CPMAI_v7 Questions and Answers

The CPMAI Glossary defines Augmented Intelligence as “enhancing human abilities with AI,” where AI outputs are leveraged by humans to improve decision-making or productivity. Using Generative AI to accelerate or improve your own work is precisely an Augmented Intelligence use case, distinct from full autonomy or simple automation .

=========

Large language models (LLMs) excel at generating, understanding, and manipulating text. According to the CPMAI Glossary:

Content summarization is a core NLP function: “the process of using AI/ML techniques to generate a concise overview of a larger body of text.”

Machine translation: “the use of AI to automatically translate text or speech from one language to another.”

Classification: LLMs can assign content to categories via fine-tuned classification heads (“classifier” term), making them suitable for content categorization.

Code generation: As generative AI, LLMs can produce new content, including code snippets, by pattern learning from programming corpora (“generative AI” term).

Search quality improvement: LLMs can rephrase queries, expand keywords, and rank results to enhance search relevance. Though not explicitly detailed in the glossary, this capability derives directly from their generative and understanding strengths.

LLMs are not designed for pure process automation (option A), which is handled by RPA or orchestrators rather than by text-centric models.

=========

In Phase II: Data Understanding, CPMAI’s Generic Task Groups focus on:

Collecting initial data (identifying sources and volumes) and describing data (location/source) .

Verifying data quality to ensure completeness and correctness .

Team composition (the makeup of your data team) is addressed earlier under Phase I: Assess Situation, not during the Data Understanding phase where data requirements (quality, volume, source) are determined.

=========

CPMAI’s glossary defines data splitting as “dividing a data set into subsets (e.g., training, validation, test) for model development and evaluation,” typically achieved via random subsampling to ensure each subset is representative of the underlying distribution and to prevent sampling bias.

In Phase II: Data Understanding, CPMAI’s Data Selection tasks require teams to “Decide on the data to be used for analysis” by first listing all available sources and then selecting only those records and attributes that meet quality and relevance criteria . Taking a company-wide data inventory ensures you don’t overlook relevant datasets before narrowing down to what truly applies.

=========

A false positive occurs when the model predicts the positive class (e.g., “tree”) but the actual label is negative (e.g., “flower”). The confusion matrix definition confirms that mislabeling a negative instance as positive maps to the false positive count.

=========

In Phase IV’s Usage of AutoML task, CPMAI expressly recommends leveraging automated machine-learning tools to accelerate model creation when specialized expertise or time is limited. Documenting how AutoML will generate, evaluate, and export models allows teams to maintain pace without sacrificing rigor.

Phase III: Data Preparation focuses on constructing and documenting reusable data pipelines—including training and inference pipelines—alongside cleansing, augmentation, and labeling tasks to prepare data for modeling . This is where teams build the end-to-end data preparation workflows for AI solutions such as chatbots.

During Phase I: Business Understanding, the Data Feasibility task explicitly mandates a Go/No-Go decision on data availability and access: “Do you have access to the data you need? If not, what do you need for access to the data? Mark as a ‘NoGo.’” Projects should not proceed until all essential data access requirements are met to avoid wasted effort and unresolvable blockages down the line

CPMAI’s Usage of AutoML task instructs teams to “Document how AutoML tools will be used for model creation” and to verify that the output can be integrated into the overall I/O flow . While AutoML excels at automating algorithm selection, model selection, hyperparameter tuning, and even preliminary performance metrics, CPMAI places iterative modeling and evaluation squarely under the manual Model Evaluation phase—where teams must interpret results against business success criteria and decide on next steps. Entrusting that high-level, iterative decision-making to an AutoML black box would undermine the human-centric evaluation that CPMAI mandates.

=========

Naive Bayes classifiers are a family of “simple probabilistic classifiers based on Bayes’ theorem with the ‘naive’ assumption of feature independence,” making them fast to train and easy to interpret—ideal for teams new to AI tackling binary tasks like spam detection .

=========

In CPMAI’s Executing the Business Understanding Phase, teams first “formulate AI-specific business questions” and “estimate time-to-ROI for various AI project types” before matching business needs to cognitive patterns . This ensures ROI-driven objectives guide the selection of one or more of the Seven Patterns of AI, rather than the reverse.

=========

Trustworthy AI is not confined to a single phase but is woven throughout the entire CPMAI lifecycle:

The CPMAI Exam Content Outline under Domain VI: Trustworthy AI specifies tasks such as “Apply ethical AI concepts throughout the development lifecycle,” “Ensure compliance with privacy/security requirements,” and “Implement transparency and explainability” at every stage .

The CPMAI Workbook’s Task Group: Trustworthy AI Requirements (covering transparency, explainability, ethics, compliance, and responsible-AI frameworks) appears as an overarching set of artifacts and considerations that map back to multiple phases—beginning with Business Understanding and continuing through Model Operationalization .

Thus, Trustworthy AI considerations apply across all CPMAI phases.

CPMAI defines analytics as “the use of statistical and computational methods to extract meaningful insights from data.” When the goal is to discover and highlight patterns and insights within existing datasets, applying data mining or analytics techniques is the appropriate action.

The CPMAI Glossary defines a naive Bayes classifier as “a family of simple probabilistic classifiers based on Bayes’ theorem with the assumption of feature independence,” making it ideal for small or limited datasets where model simplicity and interpretability are priorities.

=========

Algorithm selection/code must be documented under the Select Modeling Technique task, where teams “document the actual algorithm/modeling technique to be used” .

Supporting training data pipelines are a core artifact of Phase III: Data Cleansing, which mandates “create a reusable data pipeline to collect, ingest, and prepare data for training purposes” .

Hyperparameter settings are captured in the Hyperparameter Optimization task, where teams “list the final, optimized settings” used for model building .

The bias-variance tradeoff is a conceptual consideration during evaluation but is not a discrete artifact to include in the project deliverables.

=========

CPMAI’s Domain IV: Data for AI – Task 1: Managing Data Fundamentals and Big Data Concepts emphasizes that leaders—not just technical practitioners—must grasp the core characteristics of Big Data (the V’s: volume, velocity, variety, veracity) and its strategic role in delivering business advantage. Ensuring senior leadership is data literate and understands how to leverage Big Data concepts across teams is critical for sustaining a competitive edge; merely upskilling the technical team or distributing data literacy unevenly will leave strategic gaps.

=========

Model iteration requires regularly updating a deployed model with new data and configuration. The CPMAI Workbook’s Task: Fine-Tuning / Re-training of Pre-Trained Models prescribes defining and documenting re-training pipelines as part of the model-building lifecycle to ensure seamless iteration and ongoing performance improvements.

Phase III: Data Preparation includes the Data Labeling generic task group. Specifically, the Label data task covers “identifying methods for data labeling and engaging in data labeling efforts,” which is essential for supervised learning workflows like image recognition.

=========

CPMAI defines real-time prediction as “the generation of predictions instantly as new data is received, which is crucial for time-sensitive applications” and positions it as the go-to approach when models must deliver immediate outputs for continuous monitoring scenarios . While streaming and microservices are architectural styles or ongoing learning frameworks, real-time prediction speaks directly to the requirement of instant inference.

=========

CPMAI’s Domain VI: Trustworthy AI is designed to span the entire project lifecycle, embedding ethics, transparency, fairness, privacy, and security checks into every phase—from Business Understanding through Operationalization—so that trustworthy practices are not an afterthought but a continuous, integrated activity .

=========

The CPMAI™ v7 methodology groups all data-centric preparation activities—including both data cleansing (“Clean data”) and data augmentation (“Enhance & Augment data”)—into Phase III: Data Preparation. In this phase, teams focus squarely on constructing the dataset to be used for modeling by performing all required cleaning, transformation, and enhancement operations.

Phase III: Data Preparation is defined in the Workbook’s Table of Contents as covering Data Cleansing & Enhancement tasks (“Clean data” and “Enhance & Augment data”) .

Under Phase III, the Generic Task Group: Data Cleansing & Enhancement explicitly lists “Task: Clean data” (bringing data quality to modeling-ready levels) and “Task: Enhance & Augment data” (producing derived attributes and new records) as core activities .

=========

Phase II: Data Understanding centers on identifying just the right amount of data for model training—neither too little (risking underfitting) nor too much (wasting resources and introducing noise). This balanced—“Goldilocks”—approach ensures you collect sufficient high-quality, relevant records to meet cognitive objectives without incurring unnecessary cost or complexity.

=========

Under Domain VI: Trustworthy AI in the CPMAI Exam Content Outline, Responsible AI covers establishing policies, guidelines, and governance that ensure AI solutions are developed for positive, ethical use and prevent misuse. Defining proper-use guidelines and embedding ethical intent into facial recognition directly align with Responsible AI practices .

Tom Mitchell’s widely adopted formulation captures ML’s essence: improvement on task T, measured by P, through experience E. This aligns with CPMAI’s view that ML enables systems to learn from data and improve over time (“The ability of a machine to learn from data, improve with experience, and apply that learning to make predictions.”) .

=========