Ntọala AI

• Explain training and inference in plain language.
• Separate demonstrated capability from marketing claims.
• Evaluate AI outputs using evidence, uncertainty, and fit-for-purpose tests.

Prerequisites:None.

1. 1

   What AI is

   Nsonaazụ: Distinguish AI, machine learning, and ordinary software.

   Omume omume: Classify five everyday systems and explain which definition of AI you used.

   AI concepts and boundaries
   Kedu ihe bụ AI? →Ntọala mmụta igwe →
2. 2

   How systems learn

   Nsonaazụ: Describe the role of examples, objectives, and generalization.

   Omume omume: Diagnose why a model can perform well in testing but fail for a new population.

   Data, training, and generalization
   Kedu ka AI si amụta →Ntọala mmụta igwe →Ọzụzụ AI →
3. 3

   Models and inference

   Nsonaazụ: Explain what happens when a trained model produces an output.

   Omume omume: Trace one user request from input through model output and human review.

   Ụdị, inference, na njedebe
   Ụdị AI kọwara →Ntinye aka AI →Netwọk akwara ozi →
4. 4

   Evidence and uncertainty

   Nsonaazụ: Check claims, benchmarks, and confident-sounding outputs.

   Omume omume: Create a claim ledger for one AI product announcement.

   Nyocha na benchmark agụmakwụkwọUncertainty and confidence calibrationIsi mmalite na nkwupụta nkwenye
   AI Benchmarks →AI hallucinations →Akụkọ ifo AI →