Project Management

Please login or join to subscribe to this thread

What auditability challenges have you encountered when developing AI systems iteratively?

linkedin twitter facebook   Agile   Artificial Intelligence   Ethics  
avatar
Stelian ROMAN Project Manager| MicroSafety Carlingford, New South Wales, Australia

Artificial Intelligence (AI) systems are increasingly being developed through iterative processes, leveraging cycles of prototyping, user feedback, and continuous improvement. While this Agile approach accelerates innovation and adapts to changing requirements, it also introduces unique complexities when it comes to auditability. Auditability refers to the ability to trace, verify, and explain how an AI system was developed, how it functions, and why it produces specific outputs. As AI systems become more integral to critical decision-making in sectors like healthcare, finance, and the public sector, ensuring their auditability is not just a regulatory requirement, but a trust imperative. This blog post explores the challenges and actionable recommendations for maintaining auditability in AI systems that evolve through iterative development.

·What auditability challenges have you encountered when developing AI systems iteratively?

·Which tools or practices have you found most effective in maintaining a clear audit trail?

·How do you see auditability requirements evolving as AI systems become more complex and autonomous?

Blog post "Auditability of AI Systems Developed Iteratively"

ProjectManagement.com - The Agile Enterprise
Sort By:
avatar
Sayed Zaidi Kashif Mekhdi Architect Projects Engineer| Kuwait Oil Company Salmiya, KU, Kuwait
When developing AI systems step by step, one big challenge is keeping track of all changes and decisions. It can be hard to explain why the AI makes certain choices because it learns and changes often. To keep a clear audit trail, we use version control tools and document every update carefully. Also, testing and user feedback are recorded. As AI gets more complex, audit rules will be stricter, and we will need better tools to explain AI actions clearly. How do you keep track of changes in your projects?
avatar
Luis Branco CEO| Business Insight, Consultores de Gestão, Ldª Carcavelos, Lisboa, Portugal
An excellent perspective.

What struck me most is that auditability is often approached as a traceability challenge, when it may increasingly become a decision traceability challenge.

In complex AI systems, organizations can often reconstruct datasets, model versions and technical artifacts.
Yet they may still struggle to explain why a particular approach was chosen, which alternatives were rejected, who approved the decision and what risks were considered acceptable at the time.

As AI systems become more autonomous and consequential, I believe auditability will need to evolve beyond artifact traceability toward decision traceability.

After all, understanding what changed is important.
Understanding why it changed may become even more important.
...
1 reply by Stelian ROMAN
Aug 23, 2026 10:59 PM
Stelian ROMAN
...
Luis Branco I like your link between auditability and decision-making. A self-organised team that has the authority to make decisions should also be responsible for providing the reasons. In my opinion, AI agents should be considered as team members (developers) and their work documented and traceable to a human decision.
avatar
Stelian ROMAN Project Manager| MicroSafety Carlingford, New South Wales, Australia
Jun 24, 2026 5:19 AM
Replying to Luis Branco
...
An excellent perspective.

What struck me most is that auditability is often approached as a traceability challenge, when it may increasingly become a decision traceability challenge.

In complex AI systems, organizations can often reconstruct datasets, model versions and technical artifacts.
Yet they may still struggle to explain why a particular approach was chosen, which alternatives were rejected, who approved the decision and what risks were considered acceptable at the time.

As AI systems become more autonomous and consequential, I believe auditability will need to evolve beyond artifact traceability toward decision traceability.

After all, understanding what changed is important.
Understanding why it changed may become even more important.
Luis Branco I like your link between auditability and decision-making. A self-organised team that has the authority to make decisions should also be responsible for providing the reasons. In my opinion, AI agents should be considered as team members (developers) and their work documented and traceable to a human decision.

Please login or join to reply

Content ID:
ADVERTISEMENTS

Conformity's an obsession with me.

- George Costanza

ADVERTISEMENT

Sponsors







Newsletters

Aug 19: PM Lessons From the Himalayas
Aug 5: Invisible Work, Visible Problems
Jul 22: Your AI Onboarding Problem

See all newsletters