Artificial Intelligence and Machine Learning Foundations
Learning from data
What machine learning is
Machine learning builds models that learn patterns from data rather than relying only on hand-written rules. A model is trained on examples, evaluated on unseen data, and then used to make predictions or decisions.
Rule-based software vs. machine learning
Rule-based software
- Explicit logic written by developers
- Best when rules are clear and stable
Machine learning
- Patterns learned from data
- Best when rules are complex or hard to specify
ML vocabulary
What is overfitting?
Overfitting means poor generalization to unseen data.
Correct answer: When a model performs well on training data but poorly on new data
Why is a separate test set important?
Evaluation should not reuse the same data used for training.
Correct answer: It estimates how well a model generalizes to unseen data.
Applications and limitations
ML powers search ranking, recommendation systems, speech recognition, vision, fraud detection, and more. However, models can encode bias, make confident mistakes, and require careful monitoring, especially when decisions affect people.
Which is a common machine learning task?
Spam detection is a classic supervised learning problem.
Correct answer: Predicting whether an email is spam
Name one risk of using machine learning in high-stakes settings.
High-stakes applications need careful evaluation and oversight.
Correct answer: Bias, poor generalization, or lack of interpretability.
Data quality matters
A model is only as good as the data and objective it learns from. Bad data, weak labels, or a mismatched metric can produce misleading results.