Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

"Maybe We Need Some More Examples:" Individual and Team Drivers of Developer GenAI Tool Use

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Frequent AI users see Copilot as a collaborator, not a feature

desk verdict A well-executed paired-interview study with a genuinely useful matched-pair telemetry design; the causal framing in the abstract overreaches, but the descriptive taxonomy and the Productivity Pressure Paradox are worth taking seriously. read the letter →

arxiv 2507.21280 v1 pith:5T6UKFFX submitted 2025-07-28 cs.SE

Courtney Miller , Rudrajit Choudhuri , Mara Ulloa , Sankeerti Haniyur , Robert DeLine , Margaret-Anne Storey , Emerson Murphy-Hill , Christian Bird
show 1 more author
Jenna L. Butler
This is my paper · ORCID
classification cs.SE
keywords generativeAIsoftwareengineeringdeveloperadoptionpairedinterviewsGitHubCopilotproductivitypressureparadoxtechnologyacceptancesociotechnicalfactors
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

This paper argues that uneven adoption of generative AI coding tools is not primarily a matter of skill or tool quality; it is driven by three individual dispositions: whether developers perceive the tool as a collaborator or a feature, whether they engage experimentally or conservatively, and whether they respond to failures with adaptive persistence or quick abandonment. The evidence comes from paired interviews with 54 developers in 27 teams, each pair matched for team, role, and seniority but differing in telemetry-measured Copilot use. Because each pair shares the same codebase, manager, and policies, the comparison isolates individual factors from team context, and it lets the authors see how team scaffolding amplifies or suppresses those factors. The paper also introduces the Productivity Pressure Paradox: when management raises productivity expectations without providing learning time and context-specific support, the pressure itself blocks the experimentation and skill building that would produce the gains. If correct, the finding shifts responsibility for adoption outcomes from individual developers to organizational design.

What carries the argument

The load-bearing machinery is the matched-pair interview design: 27 pairs, each with one frequent and one infrequent user from the same team, selected from telemetry by a Hungarian-assignment pairing algorithm that maximizes usage differences within groups matched on career stage, title, level, country, and manager. The design converts team context into a controlled background, so contrasting accounts of the same codebase, manager, and policies can be attributed to individual interpretation. On top of this, the conceptual framework organizes the adoption journey into four stages—mindset formation, approach determination, experience and learning, and integration and evolution—with external factors shown as amplifiers.

What would settle it

Track a cohort of new Copilot users from first exposure: record weekly usage plus short surveys of whether they view the tool as a collaborator or feature, their willingness to experiment, and their response to a failed generation. If collaborator framing and adaptive persistence appear only several weeks after usage rises, the paper's causal direction is wrong; if they appear before usage diverges, the drivers are genuine antecedents.

Watch

Extended reading notes

Core claim

Using a paired interview design, the authors identify a systematic profile separating frequent from infrequent GenAI tool users even within the same team. Frequent users tend to frame the tool as a collaborator with realistic expectations, integrate it continuously, experiment with its boundaries, learn self-directedly, and persist adaptively when outputs fail; infrequent users tend to frame it as a feature, keep it to task-specific boilerplate uses, learn minimally, and abandon quickly after a bad result. These individual factors are not private quirks: team and organizational conditions—leadership messaging, context-specific demonstrations, social learning structures, and protected exploration time—actively shape them, and their absence leaves individual developers to solve an organizational problem alone. The paper names that dynamic the Productivity Pressure Paradox and argues that productivity gains from GenAI require systematic organizational change, not individual workflow optimization.

Load-bearing premise

The paper assumes that the perceptions, habits, and failure responses people describe in interviews are causes of how often they use the tool, not post-hoc explanations for behavior already visible in usage records.

Editorial extensions

If this is right

  • Organizations that want wider adoption should invest in context-specific resources, such as team demonstrations, use-case wikis, and designated AI champions, rather than generic training alone.
  • Protected experimentation time, such as dedicated learning days or relaxed velocity expectations, is a precondition for effective GenAI use, not a luxury.
  • Leadership messaging and peer demonstrations can cultivate a collaborator framing with realistic expectations, which the paper links to deeper engagement.
  • If the Productivity Pressure Paradox holds, measuring success by short-term velocity will predictably suppress the learning that produces long-term productivity gains.

Reading between the lines

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

  • The cross-sectional design cannot distinguish whether collaborator framing causes frequent use or emerges from it; a plausible reading is that the three individual factors are partly self-reinforcing rationalizations after usage becomes habitual.
  • The paradox suggests a testable prediction: teams that shield developers from deadlines during an initial adoption period should converge toward frequent use faster than teams under constant delivery pressure.
  • Because usage was measured only through GitHub Copilot invocation days, developers who route around Copilot via other AI tools may be misclassified, which could bias the infrequent group toward people with alternative AI workflows and strengthen the appearance of individual difference.
  • The matched-pair method generalizes to other professional tools whenever telemetry can classify usage; the same design could be applied to ChatGPT or other coding assistants inside and outside Copilot ecosystems.
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 / 5 minor

Summary. The paper reports a paired-interview study of 54 developers from 27 teams at one large software company, with participants selected from eight weeks of GitHub Copilot telemetry so that each team contributed one frequent and one infrequent user. Through thematic analysis, the authors propose a conceptual framework in which individual factors (tool perception, performance expectations, adoption approach, integration strategy, learning strategy, response to failure, role evolution, and skill evolution) distinguish frequent from infrequent users, and in which team and organizational factors (leadership messaging, context-specific resources, social learning structures, and protected learning time) amplify or suppress those individual factors. The paper introduces the 'Productivity Pressure Paradox,' arguing that organizational expectations for rapid productivity gains, without corresponding learning support, undermine the productivity benefits that motivated adoption. The claimed contributions are the conceptual framework, the taxonomy of organizational shaping factors, and the paradox concept with associated intervention strategies.

Significance. If read as a descriptive qualitative theory of how developers in similar team contexts talk about and approach GenAI tools, this is a valuable study. The paired design is a genuine methodological strength: it controls for team, codebase, manager, and policy context to a degree unusual in adoption research, and the telemetry-based classification of frequent versus infrequent users avoids relying solely on self-reported usage. The member-checking step and the detailed quotes give the taxonomy credibility, and Table 2 offers actionable, evidence-grounded interventions. The main significance risk is that the paper's packaging overclaims causal status: the abstract's 'result primarily from' and the paradox's 'undermining productivity' go beyond what cross-sectional retrospective interviews can establish. The descriptive taxonomy and the organizational-scaffolding argument remain defensible and useful after appropriate reframing.

major comments (3)
  1. [Abstract; §1; §3.2; §3.4; §3.6] The headline claim that usage differences 'result primarily from' tool perception, engagement approach, and response to failure, and that the authors 'demonstrate' this, conflicts with the limitation stated in Section 3.6 that the 'qualitative design does not support formal causal inference.' Because the interviews were conducted after the telemetry window and the interviewer/coders knew each participant's usage class (Section 3.4), the accounts of perception and strategy are retrospective and could be consequences or rationalizations of frequent use rather than antecedents. The abstract and the RQ1 framing should be revised to describe correlates and distinguishing patterns, with causal 'drivers' language moved to the hypothesis-generating discussion.
  2. [§4.3.1; Figure 3; §3.4] The category 'adaptive persistence vs. quick abandonment' is close to a re-description of the outcome variable (days of Copilot use): a developer who persists in integrating the tool into continuous workflows will, almost by definition, accumulate more usage days. Since coding was done with knowledge of which participant was frequent versus infrequent, the risk that some taxonomy categories merely restate the selection criterion is non-negligible. The authors should either provide a blindness check for coding, add an explicit acknowledgment in Section 3.6 that some categories overlap with the outcome, or show evidence that persistence manifested in domains beyond frequency of tool invocation.
  3. [§5.4; Figure 4; §6.1] The Productivity Pressure Paradox is presented as a demonstrated dynamic that 'undermines the very productivity benefits that motivate adoption,' but the study contains no productivity or time-use data. The supporting evidence consists of participants' reports that deadlines and expectations left them without time to learn (e.g., PID6, PID44) and one participant's 'chicken or egg' reflection (PID46). These reports support a perceived-pressure mechanism, not the causal claim that invested learning time would have produced the assumed gains or that pressure causally reduced realized productivity. Please present the paradox as an emergent hypothesis and add appropriate evidential hedges where the text claims 'systemic challenges' and a 'counterproductive cycle.'
minor comments (5)
  1. [§3.4; Figure 3] Section 3.4 states that the authors avoid frequency counts or percentages in line with guidance against quantifying qualitative data, yet Figure 3 displays distributions of user type proportions by factor; please clarify whether Figure 3 is illustrative of thematic prevalence and how these proportions were generated.
  2. [§4.4.1] The phrase 'whomst they viewed as the immediate threat' contains a typo; 'whomst' should be 'whom' or similar wording.
  3. [§3.4] The saturation criterion of 'two consecutive interview pairs without learning any new major insights' is reported but not justified; please explain why this criterion was sufficient, ideally with a methodological reference beyond the general citation in [32].
  4. [§3.6; §1] The paper measures only Copilot usage days and acknowledges in Section 3.6 that other GenAI tools were not captured, yet several passages (e.g., Section 1's 'hampers productivity efforts') slide from usage frequency to productivity; please consistently label the outcome variable as tool-usage frequency rather than productivity.
  5. [Throughout] There are small inconsistencies in capitalization ('github' vs. 'GitHub', 'genAI' vs. 'GenAI') that should be harmonized in a final proofreading pass.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the paper's thematic findings are interpretive summaries of interview data, not predictions or derived quantities that reduce to fitted inputs.

full rationale

This is an inductive qualitative interview study rather than a formal derivation or modeling paper. There are no equations, no fitted parameters renamed as predictions, and no prediction that is equivalent to its inputs by construction. The central categories (collaborator vs. feature, experimental vs. conservative, adaptive persistence vs. quick abandonment) were generated through iterative thematic coding of the interview corpus, and the paper explicitly acknowledges in Section 3.6 that its qualitative design 'does not support formal causal inference.' A concern that the categories may partly restate the outcome or that coders' awareness of usage groups influenced interpretation is a real validity threat, but it is not the equation-level circularity this axis targets. The overlapping-author citations ([13], [17], [18]) are background literature, not load-bearing justifications for the paper's taxonomy or paradox; the central finding does not reduce to a self-citation chain. The 'Productivity Pressure Paradox' is a synthesis of participant-reported experiences and prior technostress/productivity literature, not a renamed result smuggled in as a derivation. Accordingly, no significant circularity is present; score 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The central claims rest mainly on qualitative assumptions about telemetry validity, matched-context equivalence, self-report accuracy, and a study-specific saturation rule. The only clearly invented entity is the named Productivity Pressure Paradox, which is an interpretive synthesis of the interview data rather than an independently measured quantity.

free parameters (4)
  • Eight-week telemetry observation window (April 1 to May 31, 2025) = 8 weeks
    Hand-chosen window determines each developer's use-frequency classification; changing the window could reclassify participants.
  • Top-third usage gap threshold for pair inclusion = top third of usage differences
    Pragmatic filter to ensure contrasting pairs; no theoretical justification is given, and it affects which pairs are studied.
  • Median usage split within matched groups = median days of use
    Defines relative frequent vs infrequent within each group; frequency labels are relative rather than absolute.
  • Saturation stopping rule = two consecutive pairs with no new major insights (4 interviews)
    Stopping criterion is a judgment call; earlier or later stopping could alter theme saturation.
assumptions (4)
  • domain assumption Copilot telemetry days is a valid proxy for GenAI tool usage
    Section 3.2 uses days of Copilot use in an 8-week window as the usage measure; Section 3.6 acknowledges that general-purpose or non-approved AI tools are excluded, so developers using other AI tools could be misclassified.
  • domain assumption Matched pairs share the same relevant team-level context
    Section 3.2 matches on career stage, job title, employee level, country, direct manager, and primary language; this assumes these variables capture the environmental confounds that matter.
  • domain assumption Retrospective self-reports accurately reflect actual drivers of usage
    Section 3.4 collects perceptions and strategies via interviews after usage is known; memory and post-hoc rationalization can distort accounts, and the direction of causality is not tested.
  • ad hoc to paper Thematic saturation after two pairs with no new insights is sufficient
    Section 3.4 defines saturation pragmatically; this is a study-specific criterion rather than a standard mathematical axiom.
invented entities (1)
  • Productivity Pressure Paradox
    purpose: Named conceptual mechanism explaining why organizational expectations for fast productivity gains without learning support create a self-reinforcing cycle that reduces exploration and skill building.
    The construct is synthesized from interview themes in Sections 5.4 and 6.1 and illustrated in Figure 4; it has no falsifiable handle outside this paper's qualitative data and is not validated in other samples.

how reviews work

0 comments
Cite this review

Pith. "Pith review of "Maybe We Need Some More Examples:" Individual and Team Drivers of Developer GenAI Tool Use." pith.science (2026). https://pith.science/paper/5T6UKFFX

@misc{pith2026250721280,
  author       = {Pith},
  title        = {Pith review of: "Maybe We Need Some More Examples:" Individual and Team Drivers of Developer GenAI Tool Use},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5T6UKFFX}},
  note         = {Machine review of arXiv:2507.21280}
}
read the original abstract

Despite the widespread availability of generative AI tools in software engineering, developer adoption remains uneven. This unevenness is problematic because it hampers productivity efforts, frustrates management's expectations, and creates uncertainty around the future roles of developers. Through paired interviews with 54 developers across 27 teams -- one frequent and one infrequent user per team -- we demonstrate that differences in usage result primarily from how developers perceive the tool (as a collaborator vs. feature), their engagement approach (experimental vs. conservative), and how they respond when encountering challenges (with adaptive persistence vs. quick abandonment). Our findings imply that widespread organizational expectations for rapid productivity gains without sufficient investment in learning support creates a "Productivity Pressure Paradox," undermining the very productivity benefits that motivate adoption.

Figures

Figures reproduced from arXiv: 2507.21280 by the authors.

Figure 1
Figure 1. Conceptual theoretical framework. Purple ovals == [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. GenAI Tool Usage Distributions differences in usage. Specifically, We selected pairs that had same primary programming language and a usage gap in the top third across all pairs (measured by the difference in days of GenAI tool usage between the pair’s frequent and infrequent user). Note our definitions of frequent and infrequent users are relative to their teams’ usage (not to all developers across the company). We… view at source ↗
Figure 3
Figure 3. Distributions of user type proportions by factor. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: The Productivity Pressure Paradox. A common theme among participants was experiencing increased productivity expectations from management because they have ac￾cess to these tools e.g., “I know that management is expecting us to produce code faster. And I know that the …

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. You Shall Not Pass! Where and Why Developers Draw The Line on AI Autonomy

    cs.HC 2026-07 unverdicted novelty 6.0 of 10

    Mixed-methods survey finds developers accept AI producing work under oversight but resist autonomy on identity-defining, human-facing, and design tasks, modulated by experience, risk tolerance, and task attributes.

Reference graph

Works this paper leans on

99 extracted references · 59 canonical work pages · cited by 1 Pith paper

  1. [1]

    [n. d.]. GitHub Copilot. https://github.com/features/copilot. Accessed Jun. 2025

    2025

  2. [2]

    [n. d.]. HeyMarvin. https://heymarvin.com/. Accessed Jun. 2025

    2025

  3. [3]

    Saleema Amershi, Dan Weld, Mihaela Vorvoreanu, Adam Fourney, Besmira Nushi, Penny Collisson, Jina Suh, Shamsi Iqbal, Paul N Bennett, Kori Inkpen, et al. 2019. Guidelines for human-AI interaction. In2019 CHI Conference on Human Factors in Computing Systems. 1–13

    2019

  4. [4]

    Jessica Apotheker et al. [n. d.]. From Potential to Profit with GenAIe. https:// www.bcg.com/publications/2024/from-potential-to-profit-with-genai Accessed: 2025-05-17

    2024

  5. [5]

    Leonardo Banh, Florian Holldack, and Gero Strobel. 2025. Copiloting the future: How generative AI transforms Software Engineering.Information and Software Technology183 (2025), 107751

    2025

  6. [6]

    Victor R Basili, Forrest Shull, and Filippo Lanubile. 2002. Building knowledge through families of experiments.IEEE transactions on software engineering25, 4 (2002), 456–473

    2002

  7. [7]

    Brett A Becker, Paul Denny, James Finnie-Ansley, Andrew Luxton-Reilly, James Prather, and Eddie Antonio Santos. 2023. Programming is hard-or at least it used to be: Educational opportunities and challenges of ai code generation. In Proceedings of the 54th ACM Technical Symposium on Computer Science Education V. 1. 500–506

    2023

  8. [8]

    Joel Becker, Nate Rush, Elizabeth Barnes, and David Rein. 2025. Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity. arXiv preprint arXiv:2507.09089(2025)

    arXiv 2025

Show all 99 references
  1. [9]

    Josiah D Boucher, Gillian Smith, and Yunus Doğan Telliel. 2024. Is resistance futile?: Early career game developers, generative ai, and ethical skepticism. In Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems. 1–13

    2024

  2. [10]

    Virginia Braun and Victoria Clarke. 2006. Using thematic analysis in psychology. Qualitative research in psychology3, 2 (2006), 77–101

    2006

  3. [11]

    Craig Brod. 1984. Technostress: The human cost of the computer revolution.(No Title)(1984)

    1984

  4. [12]

    Frederick P Brooks. 1974. The mythical man-month.Datamation20, 12 (1974), 44–52

    1974

  5. [13]

    Jenna Butler, Jina Suh, Sankeerti Haniyur, and Constance Hadley. 2025. Dear Di- ary: A randomized controlled trial of Generative AI coding tools in the workplace. InProc. Int’l Conf. Software Engineering (ICSE)

    2025

  6. [14]

    Sue Cantrell and Corrie Commisso. [n. d.]. Outcomes over outputs: Why produc- tivity is no longer the metric that matters most. https://www.deloitte.com/us/ en/insights/topics/talent/measuring-productivity.html

  7. [15]

    It would work for me too

    Ruijia Cheng, Ruotong Wang, Thomas Zimmermann, and Denae Ford. 2024. “It would work for me too”: How online communities shape software developers’ trust in AI-powered code generation tools.ACM Transactions on Interactive Intelligent Systems14, 2 (2024), 1–39

    2024

  8. [16]

    Rudrajit Choudhuri, Dylan Liu, Igor Steinmacher, Marco Gerosa, and Anita Sarma. 2024. How far are we? the triumphs and trials of generative ai in learning software engineering. InProceedings of the IEEE/ACM 46th international conference on software engineering. 1–13

    2024

  9. [17]

    Rudrajit Choudhuri, Bianca Trinkenreich, Rahul Pandita, Eirini Kalliamvakou, Igor Steinmacher, Marco Gerosa, Christopher Sanchez, and Anita Sarma. 2024. What Guides Our Choices? Modeling Developers’ Trust and Behavioral Intentions Towards GenAI.arXiv preprint arXiv:2409.04099(2024)

    2024 arXiv

  10. [18]

    Rudrajit Choudhuri, Bianca Trinkenreich, Rahul Pandita, Eirini Kalliamvakou, Igor Steinmacher, Marco Gerosa, Christopher Sanchez, and Anita Sarma. 2025. What Needs Attention? Prioritizing Drivers of Developers’ Trust and Adoption of Generative AI.arXiv preprint arXiv:2505.17418(2025)

    2025

  11. [19]

    Michael Chui, Eric Hazan, Roger Roberts, Alex Singla, and Kate Smaje. 2023. The economic potential of generative AI. (2023)

    2023

  12. [20]

    2014.Basics of qualitative research: Techniques and procedures for developing grounded theory

    Juliet Corbin and Anselm Strauss. 2014.Basics of qualitative research: Techniques and procedures for developing grounded theory. Sage publications

    2014

  13. [21]

    1993.AI: the tumultuous history of the search for artificial intelli- gence

    Daniel Crevier. 1993.AI: the tumultuous history of the search for artificial intelli- gence. Basic Books, Inc

    1993

  14. [22]

    2024.Octoverse: The state of open source and rise of AI in 2023

    Kyle Daigle and GitHub Staff. 2024.Octoverse: The state of open source and rise of AI in 2023. Technical Report. GitHub. https://github.blog/news-insights/ research/the-state-of-open-source-and-ai/

    2024

  15. [23]

    Fred D Davis, Richard P Bagozzi, and Paul R Warshaw. 1989. Technology accep- tance model.J Manag Sci35, 8 (1989), 982–1003

    1989

  16. [24]

    Fabrizio Dell’Acqua, Edward McFowland III, Ethan R Mollick, Hila Lifshitz-Assaf, Katherine Kellogg, Saran Rajendran, Lisa Krayer, François Candelon, and Karim R Lakhani. 2023. Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on kno...

    2023

  17. [25]

    2018.Out of the Crisis, reissue

    W Edwards Deming. 2018.Out of the Crisis, reissue. MIT press

    2018

  18. [26]

    Paul Denny, James Prather, Brett A Becker, James Finnie-Ansley, Arto Hellas, Juho Leinonen, Andrew Luxton-Reilly, Brent N Reeves, Eddie Antonio Santos, and Sami Sarsa. 2024. Computing education in the era of generative AI.Commun. ACM67, 2 (2024), 56–67

    2024

  19. [27]

    2011.The Sage handbook of qualitative research

    Norman K Denzin and Yvonna S Lincoln. 2011.The Sage handbook of qualitative research. sage

    2011

  20. [28]

    James R Detert, Roger G Schroeder, and John J Mauriel. 2000. A framework for linking culture and improvement initiatives in organizations.Academy of management Review25, 4 (2000), 850–863. Conference’17, July 2017, Washington, DC, USA Miller et al

    2000

  21. [29]

    K Anders Ericsson, Ralf T Krampe, and Clemens Tesch-Römer. 1993. The role of deliberate practice in the acquisition of expert performance.Psychological review 100, 3 (1993), 363

    1993

  22. [30]

    Edward Feigenbaum and Howard Shrobe. 1993. The Japanese national Fifth Generation project: introduction, survey, and evaluation.Future Generation Computer Systems9, 2 (1993), 105–117

    1993

  23. [31]

    Bent Flyvbjerg. 2006. Five misunderstandings about case-study research.Quali- tative inquiry12, 2 (2006), 219–245

    2006

  24. [32]

    Jill J Francis et al. 2010. What is an adequate sample size? Operationalising data saturation for theory-based interview studies.Psychology and Health(2010)

    2010

  25. [33]

    Anna Mette Fuglseth and Øystein Sørebø. 2014. The effects of technostress within the context of employee use of ICT.Computers in human behavior40 (2014), 161–170

    2014

  26. [34]

    Janet Fulk. 1993. Social construction of communication technology.Academy of Management journal36, 5 (1993), 921–950

    1993

  27. [35]

    Janet Fulk, Joseph Schmitz, and Charles W Steinfield. 1990. A social influence model of technology use. InOrganizations and communication technology. SAGE Publications, Inc., 117–140

    1990

  28. [36]

    Fulvio Gaudioso, Ofir Turel, and Carlo Galimberti. 2017. The mediating roles of strain facets and coping strategies in translating techno-stressors into adverse job outcomes.Computers in Human Behavior69 (2017), 189–196

    2017

  29. [37]

    Katy Ilonka Gero, Zahra Ashktorab, Casey Dugan, Qian Pan, James Johnson, Werner Geyer, Maria Ruiz, Sarah Miller, David R Millen, Murray Campbell, et al

  30. [38]

    Dale L Goodhue and Ronald L Thompson. 1995. Task-technology fit and individ- ual performance.MIS quarterly(1995), 213–236

    1995

  31. [39]

    Egon Guba. 1979. Naturalistic inquiry.Improving Human Performance Qtrly. (1979)

    1979

  32. [40]

    William Harding and Matthew Kloster. 2024. Coding on copilot: 2023 data suggests downward pressure on code quality.https://www. gitclear. com/cod- ing_on_copilot_data_shows_ais_downward_pressure_on_code_quality/(2024)

    2024

  33. [41]

    1994.Organizational linkages: Understanding the productivity paradox

    Douglas H Harris. 1994.Organizational linkages: Understanding the productivity paradox. National Academies Press

    1994

  34. [42]

    1997.Mind Design II

    John Haugeland. 1997.Mind Design II. MIT Press

    1997

  35. [43]

    Heather A Haveman and Rachel Wetts. 2019. Organizational theory: From classical sociology to the 1970s.Sociology Compass13, 3 (2019), e12627

    2019

  36. [44]

    Robert R Hoffman, Shane T Mueller, Gary Klein, and Jordan Litman. 2023. Mea- sures for explainable AI: Explanation goodness, user satisfaction, mental models, curiosity, trust, and human-AI performance.Frontiers in Computer Science5 (2023), 1096257

    2023

  37. [45]

    Le Van Huy, Hien TT Nguyen, Tan Vo-Thanh, Nguyen Huu Thai Thinh, Tran Thi Thu Dung, et al. 2024. Generative AI, why, how, and outcomes: A user adoption study.AIS Transactions on Human-Computer Interaction16, 1 (2024), 1–27

    2024

  38. [46]

    Brittany Johnson, Christian Bird, Denae Ford, Nicole Forsgren, and Thomas Zimmermann. 2023. Make your tools sparkle with trust: The PICSE framework for trust in software tools. In2023 IEEE/ACM 45th International Conference on Software Engineering: Software Engineering in Pract...

    2023

  39. [47]

    Matthew R Jones and Helena Karsten. 2008. Giddens’s structuration theory and information systems research.MIS quarterly(2008), 127–157

    2008

  40. [48]

    Mladan Jovanovic and Mark Campbell. 2022. Generative artificial intelligence: Trends and prospects.Computer55, 10 (2022), 107–112

    2022

  41. [49]

    Eirini Kalliamvakou. 2024. A developer’s second brain: Reducing complexity through partnership with AI

    2024

  42. [50]

    Ranim Khojah, Mazen Mohamad, Philipp Leitner, and Francisco Gomes de Oliveira Neto. 2024. Beyond code generation: An observational study of chatgpt usage in software engineering practice.Proceedings of the ACM on Software Engineering1, FSE (2024), 1819–1840

    2024

  43. [51]

    Katherine J Klein and Steve WJ Kozlowski. 2000. A multilevel approach to theory and research in organizations: Contextual, temporal, and emergent processes. Multilevel theory, research, and methods in organizations: Foundations, extensions, and new directions(2000), 3–90

    2000

  44. [52]

    Amy J Ko. 2019. Why we should not measure productivity. InRethinking Productivity in Software Engineering. Springer

    2019

  45. [53]

    Paweł Korzyński, Susana Costa e Silva, Anna Maria Górska, and Grzegorz Mazurek. 2024. Trust in AI and top management support in generative-AI adoption.Journal of Computer Information Systems(2024), 1–15

    2024

  46. [54]

    Will I be replaced?

    Mohammad Amin Kuhail, Sujith Samuel Mathew, Ashraf Khalil, Jose Berengueres, and Syed Jawad Hussain Shah. 2024. “Will I be replaced?” Assessing ChatGPT’s effect on software development and programmer perceptions of AI tools.Science of Computer Programming235 (2024), 103111

    2024

  47. [55]

    Sreejith Kurup and Vivek Gupta. 2022. Factors influencing the AI adoption in organizations.Metamorphosis21, 2 (2022), 129–139

    2022

  48. [56]

    Stefano Lambiase, Gemma Catolino, Fabio Palomba, Filomena Ferrucci, and Daniel Russo. 2024. Investigating the role of cultural values in adopting large language models for software engineering.ACM Transactions on Software Engi- neering and Methodology(2024)

    2024

  49. [57]

    Stefano Lambiase, Gemma Catolino, Fabio Palomba, Filomena Ferrucci, and Daniel Russo. 2025. Exploring Individual Factors in the Adoption of LLMs for Specific Software Engineering Tasks.arXiv preprint arXiv:2504.02553(2025)

    2025 arXiv

  50. [58]

    Shoo K Lee, Sukhy K Mahl, and Brian H Rowe. 2021. The Induced Productivity Decline hypothesis: more physicians, higher compensation and fewer services. Healthcare Policy(2021)

    2021

  51. [59]

    Sarah Lewis. 2015. Qualitative inquiry and research design: Choosing among five approaches.Health promotion practice16, 4 (2015), 473–475

    2015

  52. [60]

    Ze Shi Li, Nowshin Nawar Arony, Ahmed Musa Awon, Daniela Damian, and Bowen Xu. 2024. AI tool use and adoption in software development by individuals and organizations: a grounded theory study.arXiv preprint arXiv:2406.17325 (2024)

    2024 arXiv

  53. [61]

    Jenny T Liang, Chenyang Yang, and Brad A Myers. 2024. A large-scale survey on the usability of ai programming assistants: Successes and challenges. In Proceedings of the 46th IEEE/ACM international conference on software engineering. 1–13

    2024

  54. [62]

    Bernd Marcus and Astrid Schütz. 2005. Who are the people reluctant to participate in research? Personality correlates of four different types of nonresponse as inferred from self-and observer ratings.Journal of personality(2005)

    2005

  55. [63]

    2025.Superagency in the workplace: Empowering people to unlock AI’s full potential

    Hannah Mayer, Michael Chui, and Roger Roberts. 2025.Superagency in the workplace: Empowering people to unlock AI’s full potential. Technical Report. McKinsey Digital. https://www.mckinsey.com/capabilities/mckinsey- digital/our-insights/superagency-in-the-workplace-empowering-p...

    2025

  56. [64]

    Microsoft. [n. d.]. AI at Work Is Here. Now Comes the Hard Part. https://www.microsoft.com/en-us/worklab/work-trend-index/ai-at-work-is- here-now-comes-the-hard-part. Accessed Mar. 2025

    2025

  57. [65]

    2014.Fundamentals of Qualitative Data Analysis

    Matthew B Miles, A Michael Huberman, and Johnny Saldana. 2014.Fundamentals of Qualitative Data Analysis. Sage Los Angeles, CA

    2014

  58. [66]

    G Ayorkor Mills-Tettey, Anthony Stentz, and M Bernardine Dias. 2007. The dynamic hungarian algorithm for the assignment problem with changing costs. Robotics Institute, Pittsburgh, PA, Tech. Rep. CMU-RI-TR-07-277 (2007)

    2007

  59. [67]

    Cal Newport. 2021. The Frustration with Productivity Culture. https://www.newyorker.com/culture/office-space/the-frustration-with- productivity-culture. Accessed: 2025-06-04

    2021

  60. [68]

    1994.Brain Makers

    Harvey Newquist. 1994.Brain Makers. Editors & Engineers, Limited

    1994

  61. [69]

    Lorelli S Nowell, Jill M Norris, Deborah E White, and Nancy J Moules. 2017. Thematic analysis: Striving to meet the trustworthiness criteria.International journal of qualitative methods16, 1 (2017), 1609406917733847

    2017

  62. [70]

    Wanda Janina Orlikowski. 1999. Technologies-in-practice: an enacted lens for studying technology in organizations. (1999)

    1999

  63. [71]

    Wanda J Orlikowski. 2000. Using technology and constituting structures: A practice lens for studying technology in organizations.Organization science11, 4 (2000), 404–428

    2000

  64. [72]

    Wanda J Orlikowski and Debra C Gash. 1994. Technological frames: making sense of information technology in organizations.ACM Transactions on Information Systems (TOIS)12, 2 (1994), 174–207

    1994

  65. [73]

    Sida Peng, Eirini Kalliamvakou, Peter Cihon, and Mert Demirer. 2023. The impact of ai on developer productivity: Evidence from github copilot.arXiv preprint arXiv:2302.06590(2023)

    2023 arXiv

  66. [74]

    Matt Perman. [n. d.]. The Six Major Factors that Determine Knowledge Worker Productivity. https://www.whatsbestnext.com/2010/02/the-six-major-factors- that-determine-knowledge-worker-productivity/. Accessed Mar. 2025

    2010

  67. [75]

    Kalyan Prasad Agrawal. 2024. Towards adoption of generative AI in organiza- tional settings.Journal of Computer Information Systems64, 5 (2024), 636–651

    2024

  68. [76]

    Steven G Rogelberg et al. 2003. Profiling active and passive nonrespondents to an organizational survey.Jrnl. of Applied Psych.(2003)

    2003

  69. [77]

    Everett M Rogers, Arvind Singhal, and Margaret M Quinlan. 2014. Diffusion of innovations. InAn integrated approach to communication theory and research. Routledge, 432–448

    2014

  70. [78]

    2002.Strategic computing: DARPA and the quest for machine intelligence, 1983-1993

    Alex Roland and Philip Shiman. 2002.Strategic computing: DARPA and the quest for machine intelligence, 1983-1993. MIT Press

    2002

  71. [79]

    Daniel Russo. 2024. Navigating the complexity of generative ai adoption in software engineering.ACM Transactions on Software Engineering and Methodology 33, 5 (2024), 1–50

    2024

  72. [80]

    Joseph Schmitz and Janet Fulk. 1991. Organizational colleagues, media richness, and electronic mail: A test of the social influence model of technology use. Communication research18, 4 (1991), 487–523

    1991

  73. [81]

    SciPy. [n. d.]. Optimization (scipy.optimize). https://docs.scipy.org/doc/scipy/ tutorial/optimize.html. Accessed Apr. 2025

    2025

  74. [82]

    Agnia Sergeyuk, Ilya Zakharov, Ekaterina Koshchenko, and Maliheh Izadi. 2025. Human-AI Experience in Integrated Development Environments: A Systematic Literature Review.arXiv preprint arXiv:2503.06195(2025)

    2025

  75. [83]

    Maybe We Need Some More Examples:

    Anastassiya Sichkarenko. 2024. The State of Developer Ecosystem 2024: The Unstoppable Rise of AI, Leading Languages, and Impact on Developer Experience. https://blog.jetbrains.com/team/2024/12/11/the-state-of-developer- ecosystem-2024-unveiling-current-developer-trends-the-uns...

    2024

  76. [84]

    Ningzhi Tang, Meng Chen, Zheng Ning, Aakash Bansal, Yu Huang, Collin McMil- lan, and T Li. 2023. An empirical study of developer behaviors for validating and repairing ai-generated code. In13th Workshop on the Intersection of HCI and PL

    2023

  77. [85]

    Monideepa Tarafdar, Qiang Tu, Bhanu S Ragu-Nathan, and TS Ragu-Nathan

  78. [86]

    2004.Scientific management

    Frederick Winslow Taylor. 2004.Scientific management. Routledge

    2004

  79. [87]

    Amirhosein Toosi, Andrea G Bottino, Babak Saboury, Eliot Siegel, and Arman Rahmim. 2021. A brief history of AI: how to prevent another winter (a critical review).PET clinics16, 4 (2021), 449–469

    2021

  80. [88]

    2024.AI Won’t Solve Your Developer Productivity Problems for You

    Uplevel. 2024.AI Won’t Solve Your Developer Productivity Problems for You. Techni- cal Report. Uplevel. https://uplevelteam.com/blog/ai-for-developer-productivity

    2024

  81. [89]

    Nelda Vendramin, Giulia Nardelli, and Christine Ipsen. 2021. Task-Technology Fit Theory: An approach for mitigating technostress. InA handbook of theories on designing alignment between people and the office environment. Routledge, 39–53

    2021

  82. [90]

    Viswanath Venkatesh, Michael G Morris, Gordon B Davis, and Fred D Davis

  83. [91]

    Viswanath Venkatesh, James YL Thong, and Xin Xu. 2012. Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology.MIS quarterly(2012), 157–178

    2012

  84. [92]

    Ruotong Wang, Ruijia Cheng, Denae Ford, and Thomas Zimmermann. 2024. Investigating and designing for trust in ai-powered code generation tools. InPro- ceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency. 1475–1493

    2024

  85. [93]

    Karl E Weick. 1990. Technology as equivoque: Sensemaking in new technologies. (1990)

    1990

  86. [94]

    Justin D Weisz, Michael Muller, Jessica He, and Stephanie Houde. 2023. To- ward general design principles for generative AI applications.arXiv preprint arXiv:2301.05578(2023)

    2023 arXiv

  87. [95]

    Justin D Weisz, Michael Muller, Stephanie Houde, John Richards, Steven I Ross, Fernando Martinez, Mayank Agarwal, and Kartik Talamadupula. 2021. Perfection not required? Human-AI partnerships in code translation. InProceedings of the 26th International Conference on Intelligen...

    2021

  88. [96]

    Charley M Wu, Eric Schulz, Timothy J Pleskac, and Maarten Speekenbrink. 2022. Time pressure changes how people explore and respond to uncertainty.Scientific reports(2022)

    2022

  89. [2003]

    User acceptance of information technology: Toward a unified view.MIS quarterly(2003), 425–478

    2003

  90. [2007]

    The impact of technostress on role stress and productivity.Journal of management information systems24, 1 (2007), 301–328

    2007

  91. [2020]

    InProceedings of the 2020 chi conference on human factors in computing systems

    Mental models of AI agents in a cooperative game setting. InProceedings of the 2020 chi conference on human factors in computing systems. 1–12

    2020

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.