Olivier Denaus highlights the critical role of clear requirements in testing, the Augmented Tester Model for balancing AI and human judgment, and challenges like overreliance on automation. He stresses that AI cannot replace human expertise in detecting business logic flaws, emphasizing the need for rigorous requirement analysis and human oversight in testing workflows.

How To Test This?

interview episodes where Mamadou N’diaye talks with with software testing experts

• https://podcasters.spotify.com/pod/show/spidey1944
• https://www.youtube.com/@HowToTestThis
• https://www.linkedin.com/in/mamadou-ndiaye-consultant/

Episode Details

• Show Notes: https://podcasters.spotify.com/pod/show/spidey1944/episodes/How-to-Test-With-ISTQB-In-the-Age-of-AI---Olivier-Denoo-e3ks9cb
• Published: 2026-06-16T15:01:38Z
• Duration: 00:44:06
• Author: Mamadou N’diaye

Overview

The podcast discusses software testing practices, emphasizing Olivier Denaus three decades of expertise in consulting, education, and governance, including leadership roles in ISTQB and contributions to AI-related testing syllabi. Key themes include the critical role of well-defined requirementshighlighting that 60% of defects stem from incomplete or ambiguous specificationsand the need for human oversight in testing, even as AI tools integrate into workflows. The “Augmented Tester Model” is presented as a framework to leverage AI for repetitive tasks while preserving human judgment in complex decision-making. Industry challenges include overreliance on automation, neglecting human elements, and underemphasizing requirement quality. Denau critiques AIs limitations, arguing that human “stupidity” (misuse or overestimation of AI capabilities) poses greater risks than AI itself, and stresses that AI cannot replace human expertise in detecting business logic flaws or ambiguous requirements.

The discussion also explores ISTQBs global structure, its 1.2 million certified members, and efforts to standardize testing education while adapting to trends like generative AI. Testing AI itself is framed as a critical area, requiring specialized methods like metamorphic testing and addressing biases. The podcast highlights the cyclical nature of technological hype, noting parallels between current AI enthusiasm and past automation trends, with a caution against overstating AIs capabilities. Career advice for testers includes foundational knowledge through certifications like ISTQB, engaging with evolving technologies like generative AI, and maintaining a human-centered approach. Practical steps for integrating AI into testing, such as structured prompting and validating AI outputs, are outlined, alongside the importance of requirement analysis frameworks like IREP. Overall, the conversation underscores the enduring relevance of human expertise in testing, even as tools and methodologies evolve.

What If

• What if you leveraged AI to audit and improve requirement clarity before starting a project?

  • Move: Use AI (e.g., GenAI tools) to analyze your projects requirements for ambiguity, missing edge cases, or conflicting definitions (e.g., vague terms like “beautiful” or “blue”).
  • Why Now?: 60% of defects stem from poor requirements (Olivier Denau), and AI can rapidly prioritize gaps. This reduces rework and aligns with the “augmented tester” model by automating repetitive analysis.
  • Expected Upside: Fewer defects downstream, faster iteration, and a lower risk of AI hallucinations in test scenarios caused by unclear requirements.

• What if you built an AI-assisted test generation workflow to prioritize human judgment on complex logic?

  • Move: Use AI to draft test cases for routine scenarios (e.g., form validation) while manually reviewing AI-generated logic for edge cases (e.g., negative ticket pricing).
  • Why Now?: The “augmented tester” model emphasizes AI for low-value tasks (Denau), freeing you to focus on business-critical logic. Current AI hype cycles demand practical integration to avoid overreliance.
  • Expected Upside: 30% faster test creation, reduced flaky test ratios, and fewer missed edge cases due to human oversight of AIs limitations.

• What if you formalized a requirements engineering process using frameworks like IREP to preempt AI/automation errors?

  • Move: Adopt IREP (Independent Requirements Engineering Process) to structure requirement reviews, ensuring clarity on definitions, constraints, and stakeholder expectations before prototyping.
  • Why Now?: Poor requirements are a leading cause of defects, and IREP aligns with the French testing boards standards. This also mitigates AI risks by ensuring inputs to automated tools are precise.
  • Expected Upside: 20% reduction in rework from requirement ambiguity, better alignment with agile documentation principles, and smoother AI integration for testing.

Takeaway

1. Prioritize Requirement Clarity: Dedicate significant time to refining and validating software requirements with stakeholders to reduce defects, as 60% of issues stem from ambiguous or incomplete requirements. Use tools like IREP frameworks to identify gaps and ensure alignment across teams.

2. Adopt the Augmented Tester Model: Integrate AI tools for repetitive tasks (e.g., test case generation, data analysis) but maintain human oversight for judgment, ethics, and complex decision-making. Avoid over-reliance on automation or AI outputs without verification.

3. Invest in ISTQB and AI-Specific Certifications: Pursue foundational ISTQB certification (e.g., Foundation Level) to establish a shared vocabulary and understanding of testing principles. For advanced roles, explore AI-focused syllabi like GenAI or CTAI to validate expertise in AI-driven testing practices.

4. Validate AI Outputs with Human Judgment: When using AI for testing or requirements analysis, apply structured prompting techniques and validate outputs for hallucinations, logic flaws, or edge cases. Use anonymized data for experiments to avoid security risks and maintain data confidentiality.

5. Stay Proactive in Skill Development: Continuously update skills in emerging trends (e.g., GenAI, DevOps, Agile) through iterative learning, collaboration with cross-functional teams, and sharing insights with peers. Avoid obsolescence by actively engaging with new tools and methodologies like AI-augmented testing workflows.