Pith. sign in

REVIEW 3 major objections 4 minor 160 references

Unaccountable Delegation, Fading Skills: Mapping the Risks of Workplace AI Agents

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read This paper argues that workplace AI agent risk is shaped by the human-agent boundary and deployment mode, not just by the agent itself.

desk verdict The taxonomy and scenario corpus are genuinely useful, but the distributional findings describe what the authors prompted an LLM to generate, not workplace risk prevalence; the paper's own limitations section says as much, yet the Conclusion overreaches. read the letter →

arxiv 2608.08601 v1 pith:P7SRS2LO submitted 2026-08-09 cs.AI cs.MA

Gabriele La Malfa , Lakmal Meegahapola , Edyta Bogucka , Jie M. Zhang , Michael Luck , Elizabeth Black , Daniele Quercia This is my paper · ORCID
classification cs.AIcs.MA
keywords AIagentsworkplacerisktaxonomyaugmentationvsautomationskillerosionhumanoversightjobtaskssociotechnical
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper maps the risks of AI agents at the level of individual job tasks. It embeds a three-layer framework of agents, goals, environments, and their interactions into a structured prompt, applies it to 2,078 computer-based job tasks, and generates 8,356 risk scenarios labelled by severity and by whether the agent automates or augments the task. The corpus yields a 15-category taxonomy and four findings: augmentation is not inherently safe because it can erode skills and oversight; erroneous agent actions form the largest and most severe category, mostly arising where humans interpret and act on agent output; automation shifts risks to organizations; and workers rated the taxonomy easier to use than two alternatives. The paper concludes that workplace AI agent risks depend as much on how people work with agents and how agents are deployed as on the agents themselves.

What carries the argument

The load-bearing mechanism is a multi-layer framework of an agentic AI system, defined as the agents, their goals, and their environment, together with the interaction pathways among them and with human workers, organized under three layers: technical capability, human interaction, and systemic impact. The framework is embedded in a structured prompt that labels each generated risk scenario with its component or interaction pathway and with a deployment mode, augmentation or automation. This same framework, extended through the generated scenarios plus documented incidents, produces the 15-category workplace AI agent risk taxonomy, and the taxonomy is then tested for structural distinctness, coverage against existing frameworks, and usability with workers.

What would settle it

Collect documented incidents from actual AI agent deployments in the same job roles and compare the category distribution and augmentation-versus-automation split against the paper's 8,356 scenarios; if real incidents concentrate in different categories, such as more technical model failures and fewer human-capability-erosion scenarios, the four findings and the taxonomy's claimed priorities would be called into question.

Watch

Extended reading notes

Core claim

The central claim, on the paper's own terms, is that workplace AI agent risks do not arise from agents alone; they also depend on how people work with agents and how agents are deployed. The supporting evidence is a corpus of 8,356 risk scenarios grounded in 2,078 real job tasks: Erroneous Agent Actions is the largest and most severe class, at 30.6% of scenarios with 88.2% rated high or critical, and it is dominated by misinterpretation and incorrect recommendations rather than outright fabrication. The second-largest class, Human Capability Erosion at 21.3%, is overwhelmingly an augmentation phenomenon, with 97.0% of those scenarios arising when an agent assists rather than replaces a worker, and with 1,420 scenarios specifically describing erosion of workers' decision and oversight capability. By contrast, automation shifts the dominant categories to Operational Failures, Financial Losses, and Employment Displacement. The paper interprets these patterns as evidence that safer workplaces require not only safer agents but also carefully designed human-agent collaboration.

Load-bearing premise

The entire risk map rests on whether the language model's imagined risk scenarios fairly represent real workplace risks, and the paper itself cautions that the category frequencies describe only the corpus, not how often risks occur in real workplaces.

Editorial extensions

If this is right

  • Keeping a human in the loop does not automatically make a deployment safe: 21.3% of risk scenarios are human capability erosion and 97.0% of those occur under augmentation, including 1,420 scenarios of eroded decision and oversight capability.
  • Risk assessment should target the human-agent boundary: the largest category, Erroneous Agent Actions, is mostly misinterpretation and wrong recommendations rather than fabrication, and 69.5% of those risks arise under augmentation.
  • Automation and augmentation need different governance: automation concentrates risk in organizational categories such as operational failures, financial losses, and employment displacement, while augmentation concentrates risk in the worker, including skill erosion and psychological and social risks.
  • A workplace-specific taxonomy fills gaps left by broader AI risk classifications, since organizational risks such as operational and strategic management failures and agent-execution risks are absent or only partially covered in the ten frameworks compared.
  • Workers using the proposed taxonomy classified 97.7% of scenarios correctly and preferred it over a generative AI risk taxonomy in 64% of non-tied comparisons, supporting its practical usability.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct testable extension would measure whether structured worker-contestability mechanisms, such as override channels and appeal routes, reduce the severity of Erroneous Agent Actions in field deployments, which would directly test the paper's human-agent-boundary thesis.
  • The paper's own caveat that its frequencies describe only the corpus implies a robustness prediction: repeating the scenario generation with different models and prompt designs should change the category shares substantially, so the taxonomy structure may persist even while the ranking of risk categories shifts.
  • The sharp augmentation-versus-automation split is likely a simplification, because real deployments often mix both modes; a finer-grained hybrid category could reveal additional risk patterns at the transition between human oversight and full automation.
  • If the paper is right about gradual skill erosion, long-horizon workplace studies are the natural next step: regular skill assessments of workers who use agents daily should show measurable decline in unaided task performance over months, something incident databases cannot capture.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. This paper develops a multi-layer framework for workplace AI agent risks, uses it to generate 8,356 risk scenarios from 2,078 O*NET job tasks with an LLM, validates a sample with workers and an LLM judge, and builds a 15-category workplace AI agent risk taxonomy extended from an existing generative AI risk taxonomy. The paper reports four findings: augmentation carries deskilling and erosion risks; Erroneous Agent Actions is the most frequent and severe category; automation concentrates risk in organizations while augmentation concentrates it in workers; and the proposed taxonomy is more usable than two comparison taxonomies. The authors conclude that workplace AI agent risks depend not only on agents but also on human-agent interaction and deployment choices.

Significance. If the quantitative claims are properly scoped, this is a useful and timely contribution: it provides a structured, job-task-grounded way to anticipate workplace AI agent risks, fills a genuine gap in existing AI risk taxonomies for organizational and agent-execution risks, and ships a concrete artefact (the 15-category taxonomy) that practitioners can adopt for risk registers and impact assessments. The strengths include the grounding in O*NET task descriptions, the explicit two-part validation of scenario plausibility and task alignment, the coverage comparison against ten established frameworks, and the comparative usability study. The main weakness is that the headline frequency and severity findings rest on LLM-generated labels whose representativeness is acknowledged in Section 5.3 to be limited, yet those same frequencies are used in Findings 1-3 and the Conclusion as empirical evidence. The taxonomy and framework can stand as a useful contribution, but the quantitative findings need to be re-scoped or independently validated.

major comments (3)
  1. [Section 3.3, Section 4, Section 5.3] The generation prompt explicitly instructs the model to consider risks at each framework layer and to label every scenario by component or interaction pathway, and the taxonomy in Section 3.5 is then built from those labeled scenarios. As a result, the reported corpus shares (e.g., Erroneous Agent Actions at 30.6%, Human Capability Erosion at 21.3%, and the 97.0% augmentation concentration of the latter) are partly an artifact of the elicitation instrument rather than an estimate of workplace risk prevalence. Section 5.3 concedes that these frequencies 'describe only the composition of our risk corpus' and 'should not be interpreted as estimates of how often these risks occur in real workplaces,' but Findings 1-3 and the Conclusion in Section 6 use exactly these frequencies to support the paper's central claim. This is an internal inconsistency in a load-bearing part of the argument. Please either rephrase every quantitative finding as a statement about the generated corpus, or provide an independent validation of representativeness (for example, a generation-baseline comparison and a human-annotated random sample of severity and deployment labels).
  2. [Section 3.4, Findings 2-3] The severity and deployment-mode labels on which Findings 2 and 3 rely were not human-validated. The worker study covers 450 of 8,356 scenarios (about 5.4%) and evaluates plausibility and task alignment, not severity or deployment mode. The LLM-as-a-judge reliability check reports band agreement only for plausibility (98.9%) and connection-to-task (98.1%); no agreement is reported for severity or deployment mode. Claims such as '88.2% of Erroneous Agent Actions are rated high or critical,' 'Financial Losses are 90.9% from automation,' and 'Employment Displacement is 92.7% high or critical' therefore rest on unvalidated LLM labels. Please add a human annotation study for severity and deployment mode on a random sample, or explicitly re-label these percentages as LLM-generated hypotheses about the scenario corpus rather than validated measurements.
  3. [Section 3.5] The taxonomy extension is performed by using gpt-4o-mini to map generated scenarios to Li et al.'s generative AI risk taxonomy, with unmatched scenarios inspected manually, but the paper reports no inter-annotator agreement for this mapping or for the manual conversion of incidents and failure cases. Because the same model family that generated the scenarios also performed the mapping, the resulting 15-category structure and the statement that the taxonomy 'covers all our risk scenarios' could reflect the generator's systematic mapping preferences. Reporting a sample-based agreement study with a second annotator or a different model, together with the number and definitions of unmatched scenarios before extension, would make the taxonomy construction reproducible and less dependent on a single model.
minor comments (4)
  1. [Figure 3] In the typeset copy, the category labels in panel A are not visually aligned with their corresponding bars and percentages, making it difficult to read which frequency belongs to which category. Please re-render the figure with explicit row labels.
  2. [Section 3.6] The structural integrity check uses a Jaccard overlap threshold of 0.30 and reports a mean semantic similarity of 0.24, but the threshold choice and the absence of a comparison baseline are not justified. Please report the full distribution of pairwise similarities and the rationale for the threshold in the supplement.
  3. [Section 4, Finding 4] The usability comparison is based on 26 participants and 130 classification tasks, with a pairwise p-value of .027 and a Friedman test p-value of .042; no multiple-comparison correction is applied. The wording 'outperforms' is stronger than this evidence supports; please temper the conclusion and report confidence intervals for the effect sizes.
  4. [Section 2.1] The phrase 'They can therefore overlooks socio-technical AI risks' contains a typographical error ('overlooks' should be 'overlook'), and the paragraph would benefit from a space or hyphen between 'overlook' and 'socio-technical'.

Circularity Check

1 steps flagged · score 5.0 of 10

The framework is loaded into the generation prompt, so the central conclusion partly restates the input; the paper's own §5.3 disclaimer concedes the frequencies describe only the corpus.

  1. self definitional [Section 3.3 (Generating Risk Scenarios) and Section 6 (Conclusion)]
    "The prompt explicitly instructed the model to consider risks at each framework layer (capability, human interaction, systemic impact) and to label each scenario with its corresponding component or interaction pathway (e.g., agent–human, agent–goal). ... Across our corpus, the main pattern is that workplace risks often emerge from breakdowns at the human–agent boundary rather than from isolated model failures."

    The framework's premise that risks arise from interactions among agents, goals, environments, and human workers is encoded as a generation instruction: the model is required to produce risks for every framework layer and to label each scenario with an interaction pathway. The paper's central conclusion that workplace risks 'emerge from breakdowns at the human–agent boundary rather than from isolated model failures' is therefore a restatement of the prompt's input framing rather than an independent empirical discovery. Because the taxonomy categories were extended to cover the same generated scenarios (§3.5), Findings 1–3 describe the composition of an instrument whose output was steered by the framework.

full rationale

The only load-bearing circularity is the way the framework is fed into the generator and then recovered as a finding. The prompt (§3.3) explicitly requires risks at each framework layer and pathway labels, so the corpus cannot fail to contain human-interaction risks; the qualitative conclusion that risks arise at the human–agent boundary is the framework's own premise returned as an empirical result. The paper's §5.3 limitation is unusually candid: it states that the category frequencies and distributional analyses 'describe only the composition of our risk corpus' and should not be read as real-world occurrence estimates, and that patterns may reflect prompt design. That admission is also the best evidence of partial circularity, because the Conclusion nevertheless uses those same frequencies (30.6% Erroneous Agent Actions, 21.3% Human Capability Erosion, 97.0% augmentation-bound erosion) to assert the general claim that workplace risks depend on human–agent collaboration and deployment mode. Not everything reduces: the taxonomy's structural checks, coverage comparison against ten frameworks, worker validation of scenario plausibility, and the comparative usability study are independent of the generation prompt and support the taxonomy as a classification tool. There is no load-bearing self-citation chain: the cited prior work on LLM-based risk foresight includes external sources, and no uniqueness theorem is imported from the authors. However, because the central distributional findings are partly constructed by the instrument, the paper is more than a minor self-citation case; the score reflects partial circularity, not full equivalence. The severity and deployment labels were generated by the LLM and were not separately human-validated, which compounds the instrument-dependence of Findings 1–3.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central empirical claims rest on two classes of upstream assumptions: the validity of LLM generation as a stand-in for expert foresight, and the adequacy of the framework-plus-prompt as an unbiased instrument. The largest free choices are the OpenAI model, the prompt contents, the per-task cap, and the O*NET subset; none are fitted to real-world risk data, and the paper itself warns that the resulting frequencies are corpus-internal.

free parameters (4)
  • Scenarios per task cap = up to 5 per task
    Asking for at most five risk scenarios per job task bounds corpus composition and category frequencies; this is a hand-set generation parameter, not derived from data.
  • Generator model selection = gpt-4o-mini
    The choice of gpt-4o-mini rather than a stronger or different-family model affects the content and distribution of generated risks and is a free modeling choice.
  • Framework prompt = structured prompt embedding multi-layer framework
    The prompt encodes the framework's layers and pathways, directly shaping which risks are generated and how they are labeled; this is an author-designed instrument.
  • Job-task subset threshold = 2,078 computer-based tasks (Shao et al. 2025b)
    The filtered O*NET subset determines the occupational scope; the inclusion criteria come from prior work and are not re-evaluated in this paper.
assumptions (5)
  • domain assumption LLM-generated risk scenarios are a valid proxy for expert risk foresight
    The pipeline treats gpt-4o-mini outputs as a reasonable basis for 8,356 prospective risk scenarios, supported only by a limited worker sample and a cross-family LLM judge (Section 3.4).
  • domain assumption The three-layer framework and its interaction pathways are the right risk surfaces
    The framework is built from prior literature and three expert interviews (Section 3.1), but is not itself empirically derived; embedding it in the prompt (Section 3.3) makes findings partly dependent on this choice.
  • domain assumption O*NET task descriptions carry enough context to generate job-specific risks
    Scenarios are generated from short task statements; the adequacy of this context for anticipating real workplace risks is assumed (Section 3.2).
  • domain assumption Automation vs augmentation is a meaningful binary for deployment risk
    The paper itself flags in Section 5.3 that many real deployments are hybrid and the distinction simplifies a continuum; the finding split by mode rests on this simplification.
  • standard math TF-IDF, Jaccard, and SBERT similarity scores measure taxonomy category distinctness
    Used in Section 3.6 structural integrity checks; they are standard text-similarity tools, but treating low similarity as evidence of meaningful category separation is a judgment call.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unaccountable Delegation, Fading Skills: Mapping the Risks of Workplace AI Agents." pith.science (2026). https://pith.science/paper/P7SRS2LO

@misc{pith2026260808601,
  author       = {Pith},
  title        = {Pith review of: Unaccountable Delegation, Fading Skills: Mapping the Risks of Workplace AI Agents},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/P7SRS2LO}},
  note         = {Machine review of arXiv:2608.08601}
}
read the original abstract

To anticipate socio-technical risks from AI agents, organizations need taxonomies to classify them. However, existing AI risk taxonomies focus on broad risks and do not capture job-specific risks introduced by agents. To address this gap, we make three main contributions. First, we developed a multi-layer framework from a literature review of AI agents. The framework models three core components and their interactions: agents, goals, and environment. Second, we embedded this framework in a structured prompt and applied it to descriptions of 2,078 job tasks from the O*NET database, producing 8,356 risk scenarios labeled by severity and deployment mode (automation or augmentation). We validated these scenarios with 45 workers across 10 job roles and an independent LLM judge, confirming their plausibility and alignment with job tasks. Finally, we extended an existing taxonomy to create a 15-category taxonomy of workplace AI agent risks that covers all our risk scenarios. Our analysis highlights four findings. First, augmentation is not inherently safe because overreliance on agents can gradually erode workers' skills and oversight. Second, Erroneous Agent Actions accounts for the largest share of risk scenarios and has the highest concentration of severe risks. Many arise at the human-agent boundary. Third, automation is associated mainly with organizational risks, while augmentation is associated mainly with risks to workers. Fourth, workers found our taxonomy easier to use for a risk classification task than two other taxonomies and preferred it in 64% of non-tied comparisons with a recent generative AI risk taxonomy. These findings show that workplace AI agent risks do not arise from agents alone; they also depend on how people work with agents and how agents are deployed. Safer workplaces require not only safer agents but also carefully designed human-AI agent collaboration.

Figures

Figures reproduced from arXiv: 2608.08601 by the authors.

Figure 1
Figure 1. Overview of the six-step methodology, from framework definition through risk scenario generation and vali￾dation to taxonomy construction and evaluation. Step 1 defines the multi-layer framework, which organizes sociotechnical risk into three layers: technical capability (AAIS including the AI agents, goals they pursue, and the environment they act in), human interaction (how workers engage with the AAIS and compone… view at source ↗
Figure 2
Figure 2. Workplace AI Agent Risk Taxonomy. 15 categories and 44 sub-categories organized across agent execution risks, human risks, organizational risks, and societal risks. both the failure cases and incidents into the same risk sce￾nario format as the generated scenarios. Next, our goal was to map as many scenarios as possible to an existing taxon￾omy. We used the generative AI risk taxonomy of Li et al. (Li et al. 2025b) … view at source ↗
Figure 3
Figure 3. Distribution of 8,356 workplace AI agent risk scenarios across the taxonomy categories. (A) Number of scenarios in each category, with its share of the full corpus; categories are ordered by frequency. (B) Severity composition of each category. (C) Deployment mode: whether the agent assists a worker (augmentation) or replaces one (automation). (D) Framework level at which the risk arises: technical capability, human… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

160 extracted references · 42 canonical work pages

  1. [1]

    2023 , eprint=

    TASRA: a Taxonomy and Analysis of Societal-Scale Risks from AI , author=. 2023 , eprint=

    2023

  2. [2]

    Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems , series =

    Zhang, Renwen and Li, Han and Meng, Han and Zhan, Jinyuan and Gan, Hongyuan and Lee, Yi-Chieh , title =. Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems , series =. 2025 , articleno =. doi:10.1145/3706598.3713429 , url =

    arXiv 2025

  3. [3]

    , title =

    Steenstra, Ian and Bickmore, Timothy W. , title =. Proceedings of the 25th ACM International Conference on Intelligent Virtual Agents , series =. 2025 , articleno =. doi:10.1145/3717511.3749286 , url =

    work page arXiv 2025

  4. [4]

    Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems , pages=

    Agent-Supported Foresight for AI Systemic Risks: AI Agents for Breadth, Experts for Judgment , author=. Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems , pages=

    2026

  5. [5]

    2025 , eprint=

    Responsible Artificial Intelligence Systems: A Roadmap to Society's Trust through Trustworthy AI, Auditability, Accountability, and Governance , author=. 2025 , eprint=

    2025

  6. [6]

    2025 , eprint=

    Multi-Agent Risks from Advanced AI , author=. 2025 , eprint=

    2025

  7. [7]

    2025 , eprint=

    Fairness in Agentic AI: A Unified Framework for Ethical and Equitable Multi-Agent Systems , author=. 2025 , eprint=

    2025

  8. [8]

    2024 , eprint=

    Emergence in Multi-Agent Systems: A Safety Perspective , author=. 2024 , eprint=

    2024

Show all 160 references
  1. [9]

    Conservative Agency via Attainable Utility Preservation , url=

    Turner, Alexander Matt and Hadfield-Menell, Dylan and Tadepalli, Prasad , year=. Conservative Agency via Attainable Utility Preservation , url=. doi:10.1145/3375627.3375851 , booktitle=

  2. [10]

    2019 , eprint=

    Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead , author=. 2019 , eprint=

    2019

  3. [11]

    On the Existence of Simpler Machine Learning Models , url=

    Semenova, Lesia and Rudin, Cynthia and Parr, Ronald , year=. On the Existence of Simpler Machine Learning Models , url=. doi:10.1145/3531146.3533232 , booktitle=

  4. [12]

    2025 , eprint=

    A Closer Look at the Existing Risks of Generative AI: Mapping the Who, What, and How of Real-World Incidents , author=. 2025 , eprint=

    2025

  5. [13]

    Risk Analysis , volume =

    Macrae, Carl , title =. Risk Analysis , volume =. doi:https://doi.org/10.1111/risa.13850 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1111/risa.13850 , year =

    doi

  6. [14]

    Elizabeth and Horowitz, Aaron and Selbst, Andrew , year=

    Raji, Inioluwa Deborah and Kumar, I. Elizabeth and Horowitz, Aaron and Selbst, Andrew , year=. The Fallacy of AI Functionality , url=. doi:10.1145/3531146.3533158 , booktitle=

  7. [15]

    2025 , eprint=

    The AI Risk Repository: A Comprehensive Meta-Review, Database, and Taxonomy of Risks From Artificial Intelligence , author=. 2025 , eprint=

    2025

  8. [16]

    2024 , eprint=

    A Collaborative, Human-Centred Taxonomy of AI, Algorithmic, and Automation Harms , author=. 2024 , eprint=

    2024

  9. [17]

    2023 , url=

    Adding Structure to AI Harm , author=. 2023 , url=

    2023

  10. [19]

    2024 , eprint=

    Deepfakes, Phrenology, Surveillance, and More! A Taxonomy of AI Privacy Risks , author=. 2024 , eprint=

    2024

  11. [20]

    2022 , eprint=

    A taxonomic system for failure cause analysis of open source AI incidents , author=. 2022 , eprint=

    2022

  12. [21]

    Decoding Real-World Artificial Intelligence Incidents , year=

    De Miguel Velázquez, Julia and Šćepanović, Sanja and Gvirtz, Andrés and Quercia, Daniele , journal=. Decoding Real-World Artificial Intelligence Incidents , year=

  13. [22]

    2023 , eprint=

    Sociotechnical Harms of Algorithmic Systems: Scoping a Taxonomy for Harm Reduction , author=. 2023 , eprint=

    2023

  14. [23]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Lessons for editors of AI incidents from the AI incident database , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  15. [24]

    arXiv preprint arXiv:2503.06550 , year=

    Bingoguard: Llm content moderation tools with risk levels , author=. arXiv preprint arXiv:2503.06550 , year=

    arXiv

  16. [25]

    arXiv preprint arXiv:2508.09224 , year=

    From hard refusals to safe-completions: Toward output-centric safety training , author=. arXiv preprint arXiv:2508.09224 , year=

    arXiv

  17. [26]

    Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society , volume=

    Vernacularizing Taxonomies of Harm is Essential for Operationalizing Holistic AI Safety , author=. Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society , volume=

  18. [27]

    Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society , pages=

    Participatory algorithmic management: Elicitation methods for worker well-being models , author=. Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society , pages=

    2021

  19. [28]

    Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society , pages=

    Learning occupational task-shares dynamics for the future of work , author=. Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society , pages=

  20. [29]

    Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency , pages=

    Investigating and designing for trust in ai-powered code generation tools , author=. Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency , pages=

    2024

  21. [30]

    Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society , pages=

    Why we need to know more: Exploring the state of AI incident documentation practices , author=. Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society , pages=

    2023

  22. [31]

    Proceedings of the 5th ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization , pages=

    From Incidents to Insights: Patterns of Responsibility following AI Harms , author=. Proceedings of the 5th ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization , pages=

  23. [32]

    Journal of Experimental Social Psychology , volume=

    Beyond the Turk: Alternative platforms for crowdsourcing behavioral research , author=. Journal of Experimental Social Psychology , volume=. 2017 , publisher=

    2017

  24. [33]

    Qualitative health research , volume=

    Three approaches to qualitative content analysis , author=. Qualitative health research , volume=. 2005 , publisher=

    2005

  25. [35]

    Datasheets for Datasets , journal =

    Gebru, Timnit and Morgenstern, Jamie and Vecchione, Briana and Vaughan, Jennifer Wortman and Wallach, Hanna and Daum. Datasheets for Datasets , journal =. 2021 , publisher =

    2021

  26. [36]

    Automatica , volume=

    Ironies of automation , author=. Automatica , volume=. 1983 , publisher=

    1983

  27. [37]

    Academy of management review , volume=

    Artificial intelligence and management: The automation--augmentation paradox , author=. Academy of management review , volume=. 2021 , publisher=

    2021

  28. [38]

    Journal of economic perspectives , volume=

    Why are there still so many jobs? The history and future of workplace automation , author=. Journal of economic perspectives , volume=. 2015 , publisher=

    2015

  29. [39]

    Human communication research , volume=

    Reliability in content analysis: Some common misconceptions and recommendations , author=. Human communication research , volume=. 2004 , publisher=

    2004

  30. [40]

    Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing , pages=

    Red teaming language models with language models , author=. Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing , pages=

    2022

  31. [41]

    Proceedings of the 2023 conference on empirical methods in natural language processing , pages=

    G-eval: NLG evaluation using gpt-4 with better human alignment , author=. Proceedings of the 2023 conference on empirical methods in natural language processing , pages=

    2023

  32. [42]

    2023 , eprint=

    Sociotechnical Safety Evaluation of Generative AI Systems , author=. 2023 , eprint=

    2023

  33. [43]

    2023 , eprint=

    Risk assessment at AGI companies: A review of popular risk assessment techniques from other safety-critical industries , author=. 2023 , eprint=

    2023

  34. [44]

    2023 , isbn =

    Turri, Violet and Dzombak, Rachel , title =. 2023 , isbn =. doi:10.1145/3600211.3604700 , booktitle =

    2023

  35. [46]

    2020 , eprint=

    Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database , author=. 2020 , eprint=

    2020

  36. [47]

    2024 , url=

    AIAAIC Repository , author=. 2024 , url=

    2024

  37. [48]

    2025 , url=

    AIM: The OECD AI Incidents Monitor, an evidence base for trustworthy AI , author=. 2025 , url=

    2025

  38. [49]

    Wood , title=

    Alex J. Wood , title=. 2021 , month=. doi:None , url=

    2021

  39. [50]

    Kim and Matthew T

    Pauline T. Kim and Matthew T. Bodie , journal =. Artificial Intelligence and the Challenges of Workplace Discrimination and Privacy , urldate =

  40. [51]

    2025 , eprint=

    Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems , author=. 2025 , eprint=

    2025

  41. [52]

    Some Moral and Technical Consequences of Automation , urldate =

    Norbert Wiener , journal =. Some Moral and Technical Consequences of Automation , urldate =

  42. [53]

    1966 , issn =

    Speculations Concerning the First Ultraintelligent Machine , editor =. 1966 , issn =. doi:https://doi.org/10.1016/S0065-2458(08)60418-0 , url =

    1966 doi

  43. [54]

    Science , volume =

    Yoshua Bengio and Geoffrey Hinton and Andrew Yao and Dawn Song and Pieter Abbeel and Trevor Darrell and Yuval Noah Harari and Ya-Qin Zhang and Lan Xue and Shai Shalev-Shwartz and Gillian Hadfield and Jeff Clune and Tegan Maharaj and Frank Hutter and Atılım Güneş Baydin and She...

    2024

  44. [55]

    2025 , eprint=

    The Alignment Problem from a Deep Learning Perspective , author=. 2025 , eprint=

    2025

  45. [56]

    2025 , eprint=

    Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development , author=. 2025 , eprint=

    2025

  46. [57]

    2023 , eprint=

    Benefits and Harms of Large Language Models in Digital Mental Health , author=. 2023 , eprint=

    2023

  47. [58]

    and Haber, Nick , year=

    Moore, Jared and Grabb, Declan and Agnew, William and Klyman, Kevin and Chancellor, Stevie and Ong, Desmond C. and Haber, Nick , year=. Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers. , url=. doi:10.1145/3715275.373203...

  48. [59]

    Artificial intelligence's creation and displacement of labor demand , journal =

    Taelim Choi and Nancey Green Leigh , keywords =. Artificial intelligence's creation and displacement of labor demand , journal =. 2024 , issn =. doi:https://doi.org/10.1016/j.techfore.2024.123824 , url =

    2024

  49. [60]

    AI-induced job impact: Complementary or substitution? Empirical insights and sustainable technology considerations , journal =

    Kuang-Hsien Wang and Wen-Cheng Lu , keywords =. AI-induced job impact: Complementary or substitution? Empirical insights and sustainable technology considerations , journal =. 2025 , issn =. doi:https://doi.org/10.1016/j.stae.2024.100085 , url =

    2025

  50. [61]

    Ethics and discrimination in artificial intelligence-enabled recruitment practices , volume =

    Chen, Zhisheng , year =. Ethics and discrimination in artificial intelligence-enabled recruitment practices , volume =. Humanities and Social Sciences Communications , doi =

  51. [62]

    2022 , eprint=

    Algorithm Fairness in AI for Medicine and Healthcare , author=. 2022 , eprint=

    2022

  52. [63]

    and Boyd, Danah and Friedler, Sorelle A

    Selbst, Andrew D. and Boyd, Danah and Friedler, Sorelle A. and Venkatasubramanian, Suresh and Vertesi, Janet , title =. 2019 , isbn =. doi:10.1145/3287560.3287598 , booktitle =

    2019

  53. [64]

    2025 , eprint=

    Agentic AI for Scientific Discovery: A Survey of Progress, Challenges, and Future Directions , author=. 2025 , eprint=

    2025

  54. [65]

    and Karkee, Manoj , year=

    Sapkota, Ranjan and Roumeliotis, Konstantinos I. and Karkee, Manoj , year=. AI Agents vs. Agentic AI: A Conceptual taxonomy, applications and challenges , volume=. doi:10.1016/j.inffus.2025.103599 , journal=

    2025

  55. [66]

    2024 , eprint=

    Foundational Challenges in Assuring Alignment and Safety of Large Language Models , author=. 2024 , eprint=

    2024

  56. [67]

    Harms from Increasingly Agentic Algorithmic Systems , url=

    Chan, Alan and Salganik, Rebecca and Markelius, Alva and Pang, Chris and Rajkumar, Nitarshan and Krasheninnikov, Dmitrii and Langosco, Lauro and He, Zhonghao and Duan, Yawen and Carroll, Micah and Lin, Michelle and Mayhew, Alex and Collins, Katherine and Molamohammadi, Maryam ...

  57. [68]

    2024 , eprint=

    The Ethics of Advanced AI Assistants , author=. 2024 , eprint=

    2024

  58. [69]

    2025 , eprint=

    A Survey of LLM-Driven AI Agent Communication: Protocols, Security Risks, and Defense Countermeasures , author=. 2025 , eprint=

    2025

  59. [70]

    2025 , eprint=

    An Economy of AI Agents , author=. 2025 , eprint=

    2025

  60. [71]

    2021 , url=

    AI, algorithmic and automation incidents and controversies repository (AIAAIC) , author=. 2021 , url=

    2021

  61. [72]

    , author=

    The health risks of generative AI-based wellness apps. , author=. Nature medicine , year=

  62. [73]

    Generative AI in Medical Practice: In-Depth Exploration of Privacy and Security Challenges , volume =

    Chen, Yan and Esmaeilzadeh, Pouyan , year =. Generative AI in Medical Practice: In-Depth Exploration of Privacy and Security Challenges , volume =. Journal of medical Internet research , doi =

  63. [74]

    2018 , eprint=

    Interventions over Predictions: Reframing the Ethical Debate for Actuarial Risk Assessment , author=. 2018 , eprint=

    2018

  64. [75]

    2021 , issue_date =

    Mehrabi, Ninareh and Morstatter, Fred and Saxena, Nripsuta and Lerman, Kristina and Galstyan, Aram , title =. 2021 , issue_date =. doi:10.1145/3457607 , journal =

    2021 doi

  65. [76]

    2022 , isbn =

    Fabris, Alessandro and Messina, Stefano and Silvello, Gianmaria and Susto, Gian Antonio , title =. 2022 , isbn =. doi:10.1145/3551624.3555286 , booktitle =

    2022

  66. [77]

    Anthropic News , year =

    Our Framework for Developing Safe and Trustworthy Agents , author =. Anthropic News , year =

  67. [78]

    What are AI agents? , year =

  68. [79]

    Rao, Pooja S. B. and. RiskRAG: A Data-Driven Solution for Improved AI Model Risk Reporting , year =. doi:10.1145/3706598.3713979 , booktitle =

  69. [80]

    O*NET Database , year =

  70. [81]

    1948 , month =

    Universal Declaration of Human Rights , howpublished =. 1948 , month =

    1948

  71. [82]

    Transforming our world: the 2030 Agenda for Sustainable Development , year =

  72. [83]

    2025 , month=

    What AI Means for the Future of Work , author=. 2025 , month=

    2025

  73. [84]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Preventing repeated real world AI failures by cataloging incidents: The AI incident database , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  74. [85]

    Proceedings of the 2020 conference on fairness, accountability, and transparency , pages=

    Closing the AI accountability gap: Defining an end-to-end framework for internal algorithmic auditing , author=. Proceedings of the 2020 conference on fairness, accountability, and transparency , pages=

    2020

  75. [86]

    Available at SSRN 3877437 , year=

    Assembling accountability: algorithmic impact assessment for the public interest , author=. Available at SSRN 3877437 , year=

  76. [87]

    arXiv preprint arXiv:2209.07858 , year=

    Red teaming language models to reduce harms: Methods, scaling behaviors, and lessons learned , author=. arXiv preprint arXiv:2209.07858 , year=

    arXiv

  77. [88]

    Frontiers of Computer Science , volume=

    A survey on large language model based autonomous agents , author=. Frontiers of Computer Science , volume=. 2024 , publisher=

    2024

  78. [89]

    arXiv preprint arXiv:2304.05332 , year=

    Emergent autonomous scientific research capabilities of large language models , author=. arXiv preprint arXiv:2304.05332 , year=

    arXiv

  79. [90]

    Superagency in the workplace: Empowering people to unlock AI’s full potential at work , author=

  80. [91]

    Pause Giant AI Experiments: An Open Letter , author=

  81. [92]

    Frank and David Autor and James E

    Morgan R. Frank and David Autor and James E. Bessen and Erik Brynjolfsson and Manuel Cebrian and David J. Deming and Maryann Feldman and Matthew Groh and José Lobo and Esteban Moro and Dashun Wang and Hyejin Youn and Iyad Rahwan , title =. Proceedings of the National Academy o...

    2019

  82. [93]

    , title=

    Shen, Y. , title=. Econ Change Restruct , volume=. 2024 , url=

    2024

  83. [94]

    Cazzaniga, Mauro , journal =. Gen-. 2024 , month =. doi:10.5089/9798400262548.006 , issn =

    2024 doi

  84. [95]

    and Yaroshenko, Oleg M

    Getman, Anatolii P. and Yaroshenko, Oleg M. and Dmytryk, Olga O. and Tykhonovych, Oleksii Y. and Hryn, Dmytro V. , title=. AI & Society , volume=. 2025 , doi=

    2025

  85. [96]

    Artificial Intelligence and Inequality: Challenges and Opportunities , journal =

    Farahani, Milad and Ghasemi, Ghazal , year =. Artificial Intelligence and Inequality: Challenges and Opportunities , journal =

  86. [97]

    AI , VOLUME =

    Albaroudi, Elham and Mansouri, Taha and Alameer, Ali , TITLE =. AI , VOLUME =. 2024 , NUMBER =

    2024

  87. [98]

    European Labour Law Journal , year=

    AI surveillance: Reclaiming privacy through informational control , author=. European Labour Law Journal , year=. doi:10.17863/CAM.113573 , url=

    doi

  88. [99]

    European Labour Law Journal , volume =

    Aislinn Kelly-Lyth and Anna Thomas , title =. European Labour Law Journal , volume =. 2023 , doi =

    2023

  89. [100]

    Algorithmic Management: The Role of AI in Managing Workforces , journal =

    Jarrahi, Mohammad Hossein and Möhlmann, Mareike and Lee, Min Kyung , year =. Algorithmic Management: The Role of AI in Managing Workforces , journal =

  90. [101]

    Journal of Managerial Psychology , volume =

    Mendy, John and Jain, Apoorva and Thomas, Asha , title =. Journal of Managerial Psychology , volume =. 2024 , month =. doi:10.1108/JMP-05-2024-0388 , url =

    2024 doi

  91. [102]

    International Journal of Environmental Research and Public Health , VOLUME =

    Fiegler-Rudol, Jakub and Lau, Karolina and Mroczek, Alina and Kasperczyk, Janusz , TITLE =. International Journal of Environmental Research and Public Health , VOLUME =. 2025 , NUMBER =

    2025

  92. [103]

    2023 , issue_date =

    Corvite, Shanley and Roemmich, Kat and Rosenberg, Tillie Ilana and Andalibi, Nazanin , title =. 2023 , issue_date =. doi:10.1145/3579600 , journal =

    2023 doi

  93. [104]

    2025 , eprint=

    Fully Autonomous AI Agents Should Not be Developed , author=. 2025 , eprint=

    2025

  94. [105]

    , title =

    Leveson, Nancy G. , title =. 2012 , month =. doi:10.7551/mitpress/8179.001.0001 , url =

    2012 doi

  95. [106]

    Science , volume =

    Tyna Eloundou and Sam Manning and Pamela Mishkin and Daniel Rock , title =. Science , volume =. 2024 , doi =. https://www.science.org/doi/pdf/10.1126/science.adj0998 , abstract =

    2024 doi

  96. [107]

    Strategic Management Journal , volume =

    Felten, Edward and Raj, Manav and Seamans, Robert , title =. Strategic Management Journal , volume =. doi:https://doi.org/10.1002/smj.3286 , url =. https://sms.onlinelibrary.wiley.com/doi/pdf/10.1002/smj.3286 , abstract =

    doi

  97. [108]

    2009 , isbn =

    Russell, Stuart and Norvig, Peter , title =. 2009 , isbn =

    2009

  98. [109]

    2009 , isbn =

    Wooldridge, Michael , title =. 2009 , isbn =

    2009

  99. [110]

    2008 , isbn =

    Shoham, Yoav and Leyton-Brown, Kevin , title =. 2008 , isbn =

    2008

  100. [111]

    Workforce , author=

    Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce , author=. 2025 , eprint=

    2025

  101. [112]

    Science China Information Sciences , volume=

    The rise and potential of large language model based agents: A survey , author=. Science China Information Sciences , volume=. 2025 , publisher=

    2025

  102. [113]

    Science , volume=

    GPTs are GPTs: Labor market impact potential of LLMs , author=. Science , volume=. 2024 , publisher=

    2024

  103. [114]

    arXiv preprint arXiv:2303.01157 , year=

    How will language modelers like ChatGPT affect occupations and industries? , author=. arXiv preprint arXiv:2303.01157 , year=

    arXiv

  104. [115]

    ExploreGen: Large Language Models for Envisioning the Uses and Risks of AI Technologies , year =

    Herdel, Viviane and S\'. ExploreGen: Large Language Models for Envisioning the Uses and Risks of AI Technologies , year =. Proceedings of the 2024 AAAI/ACM Conference on AI, Ethics, and Society , pages =

    2024

  105. [116]

    2023 , eprint=

    AHA!: Facilitating AI Impact Assessment by Generating Examples of Harms , author=. 2023 , eprint=

    2023

  106. [117]

    and Kulkarni, Chinmay and Wilcox, Lauren and Terry, Michael and Madaio, Michael , year=

    Wang, Zijie J. and Kulkarni, Chinmay and Wilcox, Lauren and Terry, Michael and Madaio, Michael , year=. Farsight: Fostering Responsible AI Awareness During AI Application Prototyping , url=. doi:10.1145/3613904.3642335 , booktitle=

  107. [118]

    Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency , pages=

    Who Audits the Auditors? Recommendations from a field scan of the algorithmic auditing ecosystem , author=. Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency , pages=

    2022

  108. [119]

    ☑ Fairness Toolkits, A Checkbox Culture?

    “☑ Fairness Toolkits, A Checkbox Culture?” On the Factors that Fragment Developer Practices in Handling Algorithmic Harms , author=. Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society , pages=

    2023

  109. [120]

    Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency , pages=

    Towards a multi-stakeholder value-based assessment framework for algorithmic systems , author=. Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency , pages=

    2022

  110. [121]

    Proceedings of the 2021 ACM conference on fairness, accountability, and transparency , pages=

    Algorithmic impact assessments and accountability: The co-construction of impacts , author=. Proceedings of the 2021 ACM conference on fairness, accountability, and transparency , pages=

    2021

  111. [122]

    Proceedings of the AAAI Conference on Human Computation and Crowdsourcing , author=

    Atlas of AI Risks: Enhancing Public Understanding of AI Risks , volume=. Proceedings of the AAAI Conference on Human Computation and Crowdsourcing , author=. 2024 , month=. doi:10.1609/hcomp.v12i1.31598 , number=

    2024 doi

  112. [123]

    2023 , eprint=

    MTEB: Massive Text Embedding Benchmark , author=. 2023 , eprint=

    2023

  113. [124]

    2024 , url=

    Linq-Embed-Mistral:Elevating Text Retrieval with Improved GPT Data Through Task-Specific Control and Quality Refinement , author=. 2024 , url=

    2024

  114. [125]

    2020 , eprint=

    UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction , author=. 2020 , eprint=

    2020

  115. [126]

    2024 , eprint=

    Dynamic data summarization for hierarchical spatial clustering , author=. 2024 , eprint=

    2024

  116. [127]

    2024 , eprint=

    Automation from the Worker's Perspective , author=. 2024 , eprint=

    2024

  117. [128]

    2024 , number=

    Ozge Demirci and Jonas Hannane and Xinrong Zhu , title=. 2024 , number=. doi:None , url=

    2024

  118. [129]

    2025 , eprint=

    Which Economic Tasks are Performed with AI? Evidence from Millions of Claude Conversations , author=. 2025 , eprint=

    2025

  119. [130]

    Advances in Neural Information Processing Systems , volume=

    Llm evaluators recognize and favor their own generations , author=. Advances in Neural Information Processing Systems , volume=

  120. [131]

    and Zhang, Hao and Gonzalez, Joseph E

    Zheng, Lianmin and Chiang, Wei-Lin and Sheng, Ying and Zhuang, Siyuan and Wu, Zhanghao and Zhuang, Yonghao and Lin, Zi and Li, Zhuohan and Li, Dacheng and Xing, Eric P. and Zhang, Hao and Gonzalez, Joseph E. and Stoica, Ion , title =. Proceedings of the 37th International Conf...

    2023

  121. [132]

    2023 , eprint=

    Can Large Language Models Be an Alternative to Human Evaluations? , author=. 2023 , eprint=

    2023

  122. [133]

    2023 , eprint=

    Sparks of Artificial General Intelligence: Early experiments with GPT-4 , author=. 2023 , eprint=

    2023

  123. [134]

    2023 , eprint=

    Large Language Models are not Fair Evaluators , author=. 2023 , eprint=

    2023

  124. [135]

    2024 , eprint=

    LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods , author=. 2024 , eprint=

    2024

  125. [136]

    2025 , eprint=

    From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judge , author=. 2025 , eprint=

    2025

  126. [137]

    2025 , url =

    Pete Bryan and Giorgio Severi and Joris de Gruyter and Daniel Jones and Blake Bullwinkel and Amanda Minnich and Shiven Chawla and Gary Lopez and Martin Pouliot and Adam Fourney and Whitney Maxwell and Katherine Pratt and Saphir Qi and Nina Chikanov and Roman Lutz and Raja Sekh...

    2025

  127. [138]

    2025 , eprint=

    With Great Capabilities Come Great Responsibilities: Introducing the Agentic Risk & Capability Framework for Governing Agentic AI Systems , author=. 2025 , eprint=

    2025

  128. [139]

    2025 , eprint=

    A Safety and Security Framework for Real-World Agentic Systems , author=. 2025 , eprint=

    2025

  129. [140]

    RISK@ICTSS , year=

    Towards Integration of Compositional Risk Analysis Using Monte Carlo Simulation and Security Testing , author=. RISK@ICTSS , year=

  130. [141]

    The Guardian , year =

    Ian Sample , title =. The Guardian , year =

  131. [142]

    The New York Times , year =

    Noam Scheiber , title =. The New York Times , year =

  132. [143]

    PC Mag , year =

    Emily Forlini , title =. PC Mag , year =

  133. [144]

    Understanding Human-AI Augmentation in the Workplace: A Review and a Future Research Agenda , journal =

    Nguyen, Trinh and Elbanna, Amany , year =. Understanding Human-AI Augmentation in the Workplace: A Review and a Future Research Agenda , journal =

  134. [145]

    Journal of Management , volume =

    Yiduo Shao and Chengquan Huang and Yifan Song and Mo Wang and Young Ho Song and Ruodan Shao , title =. Journal of Management , volume =. 2025 , doi =

    2025

  135. [146]

    2020 , issn =

    Artificial intelligence in the workplace – A double-edged sword , journal =. 2020 , issn =. doi:https://doi.org/10.1108/IJILT-02-2020-0022 , url =

    2020 doi

  136. [147]

    Intelligence augmentation: rethinking the future of work by leveraging human performance and abilities , journal =

    Harborth, David and K. Intelligence augmentation: rethinking the future of work by leveraging human performance and abilities , journal =. 2022 , volume =. doi:10.1007/s10055-021-00590-7 , url =

    2022 doi

  137. [148]

    Williams, Christopher Y. K. and Zack, Travis and Miao, Brenda Y. and Sushil, Madhumita and Wang, Michelle and Kornblith, Aaron E. and Butte, Atul J. , title =. JAMA Network Open , volume =. 2024 , month =. doi:10.1001/jamanetworkopen.2024.8895 , url =

    2024

  138. [149]

    AI , VOLUME =

    Wen, He and Parsaee, Mojtaba and Sajid, Zaman , TITLE =. AI , VOLUME =. 2025 , NUMBER =

    2025

  139. [150]

    Matthew and Campos, Daniel Vargas , title =

    Kennedy, Wm. Matthew and Campos, Daniel Vargas , title =. Proceedings of the 2024 AAAI/ACM Conference on AI, Ethics, and Society , pages =. 2025 , publisher =

    2024

  140. [151]

    GPT-4o mini: advancing cost-efficient intelligence , howpublished =

  141. [152]

    Gemini models , howpublished =

  142. [153]

    2024 , eprint=

    Can LLMs Generate Novel Research Ideas? A Large-Scale Human Study with 100+ NLP Researchers , author=. 2024 , eprint=

    2024

  143. [154]

    2024 , eprint=

    Unveiling LLM Evaluation Focused on Metrics: Challenges and Solutions , author=. 2024 , eprint=

    2024

  144. [155]

    2026 , eprint=

    Agents of Chaos , author=. 2026 , eprint=

    2026

  145. [156]

    2019 , eprint=

    Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks , author=. 2019 , eprint=

    2019

  146. [157]

    2024 , eprint=

    AI Risk Categorization Decoded (AIR 2024): From Government Regulations to Corporate Policies , author=. 2024 , eprint=

    2024

  147. [158]

    Understanding artificial intelligence ethics and safety: A guide for the responsible design and implementation of AI systems in the public sector , publisher =

    Leslie, David , keywords =. Understanding artificial intelligence ethics and safety: A guide for the responsible design and implementation of AI systems in the public sector , publisher =. 2019 , copyright =. doi:10.5281/ZENODO.3240529 , url =

    2019 doi

  148. [159]

    2023 , url =

    Fergusson, Grant and Fitzgerald, Caitriona and Frascella, Chris and Iorio, Megan and McBrien, Tom and Schroeder, Calli and Winters, Ben and Zhou, Enid , title =. 2023 , url =

    2023

  149. [160]

    2023 , url =

    Hoffman, Mia and Frase, Heather , title =. 2023 , url =

    2023

  150. [161]

    2023 , organization =

    Types of harm -. 2023 , organization =

    2023

  151. [162]

    Not My Voice! A Taxonomy of Ethical and Safety Harms of Speech Generators , url=

    Hutiri, Wiebke and Papakyriakopoulos, Orestis and Xiang, Alice , year=. Not My Voice! A Taxonomy of Ethical and Safety Harms of Speech Generators , url=. doi:10.1145/3630106.3658911 , booktitle=

  152. [163]

    2022 , eprint=

    From plane crashes to algorithmic harm: applicability of safety engineering frameworks for responsible ML , author=. 2022 , eprint=

    2022

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.