Session Details

Lessons About What Makes AI Projects Different

Abstract

What happens when an experienced project management professional decides to build an AI-powered product?

Drawing on an extensive experience developing a personalized AI tutor that teaches project management by adapting to each learner's professional context, this session will share some lessons that challenged key assumptions about managing projects.

Unlike traditional software, generative AI introduces a different level of uncertainty. Requirements are harder to define, outputs are not always predictable, improvements can unexpectedly affect previously working functionality, and quality is no longer simply a matter of passing or failing a test.

Through practical examples from developing and testing a real AI product, we’ll explore what makes AI projects different and what project managers need to rethink when it comes to planning, requirements, change management, quality assurance, and measuring progress.

The key message: Traditional project management principles remain relevant, but managing AI projects requires a different mindset and approach.

Key Takeaways

Participants will learn:

1. Rethink Requirements. Why AI projects require a shift from specifying exact outputs to defining desired outcomes, behaviors, and boundaries.

2. Manage Unpredictability. How to deal with AI systems that can produce different results from the same inputs, and why consistency doesn't necessarily mean identical outputs.

3. Approach Change Differently. Why improving one aspect of an AI product can unexpectedly affect other functionalities, requiring a different approach to change control and regression testing.

4. Redefine Quality Assurance. Why traditional pass/fail testing is insufficient for generative AI and how to evaluate quality across multiple dimensions.

5. Measure Progress Differently. Why delivering features doesn't always mean delivering value, and how to focus on demonstrated capabilities, reliability, and user outcomes.