Curriculum Overview: AI Workloads and Responsible AI (AI-900 Unit 1)
Unit 1: Describe Artificial Intelligence workloads and considerations (15–20%)
Curriculum Overview: AI Workloads and Responsible AI
This document provides a comprehensive overview of Unit 1 of the Microsoft Azure AI Fundamentals (AI-900) certification. This unit accounts for 15–20% of the total exam weight and focuses on the fundamental categories of Artificial Intelligence and the ethical framework required to deploy them.
Prerequisites
One of the unique aspects of the AI-900 curriculum is its accessibility. The following are the requirements for beginning this unit:
- Technical Knowledge: No advanced programming or mathematics skills are required. The course is designed for both technical and non-technical backgrounds.
- Cloud Awareness: A basic understanding of cloud computing concepts (Infrastructure as a Service, Platform as a Service, Software as a Service) is beneficial but not strictly mandatory.
- Motivation: An interest in how AI impacts modern business processes and the ethical implications of automated decision-making.
[!TIP] If you are pivoting into tech from a different field, this unit is the perfect "entry point" as it focuses more on conceptual understanding than technical implementation.
Module Breakdown
This unit is divided into two primary domains. The first explores the "what" (the types of AI workloads), and the second explores the "how" (the ethical principles of Responsible AI).
| Module ID | Module Title | Focus Area | Difficulty |
|---|---|---|---|
| 1.1 | Common AI Workloads | Computer Vision, NLP, Document Intelligence, Generative AI | Moderate |
| 1.2 | Responsible AI | Ethical principles: Fairness, Reliability, Transparency, etc. | Foundational |
Visual Breakdown of AI Workloads
Learning Objectives per Module
By the end of this curriculum, learners must be able to demonstrate proficiency in the following specific outcomes:
Module 1.1: Identify features of common AI workloads
- Computer Vision: Distinguish between image classification, object detection, and facial analysis.
- Natural Language Processing (NLP): Identify scenarios for sentiment analysis, key phrase extraction, and speech-to-text.
- Document Processing: Recognize how AI can extract data from structured and unstructured forms (Invoices, Receipts).
- Generative AI: Describe the features of models that create new content like text, code, or images.
Module 1.2: Identify guiding principles for Responsible AI
- Fairness: Ensure AI systems treat all people equally without bias.
- Reliability & Safety: Ensure systems perform as intended and do not cause harm during failures.
- Privacy & Security: Protect personal data and remain resilient to attacks.
- Inclusiveness: Empower everyone and engage people of all abilities.
- Transparency: Ensure users understand how the AI arrives at its conclusions.
- Accountability: Humans must remain responsible for the outcomes of AI systems.
Success Metrics
To ensure mastery of Unit 1, learners should evaluate themselves against these criteria:
- Workload Recognition: Can you look at a business problem (e.g., "We need to scan 1,000 invoices per hour") and identify which AI workload fits (Document Processing)?
- Principle Mapping: Can you explain which Responsible AI principle is violated if a facial recognition system only works for one demographic (Fairness/Inclusiveness)?
- Exam Readiness: Achieve a consistent score of 80% or higher on practice questions related to the 15-20% weighting of this unit.
[!IMPORTANT] The exam often tests your ability to match a specific scenario to a specific principle. For example: "Providing a datasheet that explains how a model was trained" → Transparency.
Real-World Application
This curriculum is not just for passing an exam; it maps directly to modern industry needs:
- Customer Support: Using NLP for sentiment analysis to prioritize angry customers in a support queue.
- Retail: Using Computer Vision for automated checkout or inventory tracking via cameras.
- Finance: Using Document Intelligence to automate the processing of mortgage applications or tax forms.
- Corporate Governance: Implementing the 6 Responsible AI principles to ensure the company avoids legal and reputational risks associated with "Black Box" AI.
The 6 Pillars of Responsible AI
\begin{tikzpicture} \node[draw, circle, minimum size=2cm, fill=blue!10] (core) at (0,0) {\textbf{Responsible AI}}; \node[draw, rectangle, fill=green!10] (f) at (90:3cm) {Fairness}; \node[draw, rectangle, fill=green!10] (r) at (150:3cm) {Reliability}; \node[draw, rectangle, fill=green!10] (p) at (210:3cm) {Privacy}; \node[draw, rectangle, fill=green!10] (i) at (270:3cm) {Inclusiveness}; \node[draw, rectangle, fill=green!10] (t) at (330:3cm) {Transparency}; \node[draw, rectangle, fill=green!10] (a) at (30:3cm) {Accountability};
\draw[->, thick] (core) -- (f);
\draw[->, thick] (core) -- (r);
\draw[->, thick] (core) -- (p);
\draw[->, thick] (core) -- (i);
\draw[->, thick] (core) -- (t);
\draw[->, thick] (core) -- (a);\end{tikzpicture}