CAIC Technology Project Management 3 — Questions and Answers
Question 1: Which technique helps AI project teams estimate story points for data-dependent tasks with high uncertainty?
- Planning poker with explicit uncertainty ranges (Correct answer)
- Fixed-price contracts with no flexibility
- Top-down budgeting from finance only
- Copying estimates from non-AI software projects
Correct answer: Planning poker with explicit uncertainty ranges
Planning poker surfaces team disagreements and allows uncertainty ranges to be encoded in estimates for AI's inherent unknowns.
Question 2: A company is deploying an AI model that will affect hiring decisions. The project manager must ensure the team conducts:
- Only performance benchmarking on test data
- A bias and fairness audit across protected demographic groups (Correct answer)
- Exclusively cost-optimization analysis
- Marketing copy review for the product announcement
Correct answer: A bias and fairness audit across protected demographic groups
Hiring AI is subject to employment law and ethical standards requiring bias audits across protected groups before deployment.
Question 3: In MLOps, a CI/CD pipeline for machine learning differs from traditional software CI/CD primarily because it must also:
- Compile source code into binaries
- Retrain, evaluate, and validate models as part of the pipeline (Correct answer)
- Manage user authentication flows
- Run only unit tests on business logic
Correct answer: Retrain, evaluate, and validate models as part of the pipeline
ML CI/CD must include automated model retraining and evaluation gates, not just code testing and deployment.
Question 4: Which stakeholder role is MOST critical to define clearly at the start of an AI consulting engagement to prevent scope creep?
- The product owner who controls the backlog and success criteria (Correct answer)
- The cloud infrastructure vendor representative
- The open-source library maintainer
- The external data labeling contractor
Correct answer: The product owner who controls the backlog and success criteria
A clear product owner with authority over scope and success criteria prevents stakeholders from continuously expanding requirements.
Question 5: A project team is selecting between building a custom AI model versus using a pre-trained foundation model API. The PRIMARY project management factor favoring the API approach is:
- Greater control over model weights
- Faster time-to-value and lower initial development cost (Correct answer)
- Ability to train on proprietary data exclusively
- Elimination of all ongoing inference costs
Correct answer: Faster time-to-value and lower initial development cost
Foundation model APIs reduce development cycles and upfront cost, accelerating delivery even though they introduce vendor dependency.
Question 6: Which risk category is unique to AI projects compared to traditional software projects?
- Budget overrun risk
- Training data quality and availability risk (Correct answer)
- Staff turnover risk
- Network latency risk
Correct answer: Training data quality and availability risk
AI projects uniquely depend on data quality, volume, and access, which can derail the entire project if not secured early.
Question 7: When a project retrospective reveals that data labeling took 60% longer than estimated, the corrective action for future AI projects should include:
- Eliminating the labeling phase entirely
- Building labeling time buffers and exploring semi-supervised or active learning approaches (Correct answer)
- Outsourcing all future labeling with no quality review
- Reducing model accuracy requirements to need less data
Correct answer: Building labeling time buffers and exploring semi-supervised or active learning approaches
Adding buffers and exploring label-efficient methods addresses the root cause of data labeling underestimation systematically.
CAIC Technology Project Management 3
Which technique helps AI project teams estimate story points for data-dependent tasks with high uncertainty?