Models learn patterns from data to classify, estimate, detect or recommend without every rule being hand-coded.
CCNA focus: Treat AI output as evidence or a proposal, not authority. Protect sensitive data, understand confidence and bias, verify against live state and keep humans accountable for impactful changes.
02 // Building blocks
Know what each part does.
01
Machine learning
Models learn patterns from data to classify, estimate, detect or recommend without every rule being hand-coded.
02
Predictive AI
Uses historical and live signals to estimate likely future events such as capacity pressure or failure risk.
03
Generative AI
Produces new text, explanations, queries, code or configuration suggestions from prompts and context.
04
Operational use
AI can summarise incidents, correlate signals, detect anomalies and assist troubleshooting, but output must be verified.
03 // Compare and recognise
Read the clues.
Item
What to remember
Training data
Examples used to fit model parameters; quality and representativeness matter.
Inference
Using a trained model to produce a prediction or generated result.
False positive
A benign event incorrectly flagged as a problem.
Hallucination
Plausible-looking generated output that is unsupported or incorrect.
04 // AI-assisted workflow
AI-assisted workflow.
Telemetry and logs -> feature/context preparation -> model detects unusual behaviour -> system ranks likely causes -> generative assistant summarises evidence -> engineer validates against live state -> approved remediation runs with guardrails -> outcome feeds monitoring and review
Read the example from top to bottom, then verify the resulting state. Configuration is only complete when the output matches the intended design.
05 // Exam and troubleshooting
Turn facts into a method.
Examples used to fit model parameters; quality and representativeness matter.
Using a trained model to produce a prediction or generated result.
A benign event incorrectly flagged as a problem.
Plausible-looking generated output that is unsupported or incorrect.
Exam checkpoint: Treat AI output as evidence or a proposal, not authority. Protect sensitive data, understand confidence and bias, verify against live state and keep humans accountable for impactful changes.
06 // Check yourself
AI & Machine Learning quiz.
1. Which AI category creates new text or configuration suggestions?
2. What does predictive AI commonly estimate?
3. What is inference?
4. What is an AI hallucination?
5. Who remains accountable for an impactful network change suggested by AI?