pith. sign in

arxiv: 2604.21958 · v1 · submitted 2026-04-23 · 💻 cs.SE · cs.AI

A systematic review of generative AI usage for IT project management

Ionut Anghel , Tudor Cioara This is my paper

Pith reviewed 2026-05-09 21:06 UTC · model grok-4.3

classification 💻 cs.SE cs.AI
keywords generative AIIT project managementsystematic reviewPRISMAOpenAI GPTprompt engineeringAI agentsproject management tools
0
0 comments X

The pith

A systematic review finds OpenAI GPT dominates generative AI in IT project management through prompt engineering, placing the field at an exploratory stage.

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

This paper applies the PRISMA methodology to synthesize existing studies on generative AI applications in IT project management, covering techniques, uses, adoption patterns, limitations, and tool integrations across process groups. The review concludes that OpenAI's GPT models appear in most included work but mainly through basic prompt engineering rather than deeper customization or training. This pattern leads the authors to characterize current research as early-stage exploration instead of established practice. The paper closes by naming three concrete directions for progress: AI agents tailored to specific process groups, agents aligned with project roles, and hybrid networks that combine AI with human oversight. A reader would care because these findings clarify where effort is currently concentrated and where gaps in integration and sophistication remain.

Core claim

The analysis reveals a clear dominance of OpenAI's GPT in the included studies but relying primarily on prompt engineering, suggesting that research in this area remains at an exploratory stage. Finally, it identifies and discusses three promising research directions for AI-enabled project management, including process group-specific AI agents, project role-based AI agents, and hybrid collaborative networks that enable human-guided orchestration.

What carries the argument

The PRISMA systematic review protocol, which filters and synthesizes literature on generative AI techniques and their application to IT project management process groups and tools.

If this is right

  • Process group-specific AI agents would target distinct phases such as initiation, planning, execution, monitoring, and closure with tailored capabilities.
  • Project role-based AI agents would deliver support differentiated by stakeholder positions including project managers, team members, and sponsors.
  • Hybrid collaborative networks would combine multiple AI components under human direction to orchestrate workflows across existing project tools.
  • Current heavy use of prompt engineering implies limited scalability and calls for more automated or domain-adapted approaches.
  • Integration of generative AI across standard project management tools and process groups is still limited, requiring focused development.

Where Pith is reading between the lines

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

  • The dominance of accessible GPT models may stem more from ease of use than technical superiority, opening room for specialized or open alternatives to gain traction.
  • The three proposed agent directions could be tested first in controlled pilot projects to measure gains in efficiency or decision quality over prompt-only methods.
  • If the exploratory characterization holds, commercial project management platforms may incorporate generative features slowly until evidence of reliability improves.
  • The pattern observed here might recur in adjacent fields such as software engineering or risk management, where prompt-based AI is also common.

Load-bearing premise

The studies located and retained by the PRISMA search and screening steps form a representative sample of all relevant generative AI work in IT project management.

What would settle it

A follow-up search or industry survey that uncovers a large share of studies using non-GPT generative models or techniques beyond prompt engineering in IT project management would undermine the reported dominance and exploratory-stage conclusion.

Figures

Figures reproduced from arXiv: 2604.21958 by Ionut Anghel, Tudor Cioara.

Figure 1
Figure 1. Figure 1: PRISMA 2000 flow diagram for selecting the review articles The subsequent Screening phase involved the application of predefined inclusion & exclusion criteria, which further filtered the records and reduced the number of eligible papers. The main inclusion criteria that narrowed the number of articles to 98 were: (i) publication year between 2021-2026, (ii) publication type (article or proceeding paper), … view at source ↗
Figure 2
Figure 2. Figure 2: Selected articles statistics The analysis reveals a relatively balanced distribution between open access and subscription-based publications, with 14 papers (47%) available as open access and 16 papers (53%) published under subscription models. The geographic distribution of author affiliations demonstrates the global nature of research in this domain, with contributions from 20 different countries across … view at source ↗
read the original abstract

This paper aims to synthesize current knowledge on generative AI in IT project management using the PRISMA methodology to provide researchers with a comprehensive perspective on techniques, applications, adoption trends, limitations, and integration across project management tools and process groups. The analysis reveals a clear dominance of OpenAI's GPT in the included studies but relying primarily on prompt engineering, suggesting that research in this area remains at an exploratory stage. Finally, it identifies and discusses three promising research directions for AI-enabled project management, including process group-specific AI agents, project role-based AI agents, and hybrid collaborative networks that enable human-guided orchestration.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. This manuscript presents a PRISMA-based systematic review synthesizing research on generative AI applications in IT project management. It reports a clear dominance of OpenAI GPT models (primarily via prompt engineering) across the included studies, characterizes the field as still exploratory, and proposes three future directions: process group-specific AI agents, project role-based AI agents, and hybrid collaborative human-AI networks.

Significance. If the retrieval and screening process yields a representative sample, the review could usefully map an emerging intersection of generative AI and project management, highlighting the current reliance on lightweight prompt-based techniques and surfacing concrete research avenues. The explicit call-out of three directions provides a constructive forward-looking element.

major comments (1)
  1. [Methods] Methods section (PRISMA protocol description): No quantitative details are supplied on databases searched, exact search strings, number of records identified, duplicates removed, titles/abstracts screened, full texts assessed, or final inclusions. Without these, the claim of 'clear dominance' of OpenAI GPT cannot be evaluated for robustness against possible retrieval bias arising from rapidly evolving terminology or non-standard keywords.
minor comments (2)
  1. [Abstract] Abstract: The summary of findings would be strengthened by including at least the final number of included studies and a brief note on the dominant technique (prompt engineering) to give readers an immediate sense of scope.
  2. [Discussion / Future Work] The three proposed research directions are presented at a high level; a short paragraph sketching concrete evaluation metrics or example use cases for each would improve actionability without lengthening the paper substantially.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback. We address the single major comment below and will revise the manuscript to improve methodological transparency.

read point-by-point responses

  1. Referee: [Methods] Methods section (PRISMA protocol description): No quantitative details are supplied on databases searched, exact search strings, number of records identified, duplicates removed, titles/abstracts screened, full texts assessed, or final inclusions. Without these, the claim of 'clear dominance' of OpenAI GPT cannot be evaluated for robustness against possible retrieval bias arising from rapidly evolving terminology or non-standard keywords.

    Authors: We agree that the Methods section currently lacks the quantitative PRISMA flow details. In the revised manuscript we will add a PRISMA flow diagram together with the exact numbers for databases searched, search strings, records identified, duplicates removed, titles/abstracts screened, full texts assessed, and final inclusions. This addition will allow readers to assess the robustness of the reported dominance of OpenAI GPT models and any potential retrieval bias. revision: yes

Circularity Check

0 steps flagged

No circularity: literature synthesis with no derivations or self-referential definitions

full rationale

The paper is a systematic review employing the PRISMA methodology to synthesize existing studies on generative AI in IT project management. Its central claims (GPT dominance via prompt engineering, exploratory stage of research, and suggested future directions) are derived directly from the content of the included papers rather than from any internal equations, fitted parameters, predictions, or self-citations that reduce to the paper's own inputs. No self-definitional loops, fitted-input predictions, uniqueness theorems, or ansatzes are present. The PRISMA process is a standard external protocol for literature retrieval; any limitations in sample representativeness affect generalizability but do not constitute circularity in the derivation chain. The work is self-contained as a synthesis against external benchmarks (the reviewed literature).

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper rests on the established PRISMA guidelines for evidence synthesis and the assumption that published literature adequately captures current generative AI usage patterns. No free parameters, new entities, or ad-hoc axioms are introduced beyond standard review methodology.

axioms (1)
  • standard math PRISMA methodology provides a rigorous and reproducible framework for systematic literature reviews
    Invoked in the abstract as the chosen synthesis method; this is a domain-standard assumption in evidence-based research.

pith-pipeline@v0.9.0 · 5388 in / 1326 out tokens · 39968 ms · 2026-05-09T21:06:42.393290+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

  1. [1]

    Abdelzaher, T., Hu, Y ., Kara, D., et al. (2025). The bottlenecks of AI: Challenges for embedded and real- time research in a data-centric age. Real-Time Systems, 61, 185–236. https://doi.org/10.1007/s11241-025- 09452-w Abdulghafour, M., & Chirchir, B. (2024). Challenges of integrating artificial intelligence in software project planning: A systematic lit...

    work page doi:10.1007/s11241-025- 2025

  2. [2]

    Y ., Temizel, T

    https://doi.org/10.3390/fi16050177 Akgül, N. Y ., Temizel, T. T., Top, Ö. Ö., & Akman, P. D. (2025). Aligning data debt with AI -integrated software project lifecycle processes: A standard-based mapping approach. In 2025 IEEE/ACM International Conference on Technical Debt (TechDebt) (pp. 1 –11). IEEE. https://doi.org/10.1109/TechDebt66644.2025.00008 Alam,...

    work page doi:10.3390/fi16050177 2025

  3. [3]

    Alliata, Z., Singhal, T., & Bozagiu, A. M. (2024). The AI Scrum Master: Using large language models (LLMs) to automate agile project management tasks. In Agile Processes in Software Engineering and Extreme Programming – Workshops (XP

    work page 2024

  4. [4]

    (pp. 1–15). Springer. https://doi.org/10.1007/978-3-031- 72781-8_12 Almeida, P. M., Fernandes, G., & Santos, J. M. R. C. A. (2025). Artificial intelligence tools for project management: A knowledge -based perspective. Project Leadership and Society, 6 , 100196. https://doi.org/10.1016/j.plas.2025.100196 Aramali, V ., Cho, N., Pande, F., Al-Mhdawi, M. K. S...

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

  5. [5]

    AsanaAI. (2026). Get started with Asana AI. https://help.asana.com/s/article/get-started-with-asana-ai Assalaarachchi, L. I., Masood, Z., Hoda, R., & Grundy, J. (2025). Generative AI for software project management: Insights from a review of software practitioner literature. IEEE Software . https://doi.org/10.1109/MS.2025.3619936 Atlassian. (2025, January...

    work page doi:10.1109/ms.2025.3619936 2026

  6. [6]

    https://doi.org/10.3390/electronics14010087 Colavito, G., Lanubile, F., & Novielli, N. (2025). Benchmarking large language models for automated labeling. Information and Software Technology, 184, 107758. https://doi.org/10.1016/j.infsof.2025.107758 Dhar, R., Vaidhyanathan, K., & Varma, V . (2024). Leveraging generative AI for architecture knowledge manage...

    work page doi:10.3390/electronics14010087 2025

  7. [7]

    https://doi.org/10.3390/app14052096 Li, K., Rollins, J., & Yan, E. (2018). Web of Science use in published research. Scientometrics, 115, 1–20. Li, X., Cheng, Y ., Møller, C., & Lee, J. (2025). Data issues in industrial AI systems. Computers in Industry, 173, 104361. https://doi.org/10.1016/j.compind.2025.104361 Mahbub, T., et al. (2024). Can GPT -4 aid i...

    work page doi:10.3390/app14052096 2018

  8. [8]

    Papagiannidis, E., Mikalef, P., & Conboy, K. (2025). Responsible artificial intelligence governance. Journal of Strategic Information Systems, 34(2), 101885. https://doi.org/10.1016/j.jsis.2024.101885 Pasam, V . K., Pati, S., & Hernandez, C. T. (2025). AI -powered comment triage. In SAC 2025 (pp. 971– 979). ACM. PMBOK Guide. (2026). A guide to the project...

    work page doi:10.1016/j.jsis.2024.101885 2025

  9. [9]

    I., & Karkee, M

    https://doi.org/10.3390/buildings15071130 Sapkota, R., Roumeliotis, K. I., & Karkee, M. (2026). AI agents vs. agentic AI. Information Fusion, 126, 103599. https://doi.org/10.1016/j.inffus.2025.103599 Sarim, M., et al. (2025). Generating reliable software project task flows. Scientific Reports, 15 , 35194. https://doi.org/10.1038/s41598-025-19170-9 Scrum S...

    work page doi:10.3390/buildings15071130 2026

  10. [10]

    ChatGPT Labs

    (2026). ChatGPT Labs. https://chatgpt.com/g/g-PSusbDrAK-scrum- sage-zen-edition-version-2 Sheikhaei, M. S., et al. (2024). An empirical study on large language models for SATD identification. Empirical Software Engineering, 29,

    work page 2026

  11. [11]

    Singh, P., Choudhary, R., & Zhou, L. (2025). AI -based project planning using generative models. arXiv. https://arxiv.org/abs/2510.10887 Slama, F., & Lemire, D. (2025). Enhancing developer productivity. In IDS 2025 (pp. 39 –45). IEEE. https://doi.org/10.1109/IDS66066.2025.00011 Su, X., & Ayob, A. H. (2025). Artificial intelligence in project success. Info...

    work page doi:10.1109/ids66066.2025.00011 2025

  12. [12]

    https://doi.org/10.3390/info16080682 Taboada, I., Daneshpajouh, A., Toledo, N., & de Vass, T. (2023). Artificial intelligence enabled project management. Applied Sciences, 13,

    work page doi:10.3390/info16080682 2023

  13. [13]

    Wang, H., Zhao, L., & Chen, Y . (2025). Artificial intelligence in project management. Journal of Project Analytics, 5, 15–28. Wysocki, W., & Ochodek, M. (2026). Leveraging LLM-based data augmentation. Journal of Systems and Software, 231, 112641. https://doi.org/10.1016/j.jss.2025.112641 Yang, H., Zhou, Y ., Liang, T., & Kuang, L. (2025). ChatDL: An LLM-...

    work page doi:10.1016/j.jss.2025.112641 2025