A retail bank is building an AI credit scoring model. The project manager needs to identify the right subject matter experts to ensure the model accurately reflects real-world credit risk.Which approach correctly identifies the most qualified data SMEs?
Answer(s): A
PMI-CPMAI's AI data governance framework emphasizes that data SMEs must combine domain expertise with deep familiarity with the organization's specific data structures and quality characteristics. For credit risk modeling, this means involving both data analysts who understand data pipelines, feature behavior, and quality issues, and financial/credit experts who can validate that input variables and model outputs align with regulatory requirements and risk policies. Selecting SMEs based on availability or generic IT skills ignores the specialized knowledge required to build accurate, compliant credit AI.
An enterprise cybersecurity team is deploying an AI-driven network monitoring system to distinguish between benign alerts and genuine intrusion attempts. The project manager is concerned about both false alarm fatigue and missed attacks.Which ongoing action most effectively addresses these dual risks?
Answer(s): D
PMI-CPMAI AI risk management highlights that false positives and false negatives in security systems require continuous human-AI collaboration, not just algorithm selection. A continuous feedback loop means security analysts review flagged alerts, classify them as true or false, and feed those labels back into the model pipeline—enabling ongoing threshold recalibration and retraining. This closed-loop approach incrementally reduces misclassification rates. Algorithm selection and model combinations are design-time decisions; a feedback loop provides the ongoing governance mechanism that sustains accuracy in a changing threat landscape.
A project team has completed all data preparation activities and is ready to begin model training. The project manager wants to ensure the team has a shared and documented understanding of how model success will be measured.What should the project manager do next?
Answer(s): B
PMI-aligned AI lifecycle guidance calls for explicit documentation of performance metrics and success criteria before model training begins. This ensures that all team members agree on which metrics will be used—accuracy, precision, recall, F1, AUC, business KPIs—and what thresholds constitute acceptable performance. Pre-defined metrics support traceability, objective model evaluation, and transparent go/no-go decisions in Phase V. Since data preparation is described as complete, another data quality assessment is less critical at this point. Go/no-go decisions and scalability reports depend on having metrics defined first.
A regional hospital has deployed an AI-based patient triage chatbot in its emergency intake workflow. Users have filed complaints stating they expected to speak with a nurse but received confusing automated responses instead.What governance control should be implemented to prevent this situation?
Answer(s): C
PMI-CPMAI responsible AI and ethics guidance establishes that user-facing AI systems must clearly disclose their automated nature. A disclosure notice—such as 'You are now interacting with an AI-powered assistant'—manages user expectations, prevents the frustration of mistaken identity, and supports the transparency principle. Diverse training data and advanced algorithms may improve answer quality but do not solve the misidentification issue. Interpretable models address explainability for technical stakeholders, not patient-facing identity transparency.
A project manager is leading the rollout of a new AI platform for a logistics company. The platform must support eight different cognitive capabilities across the business, and the manager is worried that deploying all capabilities simultaneously will overwhelm the organization.What should be the first action?
PMI-CPMAI's guidance on transitioning to AI recommends incremental, phased adoption to control complexity and risk.When multiple cognitive patterns are available, the implementation roadmap should prioritize a limited set of high-value capabilities in early iterations, validate their impact, and expand scope only after each pattern is proven. Deploying all capabilities at once increases coordination complexity, resistance to change, and technical failure risk. The framework explicitly warns against 'big bang' AI deployments and advocates for sequenced releases that deliver value while minimizing operational disruption.
During a workflow analysis, a project team discovers that several steps in an insurance claims process—including form validation, policy lookup, and status updates—follow deterministic, rule-based logic requiring no judgment.Based on this finding, what should the team do with these steps?
PMI-CPMAI clearly distinguishes cognitive AI applications from noncognitive tasks.When project teams identify deterministic, rule-based steps that do not require learning or probabilistic reasoning, the recommended approach is to apply conventional automation technologies—scripts, RPA, rule engines, or workflow systems. Deploying AI to noncognitive tasks introduces unnecessary model risk, maintenance overhead, and governance complexity. The framework calls for 'fit-for-purpose' technology selection: reserve AI for tasks that genuinely require pattern recognition or adaptive decision-making.
A pediatric hospital is developing an AI model to predict which patients are at elevated risk for 30-day readmission. The data science team is identifying data sources and assessing whether the available records are suitable.Which method best supports this objective?
PMI-CPMAI AI data guidance emphasizes that teams must systematically profile candidate data sources to evaluate completeness, consistency, validity, and population coverage before committing to modeling. Data profiling tools allow the team to inspect distributions, missing values, outliers, and anomalies across clinical, administrative, and claims data. They provide the evidence needed to determine whether the dataset is fit for training and deployment. Augmentation, real-time pipelines, and data catalogs all play downstream roles but do not directly provide the initial quality evidence needed to decide whether to proceed.
A shipping company's AI fleet optimization team is working with a mixed dataset including structured GPS telemetry, structured fuel consumption logs, and unstructured driver dispatch notes. The project manager decides to skip the data variety harmonization step to stay on schedule.What is the most likely consequence of this decision?
PMI-CPMAI data preparation guidance states that data variety must be explicitly addressed through normalization, harmonization, and feature extraction so that models receive coherent, compatible inputs. Ignoring data variety leads to misaligned schemas, inconsistent encodings, and improperly processed unstructured content. The guidance notes that such issues manifest as degraded model performance, instability, and reduced generalizability—not processing slowdowns or automatic correction during training. In a fleet management context, failing to harmonize diverse data sources means the model cannot fully capture relevant patterns.
Post your Comments and Discuss PMI PMI-CPMAI exam prep with other Community members:
Our website is free, but we have to fight against AI bots and content theft. We're sorry for the inconvenience caused by these security measures. You can access the rest of the PMI-CPMAI content, but please register or login to continue.