ISACA Artificial Intelligence Fundamentals Certificate

Learning from examples is the core idea. You see how supervised, unsupervised and reinforcement learning differ, what training and testing mean, and why a model can memorize rather than learn.

• Supervised learning with labelled examples
• Unsupervised learning and finding groups
• Reinforcement learning through reward
• Training, validation and testing
• Overfitting and underfitting
• A simple model trained step by step

You build A small model trained on a public dataset, with a note on how well it learned.

A model reflects its data. You look at kinds of data, how data is collected and labelled, what makes it good or poor, and how bias and privacy concerns enter at this stage.

• Structured and unstructured data
• Collecting and labelling data
• Data quality and cleaning
• Bias in data
• Privacy and consent
• Big data and the cost of scale

You build A review of one dataset, noting its source, quality problems and possible bias.

Neural networks power much of modern AI. You see how neurons, layers and training fit together, and how different network types handle images, sequences and language.

• Neurons, layers and weights
• How a network learns from error
• Networks for images
• Networks for sequences and language
• Why deep learning needs data and computing power
• Limits of neural networks

You build A labelled drawing of a small neural network, with a trace of one example through it.

Generative systems create text, images and code. You learn how large language models predict text, how prompts steer them, why they make things up, and where they fit alongside other tools.

• How language models predict text
• Prompts and how to write them
• Image and other media generation
• Hallucination and unreliable answers
• Assistants and copilots in applications
• Choosing generative AI or another tool

You build A set of prompts for one task, with notes on which answers were right, wrong or unclear.

Techniques fit problems. You survey the main uses of AI, from vision and language to prediction, recommendation and robotics, and learn to match a problem to a suitable approach.

• Computer vision uses
• Language and speech uses
• Prediction and recommendation
• Robotics and automation
• AI in health, finance and public services
• Matching a problem to a technique

You build A table that matches six business problems to AI techniques, with reasons.

AI applications are assembled from standard parts. You meet the common algorithm families, the programming languages and libraries used to build them, and the services that provide ready-made AI.

• Regression, decision trees and clustering
• Nearest neighbors and simple classifiers
• Python and the libraries around it
• Frameworks for neural networks
• Notebooks and development tools
• Cloud AI services and APIs

You build A notebook that runs three algorithms on one dataset and compares their results.

Building AI is a process, not a single step. You follow a project from problem definition through data preparation, training, evaluation, deployment and monitoring, and see where people stay involved.

• Defining the problem and the measure of success
• Preparing data
• Training and evaluating a model
• Deploying and integrating the model
• Monitoring and updating after launch
• Where people review and decide

You build A project outline for one AI application, from problem statement to monitoring plan.

Impressive demonstrations can hide weak results. You learn accuracy and the other common measures in plain terms, read a confusion matrix, and see why the cost of different errors matters.

• Accuracy and why it can mislead
• False positives and false negatives
• Precision and recall in plain terms
• Reading a confusion matrix
• Testing on new and varied data
• Comparing two models fairly

You build An evaluation of a simple classifier, with a confusion matrix and a plain-language verdict.

AI creates risk as well as value. You look at bias, privacy, security, misinformation and accountability, learn the ethical principles organizations use, and see how rules and regulation are developing.

• Bias and unfair outcomes
• Privacy and security risks
• Misinformation and misuse
• Transparency and explainability
• Accountability and human oversight
• Ethical principles and emerging regulation

You build A risk review of one AI product, listing harms, who is affected and how they could be reduced.

Most people will use AI before they build it. You learn how to check AI output, handle information safely, respect intellectual property and adapt your own work as AI tools arrive.

• Checking and correcting AI output
• What not to put into an AI tool
• Intellectual property and attribution
• Using AI within an organization’s rules
• Skills that matter alongside AI
• Staying informed as tools change

You build A personal guide to using AI tools safely and well in your own work or study.

One problem draws the concepts and the implementation together. You choose a real problem, decide whether AI suits it, outline the data, approach and risks, and present the plan.

• Choosing a problem and testing whether AI fits
• Identifying the data and approach
• Planning evaluation and oversight
• Listing risks and safeguards
• Presenting the plan

You build A short proposal for one AI application, covering problem, data, method, evaluation and risks.