AB-731 Course Notes

Generative AI creates or transforms content from prompts and context, while predictive AI usually scores, forecasts, classifies, or detects patterns from structured data. AB-731 questions often ask leaders to choose the right AI approach for a business need: generative AI fits drafting, summarizing, search, ideation, and natural-language assistance; predictive AI fits repeatable decisions such as demand forecasts, churn scoring, anomaly detection, and risk prediction. The business value test is not whether AI sounds impressive, but whether the capability improves a measurable outcome.

• Treating generative AI as the right answer for every AI scenario instead of separating content generation from predictive scoring, forecasting, or classification.

• If the scenario asks for new text, summaries, drafts, or natural-language answers, think generative AI before predictive AI.

Microsoft 365 Copilot and Microsoft Copilot solve different levels of work. Microsoft 365 Copilot is grounded in Microsoft 365 work data and appears across apps such as Word, Excel, PowerPoint, Outlook, Teams, and chat experiences; Microsoft Copilot and Microsoft 365 Copilot Chat provide broader chat assistance with web grounding and enterprise controls depending on the experience. AB-731 questions usually ask leaders to map a business process to the right Copilot surface, not to assume every Copilot has the same data access or app integration.

• Confusing Microsoft 365 Copilot, Microsoft Copilot, Copilot Studio, and Azure AI Foundry as interchangeable products.

• Use Microsoft 365 Copilot when the task lives in Microsoft 365 apps and Microsoft Graph context matters.

An AI implementation strategy starts with responsible AI principles and ownership. Fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability should be translated into decision gates, not left as slogans. An AI council or steering committee gives legal, security, IT, risk, data, business, and operations leaders a place to approve use cases, resolve conflicts, and stop teams from creating shadow AI with unmanaged data exposure.

• Rolling out licenses before establishing an AI council, executive sponsor, data rules, success metrics, and support ownership.

• When the question describes organization-wide adoption, look for steering committee, sponsor, champions, and role-based training answers.