Part A: The Project Premortem
1. Title
AI Initiative Risk Assessment: The Project Premortem
2. Objective
The objective of this assignment is to provide a hands-on experience in identifying, analyzing, and planning for the potential risks of an AI initiative, moving beyond technical metrics to consider strategic, operational, and ethical consequences.
3. Tools/Resources Needed
- Provided Scenario Brief: The description of the AI initiative below.
- Generative AI Tool (Optional): Access to an LLM like ChatGPT, Claude, or Gemini to assist in brainstorming.
- Table Tool: Word, Excel, or similar software to create the Risk Register.
4. Description of Assignment
This lab directly applies the premortem technique mentioned in Chapters 4 and 5. Students will be given a brief for a new AI initiative and will be tasked with imagining its complete failure in order to proactively identify potential risks.
Scenario:
A retail company plans to deploy an AI-powered dynamic pricing system for its e-commerce site. The AI will adjust prices in real-time based on user demand, competitor pricing, inventory levels, and even the user's Browse history. The operating logic is that by raising the price slightly at the point of purchase, motivated buyers will proceed with the purchase. The system is designed to create an "urgency score" for each customer transaction (i.e., an active Browse session and the matching shopping cart) and use that score to adjust prices in real-time.
Tasks:
- Imagine Failure: Start with the following premise: "It’s one year from now. The dynamic pricing project has been a complete disaster. It has been shut down, and it has cost the company dearly in both money and reputation."
- Generate Causes: Brainstorm all the possible reasons for this failure. Think broadly, considering technical issues, performance of the urgency score model, customer backlash, competitor reactions, and internal resistance. You may use generative AI to help think broadly about this failure scenario. Consider the following questions:
- What technical flaws could have caused this? (e.g., model drift, bad data)
- How did customers react? Why?
- Which internal department or stakeholder group is secretly happy the project failed, and what were their original concerns?
- What ethical "tigers" did we miss while hunting for efficiency "hares"?
- Create a Risk Register: Convert your list of failure causes into a formal risk register. For each identified risk, you must categorize it (e.g., Operational, Reputational, Financial, Ethical) and propose a potential mitigation strategy.
5. Deliverables
- List of Failure Causes: A detailed list of at least 10 potential causes for the project's failure, covering a diverse range of technical, ethical, and business-related issues.