Skip to content
MMISSION AIACADEMYLearner sign in
PUBLIC SAMPLE · FOUNDATIONS

From rules to learning systems

Distinguish rules, machine learning, and generative AI in a workplace task.

An actual lesson from the program. Approximately 15 minutes. No account required. This sample does not save progress or issue a certificate.
01 / LESSON

Three different ways to solve a problem

A rules-based system follows instructions written by people: if a training record is expired, flag it. Machine learning learns statistical patterns from examples: predict which service requests may take longer using past outcomes. Generative AI creates a new output, such as a draft summary. It may use machine learning, but generating text is not the same as finding a verified fact. AI is the broad category; machine learning is one approach within it; deep learning uses neural networks with multiple layers.

02 / LESSON

How learning works

Supervised learning uses examples with known labels, such as a request and its correct category. Unsupervised learning looks for patterns without supplied labels, such as clusters of similar requests. Reinforcement learning changes a policy based on rewards from interaction. Training fits model parameters; inference uses the trained model on a new input. Evaluation must use examples that were not used to train the system, or the apparent performance may be misleading.

03 / LESSON

A leadership decision

A G-6 team wants to route routine help requests. A fixed rule can route a known form field. A classifier may help with varied free text. A language model may draft a response. Start with the lowest-complexity approach that meets the need. A model adds uncertainty, monitoring, data requirements, and maintenance; it is not automatically an improvement.

Apply the distinction.

Classify these tasks: (1) flag a date older than 365 days; (2) learn request categories from labeled examples; (3) draft an orientation message. For each, name the approach and the person who verifies the result.

Practice only. These selections and notes stay in this page’s memory and disappear when you reload. They are not submitted to the training database. Use fictional information.

1. Flag a date older than 365 days.
2. Learn request categories from labeled examples.
3. Draft an orientation message.

Key takeaway

Choose the method to fit the task. A deterministic rule is often the strongest baseline.

Continue with your team.

The full program adds saved exercises, five module assessments, completion certificates, and administrator reporting. Written activities record participation; scored assessments check comprehension.