.

Circularity Check

0 steps flagged

No significant circularity; framework derived from literature review and conceptual analysis

full rationale

The paper presents a conceptual proposal: a three-axis taxonomy (target, functional role, mode of justification) and three-step framework to translate abstract XAI desiderata into concrete, benchmarkable tasks by identifying dependency structures. This is explicitly built on a systematic literature review and conceptual analysis, with utility shown via two illustrative cases. No mathematical derivations, equations, fitted parameters, or predictions exist. No self-citation chains, uniqueness theorems, or ansatzes are invoked in a load-bearing manner that reduces the central claims to inputs by construction. The argument is self-contained as a clarifying tool for scoping research questions, with independent support from the review synthesis and examples rather than circular reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The approach assumes that XAI desiderata can be organized into identifiable dependency structures and that a new taxonomy can delimit the design space without loss of essential context; no free parameters or new physical entities are introduced.

axioms (1)
  • domain assumption Desiderata in XAI form dependency structures where higher-level goals rely on foundational properties such as faithfulness and robustness.
    Invoked in the abstract as the basis for focusing on subsets rather than all desiderata simultaneously.
invented entities (1)
  • Three-axis taxonomy consisting of target, functional role, and mode of justification no independent evidence
    purpose: To classify and scope XAI tasks from abstract desiderata
    New classification scheme introduced to support the framework; no independent empirical evidence provided in the abstract.

pith-pipeline@v0.9.0 · 5816 in / 1375 out tokens · 44523 ms · 2026-05-20T03:37:48.116855+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

79 extracted references · 79 canonical work pages · 4 internal anchors

  1. [1]

    Acikgoz, E.C., Qian, C., Wang, H., Dongre, V ., Chen, X., Ji, H., Hakkani-T¨ur, D., Tur, G.: A desideratum for conversational agents: Capabilities, challenges, and future directions (2025), https://arxiv.org/abs/2504.16939

    work page arXiv 2025

  2. [2]

    Akoush, S., Paleyes, A., Van Looveren, A., Cox, C.: Desiderata for next generation of ml model serving (2022), https://arxiv.org/abs/2210.14665

    work page arXiv 2022

  3. [3]

    Studies in Health Technology and Informatics305, 164–167 (Jun 2023)

    Alvarez-Romero, C., Rodr ´ıguez-Mejias, S., Parra-Calder´on, C.: Desiderata for the data gov- ernance and fair principles adoption in health data hubs. Studies in Health Technology and Informatics305, 164–167 (Jun 2023). https://doi.org/10.3233/SHTI230452

    work page doi:10.3233/shti230452 2023

  4. [4]

    Aryal, S., Keane, M.T.: Even if explanations: Prior work, desiderata & benchmarks for semi- factual xai (2023), https://arxiv.org/abs/2301.11970

    work page arXiv 2023

  5. [5]

    Ashmore, R., Calinescu, R., Paterson, C.: Assuring the machine learning lifecycle: Desider- ata, methods, and challenges (2019), https://arxiv.org/abs/1905.04223

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  6. [6]

    Aydin, I., Diebel-Fischer, H., Freiberger, V ., M ¨oller-Klapperich, J., Buchmann, E., F ¨arber, M., Lauber-R¨onsberg, A., Platow, B.: Assessing privacy policies with ai: Ethical, legal, and technical challenges (2024), https://arxiv.org/abs/2410.08381

    work page arXiv 2024

  7. [7]

    arXiv preprint arXiv:2304.14094 (2023)

    Barbiero, P., Fioravanti, S., Giannini, F., Tonda, A., Lio, P., Di Lavore, E.: Categorical foun- dations of explainable ai: A unifying theory. arXiv preprint arXiv:2304.14094 (2023)

    work page arXiv 2023

  8. [8]

    In: World Conference on Explainable Artificial Intelligence (2023)

    Baum, D., Baum, K., Gros, T.P., Wolf, V .: Xai requirements in smart production processes: A case study. In: World Conference on Explainable Artificial Intelligence (2023)

    work page 2023

  9. [9]

    Baum, K.: Disentangling ai alignment: A structured taxonomy beyond safety and ethics (2025), https://arxiv.org/abs/2506.06286

    work page arXiv 2025

  10. [10]

    In: Model Checking Software

    Baum, K., Biewer, S., Hermanns, H., ...: Taming the ai monster: Monitoring of individual fairness for effective human oversight. In: Model Checking Software. pp. 221–242 (2025)

    work page 2025

  11. [11]

    arXiv preprint arXiv:2506.14698 (2025)

    Bender, S., Herrmann, J., M ¨uller, K.R., Montavon, G.: Towards desiderata-driven design of visual counterfactual explainers. arXiv preprint arXiv:2506.14698 (2025)

    work page arXiv 2025

  12. [12]

    Berlot-Attwell, I.: Neuro-symbolic vqa: A review from the perspective of agi desiderata (2021), https://arxiv.org/abs/2104.06365 From Abstract Desiderata to Concrete Tasks 17

    work page arXiv 2021

  13. [13]

    Biewer, S., Baum, K., Sterz, S., Hermanns, H., Hetmank, S., Langer, M., Lauber-R ¨onsberg, A., Lehr, F.: Software doping analysis for human oversight (2023), https://arxiv.org/abs/ 2308.06186

    work page arXiv 2023

  14. [14]

    In: Proceedings of the Eighteenth Interna- tional Conference on Artificial Intelligence and Law

    Borges, G.: Ai systems and product liability. In: Proceedings of the Eighteenth Interna- tional Conference on Artificial Intelligence and Law. pp. 32–39. Association for Com- puting Machinery, New York, NY , USA (2021). https://doi.org/10.1145/3462757.3466099, https://doi.org/10.1145/3462757.3466099

    work page doi:10.1145/3462757.3466099 2021

  15. [16]

    199–231 (May 2022)

    Borges, G.: The Legal Framework for IT Security in the Industry 4.0, pp. 199–231 (May 2022). https://doi.org/10.1007/978-3-030-90513-2 11

    work page doi:10.1007/978-3-030-90513-2 2022

  16. [17]

    Broniatowski, D.A., Broniatowski, D.A.: Psychological foundations of explainability and in- terpretability in artificial intelligence, vol. 4. US Department of Commerce, National Institute of Standards and Technology (2021)

    work page 2021

  17. [18]

    npj Digital Medicine8(Nov 2025)

    Br ¨uckner, S., Dridi, A., Deshmukh, A., Kirsten, T., Lauber-R ¨onsberg, A., Riedel, R., Hetmank, S., Welzel, C., Gilbert, S.: A user-driven consent platform for health data sharing in digital health applications. npj Digital Medicine8(Nov 2025). https://doi.org/10.1038/s41746-025-02147-3

    work page doi:10.1038/s41746-025-02147-3 2025

  18. [19]

    Artificial Intelligence Review58(5), 150 (2025)

    Calzarossa, M.C., Giudici, P., Zieni, R.: An assessment framework for explainable ai with applications to cybersecurity. Artificial Intelligence Review58(5), 150 (2025)

    work page 2025

  19. [20]

    In: NeurIPS 2024 Workshop on Behavioral Machine Learning (2024)

    Chitgopkar, S., Dohrmann, N., Monson, S., Mendez, J., Doshi-Velez, F., Pan, W.: Accu- racy isn’t everything: Understanding the desiderata of ai tools in legal-financial settings. In: NeurIPS 2024 Workshop on Behavioral Machine Learning (2024)

    work page 2024

  20. [21]

    In: Advances in Neural Information Processing Sys- tems

    Christiano, P.F., Leike, J., Brown, T.B., Martic, M., Legg, S., Amodei, D.: Deep reinforce- ment learning from human preferences. In: Advances in Neural Information Processing Sys- tems. vol. 30 (2017)

    work page 2017

  21. [22]

    International Review of Law, Computers & Technology pp

    Colmenarejo, A.B., State, L., Comand ´e, G.: How should an explanation be? a mapping of technical and legal desiderata of explanations for machine learning models. International Review of Law, Computers & Technology pp. 1–32 (2025)

    work page 2025

  22. [23]

    In: Workshop on Sparse Neural Networks

    Cranmer, M., Cui, C., Fielding, D.B., Ho, S., Sanchez-Gonzalez, A., Stachenfeld, K., Pfaff, T., Godwin, J., Battaglia, P., Kochkov, D.: Disentangled sparsity networks for explainable ai. In: Workshop on Sparse Neural Networks. vol. 7 (2021)

    work page 2021

  23. [24]

    Deck, L., Schom ¨acker, A., Speith, T., Sch ¨offer, J., K ¨astner, L., K ¨uhl, N.: Mapping the po- tential of explainable ai for fairness along the ai lifecycle (2024), https://arxiv.org/abs/2404. 18736

    work page 2024

  24. [25]

    arXiv preprint arXiv:2506.15408 (2025)

    Dembinsky, D., Lucieri, A., Frolov, S., Najjar, H., Watanabe, K., Dengel, A.: Unifying vxai: a systematic review and framework for the evaluation of explainable ai. arXiv preprint arXiv:2506.15408 (2025)

    work page arXiv 2025

  25. [26]

    In: International Working Conference on Requirements Engineering: Foundation for Software Quality

    Deters, H., Droste, J., Obaidi, M., Schneider, K.: How explainable is your system? towards a quality model for explainability. In: International Working Conference on Requirements Engineering: Foundation for Software Quality. pp. 3–19. Springer (2024)

    work page 2024

  26. [27]

    Towards A Rigorous Science of Interpretable Machine Learning

    Doshi-Velez, F., Kim, B.: Towards a rigorous science of interpretable machine learning. arXiv preprint arXiv:1702.08608 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  27. [28]

    Proceedings of the ACM on human- computer interaction7(CSCW1), 1–32 (2023)

    Ehsan, U., Saha, K., De Choudhury, M., Riedl, M.O.: Charting the sociotechnical gap in explainable ai: A framework to address the gap in xai. Proceedings of the ACM on human- computer interaction7(CSCW1), 1–32 (2023)

    work page 2023

  28. [29]

    Freiesleben, T., K ¨onig, G.: Dear xai community, we need to talk! fundamental misconcep- tions in current xai research (2023), https://arxiv.org/abs/2306.04292

    work page arXiv 2023

  29. [30]

    In: Pro- 18 Hanwei Zhang, Jingwen Wang, and Holger Hermanns ceedings of the AAAI/ACM Conference on AI, Ethics, and Society

    Fresz, B., Dubovitskaya, E., Brajovic, D., Huber, M.F., Horz, C.: How should ai decisions be explained? requirements for explanations from the perspective of european law. In: Pro- 18 Hanwei Zhang, Jingwen Wang, and Holger Hermanns ceedings of the AAAI/ACM Conference on AI, Ethics, and Society. vol. 7, pp. 438–450 (2024)

    work page 2024

  30. [31]

    Heger, A.K., Marquis, L.B., V orvoreanu, M., Wallach, H., Wortman Vaughan, J.: Under- standing machine learning practitioners’ data documentation perceptions, needs, challenges, and desiderata (2022), https://arxiv.org/abs/2206.02923

    work page arXiv 2022

  31. [32]

    In: International Conference on Learning Representations (ICLR) (2019)

    Hendrycks, D., Dietterich, T.: Benchmarking neural network robustness to common corrup- tions and perturbations. In: International Conference on Learning Representations (ICLR) (2019)

    work page 2019

  32. [33]

    In: International Conference on Learning Representations (ICLR) (2017)

    Hendrycks, D., Gimpel, K.: A baseline for detecting misclassified and out-of-distribution ex- amples in neural networks. In: International Conference on Learning Representations (ICLR) (2017)

    work page 2017

  33. [34]

    Hermanns, H., Lauber-R ¨onsberg, A., Meinel, P., Sterz, S., Zhang, H.: Ai act for the working programmer (2024), https://arxiv.org/abs/2408.01449

    work page arXiv 2024

  34. [35]

    PsyArXiv Preprints (2021)

    Hoffman, R., Mueller, S.T., Klein, G., Litman, J.: Measuring trust in the xai context. PsyArXiv Preprints (2021)

    work page 2021

  35. [36]

    arXiv preprint arXiv:2102.02437 (2021)

    Jin, W., Fan, J., Gromala, D., Pasquier, P., Hamarneh, G.: Euca: The end-user-centered ex- plainable ai framework. arXiv preprint arXiv:2102.02437 (2021)

    work page arXiv 2021

  36. [37]

    Kares, F., Speith, T., Zhang, H., Langer, M.: What makes for a good saliency map? comparing strategies for evaluating saliency maps in explainable ai (xai) (2025), https://arxiv.org/abs/ 2504.17023

    work page arXiv 2025

  37. [38]

    Synthese201(1) (2022), https://doi.org/10.1007/s11229-022-04008-y

    K ¨astner, L.: Modeling psychopathology: 4d multiplexes to the rescue. Synthese201(1) (2022), https://doi.org/10.1007/s11229-022-04008-y

    work page doi:10.1007/s11229-022-04008-y 2022

  38. [39]

    European Journal for Philosophy of Science14(4) (2024), https://epub.uni-bayreuth.de/id/eprint/8273/

    K ¨astner, L., Crook, B.: Explaining ai through mechanistic interpretability. European Journal for Philosophy of Science14(4) (2024), https://epub.uni-bayreuth.de/id/eprint/8273/

    work page 2024

  39. [40]

    K ¨astner, L., Langer, M., Lazar, V ., Schom¨acker, A., Speith, T., Sterz, S.: On the relation of trust and explainability: Why to engineer for trustworthiness (2021), https://arxiv.org/abs/ 2108.05379

    work page arXiv 2021

  40. [41]

    In: International conference on machine learning

    Koh, P.W., Nguyen, T., Tang, Y .S., Mussmann, S., Pierson, E., Kim, B., Liang, P.: Concept bottleneck models. In: International conference on machine learning. pp. 5338–5348. PMLR (2020)

    work page 2020

  41. [42]

    In: Advances in Neural Information Processing Systems

    Lakshminarayanan, B., Pritzel, A., Blundell, C.: Simple and scalable predictive uncertainty estimation using deep ensembles. In: Advances in Neural Information Processing Systems. vol. 30 (2017)

    work page 2017

  42. [43]

    Artificial intelligence 296, 103473 (2021)

    Langer, M., Oster, D., Speith, T., Hermanns, H., K ¨astner, L., Schmidt, E., Sesing, A., Baum, K.: What do we want from explainable artificial intelligence (xai)?–a stakeholder perspective on xai and a conceptual model guiding interdisciplinary xai research. Artificial intelligence 296, 103473 (2021)

    work page 2021

  43. [44]

    29–44 (Jan 2018)

    Lauber-R ¨onsberg, A.: Data Protection Laws, Research Ethics and Social Sciences, pp. 29–44 (Jan 2018). https://doi.org/10.1007/978-3-658-12909-5 4

    work page doi:10.1007/978-3-658-12909-5 2018

  44. [45]

    81–100 (Feb 2025)

    Lauber-R ¨onsberg, A.: Regulatory Competition: A Perspective from Data Protection Law, pp. 81–100 (Feb 2025). https://doi.org/10.1007/978-3-031-81089-3 4

    work page doi:10.1007/978-3-031-81089-3 2025

  45. [46]

    Queue16(3), 31–57 (2018)

    Lipton, Z.C.: The mythos of model interpretability: In machine learning, the concept of in- terpretability is both important and slippery. Queue16(3), 31–57 (2018)

    work page 2018

  46. [47]

    Information Fusion106, 102301 (Jun 2024)

    Longo, L., Brcic, M., Cabitza, F., Choi, J., Confalonieri, R., Del Ser, J., Guidotti, R., Hayashi, Y ., Herrera, F., Holzinger, A., Jiang, R., Khosravi, H., Lecue, F., Malgieri, G., P´aez, A., Samek, W., Schneider, J., Speith, T., Stumpf, S.: Explainable artificial intelli- gence (xai) 2.0: A manifesto of open challenges and interdisciplinary research dir...

    work page doi:10.1016/j.inffus.2024.102301 2024

  47. [48]

    Mann, S., Crook, B., K ¨astner, L., Schom¨acker, A., Speith, T.: Sources of opacity in computer systems: Towards a comprehensive taxonomy (2023), https://arxiv.org/abs/2307.14232 From Abstract Desiderata to Concrete Tasks 19

    work page arXiv 2023

  48. [49]

    Artificial Intelligence267, 1–38 (2019)

    Miller, T.: Explanation in artificial intelligence: Insights from the social sciences. Artificial Intelligence267, 1–38 (2019)

    work page 2019

  49. [50]

    Mota, N., Chakraborty, A., Biega, A.J., Gummadi, K.P., Heidari, H.: On the desiderata for online altruism: Nudging for equitable donations. Proc. ACM Hum.-Comput. Inter- act.4(CSCW2), 126 (Oct 2020). https://doi.org/10.1145/3415197, https://doi.org/10.1145/ 3415197

    work page doi:10.1145/3415197 2020

  50. [51]

    In: Joint European Conference on Machine Learning and Knowledge Discovery in Databases

    Navarro, C.M., Kanellos, G., Gottron, T.: Desiderata for explainable ai in statistical pro- duction systems of the european central bank. In: Joint European Conference on Machine Learning and Knowledge Discovery in Databases. pp. 575–590. Springer (2021)

    work page 2021

  51. [52]

    arXiv preprint (2025)

    N ˇemeˇcek, J., Krutsk ´y, M., Pele ˇska, J., G ¨urtler, P., ˇS´ır, G.: Xai desiderata for ethical ai: In- sights from the ai act. arXiv preprint (2025)

    work page 2025

  52. [53]

    In: Proceedings of the IEEE/CVF international con- ference on computer vision

    Palacio, S., Lucieri, A., Munir, M., Ahmed, S., Hees, J., Dengel, A.: Xai handbook: towards a unified framework for explainable ai. In: Proceedings of the IEEE/CVF international con- ference on computer vision. pp. 3766–3775 (2021)

    work page 2021

  53. [54]

    Briefings in Bioinformatics25(5), bbae447 (Sep 2024)

    Pallocca, M., Betti, M., Baldinelli, ., Ciliberto, G.: Clinical bioinformatics desiderata for molecular tumor boards. Briefings in Bioinformatics25(5), bbae447 (Sep 2024). https://doi.org/10.1093/bib/bbae447

    work page doi:10.1093/bib/bbae447 2024

  54. [55]

    Parliament, E., of the EU, C.: Regulation (eu) 2024/1689 of the european parliament and of the council of 13 june 2024 laying down harmonised rules on artificial intelligence and amending regulations (ec) no 300/2008, (eu) no 167/2013, (eu) no 168/2013, (eu) 2018/858, (eu) 2018/1139 and (eu) 2019/2144 and directives 2014/90/eu, (eu) 2016/797 and (eu) 2020...

    work page 2024

  55. [56]

    RISE: Randomized Input Sampling for Explanation of Black-box Models

    Petsiuk, V ., Das, A., Saenko, K.: Rise: Randomized input sampling for explanation of black- box models. arXiv preprint arXiv:1806.07421 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  56. [57]

    arXiv preprint arXiv:2206.13888 (2022)

    Renftle, M., Trittenbach, H., Poznic, M., Heil, R.: Explaining any ml model?–on goals and capabilities of xai. arXiv preprint arXiv:2206.13888 (2022)

    work page arXiv 2022

  57. [58]

    IEEE transactions on pattern analysis and machine intelligence46(4), 2104– 2122 (2023)

    Rong, Y ., Leemann, T., Nguyen, T.T., Fiedler, L., Qian, P., Unhelkar, V ., Seidel, T., Kasneci, G., Kasneci, E.: Towards human-centered explainable ai: A survey of user studies for model explanations. IEEE transactions on pattern analysis and machine intelligence46(4), 2104– 2122 (2023)

    work page 2023

  58. [59]

    Technology, Mind, and Behavior6(2) (2025)

    Schlicker, N., Lechner, F., Wehrle, K., Greulich, B., Hirsch, M.C., Langer, M.: Trustwor- thy enough? examining trustworthiness assessments of large language model-based medical agents. Technology, Mind, and Behavior6(2) (2025)

    work page 2025

  59. [60]

    Data Mining and Knowledge Dis- covery38(5), 3043–3101 (2024)

    Schwalbe, G., Finzel, B.: A comprehensive taxonomy for explainable artificial intelligence: a systematic survey of surveys on methods and concepts. Data Mining and Knowledge Dis- covery38(5), 3043–3101 (2024)

    work page 2024

  60. [61]

    In: International conference on machine learning

    Shin, S., Jo, Y ., Ahn, S., Lee, N.: A closer look at the intervention procedure of concept bot- tleneck models. In: International conference on machine learning. pp. 31504–31520. PMLR (2023)

    work page 2023

  61. [62]

    In: Proceedings of the AAAI conference on artificial intelligence

    Sokol, K., Flach, P.: Desiderata for interpretability: Explaining decision tree predictions with counterfactuals. In: Proceedings of the AAAI conference on artificial intelligence. vol. 33, pp. 10035–10036 (2019)

    work page 2019

  62. [63]

    Sokol, K., Flach, P.: Explainable ai (xai): The shape of the field and a taxonomy of methods (2020)

    work page 2020

  63. [64]

    Speith, J., Becker, S., Speith, T., Weber, M., Zou, Y ., Biega, A., Paar, C.: ”make the voodoo box go bleep bloop:” exploring end users’ understanding and information needs regarding microchips (2025), https://arxiv.org/abs/2502.13749

    work page arXiv 2025

  64. [65]

    Ethics and Informa- tion Technology26, 1–15 (Jun 2024)

    Speith, T., Crook, B., Mann, S., Schom¨acker, A., Langer, M.: Conceptualizing understanding in explainable artificial intelligence (xai): an abilities-based approach. Ethics and Informa- tion Technology26, 1–15 (Jun 2024). https://doi.org/10.1007/s10676-024-09769-3 20 Hanwei Zhang, Jingwen Wang, and Holger Hermanns

    work page doi:10.1007/s10676-024-09769-3 2024

  65. [66]

    [Journal name needed] (2023)

    Speith, T., Langer, M.: A new perspective on evaluation methods for explainable artificial intelligence (xai). [Journal name needed] (2023)

    work page 2023

  66. [67]

    Speith, T., Speith, J., Becker, S., Zou, Y ., Biega, A., Paar, C.: Explainability as a require- ment for hardware: Introducing explainable hardware (xhw) (2024), https://arxiv.org/abs/ 2302.14661

    work page arXiv 2024

  67. [68]

    In: Bridging the Gap Between AI and Reality: Second International Conference, AISoLA 2024, Crete, Greece, October 30 – November 3, 2024, Selected Papers

    Sterz, S.: What computing professionals should know about ethics: Perspectives of philoso- phers. In: Bridging the Gap Between AI and Reality: Second International Conference, AISoLA 2024, Crete, Greece, October 30 – November 3, 2024, Selected Papers. pp. 143–

    work page 2024

  68. [69]

    https://doi.org/10.1007/978-3-032-01377- 4 7, https://doi.org/10.1007/978-3-032-01377-4 7

    Springer-Verlag, Berlin, Heidelberg (2025). https://doi.org/10.1007/978-3-032-01377- 4 7, https://doi.org/10.1007/978-3-032-01377-4 7

    work page doi:10.1007/978-3-032-01377- 2025

  69. [70]

    In: The 2024 ACM Conference on Fairness Accountability and Transparency

    Sterz, S., Baum, K., Biewer, S., Hermanns, H., Lauber-R ¨onsberg, A., Meinel, P., Langer, M.: On the quest for effectiveness in human oversight: Interdisciplinary perspectives. In: The 2024 ACM Conference on Fairness Accountability and Transparency. pp. 2495–

    work page 2024

  70. [71]

    https://doi.org/10.1145/3630106.3659051, https://doi.org/10.1145/ 3630106.3659051

    ACM (Jun 2024). https://doi.org/10.1145/3630106.3659051, https://doi.org/10.1145/ 3630106.3659051

    work page doi:10.1145/3630106.3659051 2024

  71. [72]

    Sullivan, E., Kasirzadeh, A.: Explanation hacking: The perils of algorithmic recourse (2024), https://arxiv.org/abs/2406.11843

    work page arXiv 2024

  72. [73]

    Computers in Human Behavior: Artificial Humans6, 100216 (2025)

    Tim Hunsicker, C.J.K., Langer, M.: Investigating choices regarding the accuracy- transparency trade-off of ai-based systems across contexts. Computers in Human Behavior: Artificial Humans6, 100216 (2025)

    work page 2025

  73. [74]

    Walker, N., Jiang, Y ., Cakmak, M., Stone, P.: Desiderata for planning systems in general- purpose service robots (2019), https://arxiv.org/abs/1907.02300

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  74. [75]

    Journal of Machine Learning Research25(275), 1–65 (2024)

    Wang, Y ., Jordan, M.I.: Desiderata for representation learning: A causal perspective. Journal of Machine Learning Research25(275), 1–65 (2024)

    work page 2024

  75. [76]

    Wollschl ¨ager, T., Gao, N., Charpentier, B., Ketata, M.A., G ¨unnemann, S.: Uncertainty esti- mation for molecules: Desiderata and methods (2023), https://arxiv.org/abs/2306.14916

    work page arXiv 2023

  76. [77]

    arXiv preprint arXiv:2601.12804 (2026)

    Zhang, H., Cheng, L., Wen, R., Zhang, Y ., Zhang, L., Hermanns, H.: Sl-cbm: Enhancing concept bottleneck models with semantic locality for better interpretability. arXiv preprint arXiv:2601.12804 (2026)

    work page arXiv 2026

  77. [78]

    In: The 16th Asian Conference on Machine Learning (Conference Track) (2024)

    Zhang, H., Figueroa, F.T., Hermanns, H.: Saliency maps give a false sense of explanability to image classifiers: An empirical evaluation across methods and metrics. In: The 16th Asian Conference on Machine Learning (Conference Track) (2024)

    work page 2024

  78. [79]

    Computer Vision and Image Understanding248, 104101 (2024)

    Zhang, H., Torres, F., Sicre, R., Avrithis, Y ., Ayache, S.: Opti-cam: Optimizing saliency maps for interpretability. Computer Vision and Image Understanding248, 104101 (2024)

    work page 2024

  79. [80]

    Zhang, Y ., Ti ˇno, P., Leonardis, A., Tang, K.: A survey on neural network interpretability. IEEE transactions on emerging topics in computational intelligence5(5), 726–742 (2021) Desideratum Other Name(s) Description Cite Acceptance Adoption, Willingness- to-use Users are willing to adopt, accept, and continuously use a sys- tem in practice [43] Account...

    work page 2021