2. Technical AI and Machine Learning for Grades 11-12

Course Positioning

A rigorous pre-university AI course for students who may pursue computer science, engineering, statistics, data science, design, medicine, business analytics, or scientific research.

Learning outcomes

Implement linear regression, logistic regression, gradient descent, and a simple neural network in Python.
Use vectors, matrices, loss functions, gradients, and optimization to explain how models learn.
Compare classical ML, deep learning, transformers, RAG, and fine-tuning at a high level.
Evaluate models using classification and regression metrics, validation sets, and failure slices.
Build a small ML or LLM-assisted application with documented data, evaluation, and deployment choices.
Read a simplified AI research paper and extract the problem, method, experiment, and limitation.

Course Design Snapshot

Positioning: A rigorous pre-university AI course for students who may pursue computer science, engineering, statistics, data science, design, medicine, business analytics, or scientific research.
Audience: Students in grades 11-12, ideally with basic Python exposure and school-level mathematics.
Duration: 12 weeks, 2 sessions per week, 90 minutes per session, with optional weekend project clinics.
Prerequisites: Basic Python, algebra, graphs, functions, probability intuition, and willingness to work through notebooks.
Format: Concept lecture, math intuition, code walkthrough, lab, error analysis, and project checkpoint.

Expanded Topic-by-Topic Coverage

Module 1. AI problem framing

Module focus: AI problem framing: prediction, classification, generation, control, ranking, retrieval, and decision support. Primary live activity or lab: Turn five real-world problems into formal ML tasks with inputs, labels, metrics, and risks.

Topics and coverage

prediction

What it means: define prediction clearly and connect it to the module focus: AI problem framing: prediction, classification, generation, control, ranking, retrieval, and decision support.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

classification

What it means: place classification inside the AI system stack so learners know what problem it solves and what tradeoffs it introduces.
What to cover: inputs, outputs, system boundaries, evaluation criteria, cost or latency implications, and common failure cases.
Demonstration: use a diagram, small code sample, worksheet, or tool trace to make the mechanism visible.
Evidence of learning: learners compare two approaches and explain which one they would choose for a realistic constraint.

generation

What it means: define generation clearly and connect it to the module focus: AI problem framing: prediction, classification, generation, control, ranking, retrieval, and decision support.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

control

What it means: define control clearly and connect it to the module focus: AI problem framing: prediction, classification, generation, control, ranking, retrieval, and decision support.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

ranking

What it means: define ranking clearly and connect it to the module focus: AI problem framing: prediction, classification, generation, control, ranking, retrieval, and decision support.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

retrieval

What it means: explain how retrieval changes the interaction between human intent, model behavior, external information, and final output.
What to cover: inputs, constraints, examples, output format, grounding, iteration, failure modes, and when a human must intervene.
Demonstration: show a weak attempt, a stronger structured attempt, and a reviewed final version with explicit checks.
Evidence of learning: learners create a reusable prompt, schema, retrieval note, or workflow pattern and test it on at least two examples.

decision support

What it means: show where decision support appears in the learner's real workflow and which parts are judgment-heavy versus draftable.
What to cover: current workflow, pain points, AI-assisted steps, human review checkpoints, quality standard, and ownership of the final decision.
Demonstration: convert one messy real-world input into a structured brief, draft, analysis, checklist, or next action.
Evidence of learning: learners produce a reusable template or playbook entry that can be used after the course.

Practice and evidence of learning

Learners complete or discuss: Turn five real-world problems into formal ML tasks with inputs, labels, metrics, and risks.
Learners produce: Turn five real-world problems into formal ML tasks with inputs, labels, metrics, and risks.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 2. Python for data

Module focus: Python for data: arrays, dataframes, visualization, preprocessing, missing data, leakage, and reproducibility. Primary live activity or lab: Clean a dataset, create train/validation/test splits, and write a short data report.

Topics and coverage

arrays

What it means: define arrays clearly and connect it to the module focus: Python for data: arrays, dataframes, visualization, preprocessing, missing data, leakage, and reproducibility.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

dataframes

What it means: connect dataframes to the data lifecycle from source and structure through analysis, interpretation, and decision-making.
What to cover: source reliability, missing or biased data, leakage, assumptions, calculations, and the difference between correlation and decision-ready evidence.
Demonstration: walk through a small dataset or example table and mark the checks required before trusting the result.
Evidence of learning: learners produce a short analysis note that includes assumptions, limitations, and verification steps.

visualization

What it means: define visualization clearly and connect it to the module focus: Python for data: arrays, dataframes, visualization, preprocessing, missing data, leakage, and reproducibility.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

preprocessing

What it means: show where preprocessing appears in the learner's real workflow and which parts are judgment-heavy versus draftable.
What to cover: current workflow, pain points, AI-assisted steps, human review checkpoints, quality standard, and ownership of the final decision.
Demonstration: convert one messy real-world input into a structured brief, draft, analysis, checklist, or next action.
Evidence of learning: learners produce a reusable template or playbook entry that can be used after the course.

missing data

What it means: connect missing data to the data lifecycle from source and structure through analysis, interpretation, and decision-making.
What to cover: source reliability, missing or biased data, leakage, assumptions, calculations, and the difference between correlation and decision-ready evidence.
Demonstration: walk through a small dataset or example table and mark the checks required before trusting the result.
Evidence of learning: learners produce a short analysis note that includes assumptions, limitations, and verification steps.

leakage

What it means: define leakage clearly and connect it to the module focus: Python for data: arrays, dataframes, visualization, preprocessing, missing data, leakage, and reproducibility.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

reproducibility

What it means: define reproducibility clearly and connect it to the module focus: Python for data: arrays, dataframes, visualization, preprocessing, missing data, leakage, and reproducibility.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

Practice and evidence of learning

Learners complete or discuss: Clean a dataset, create train/validation/test splits, and write a short data report.
Learners produce: Clean a dataset, create train/validation/test splits, and write a short data report.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 3. Regression

Module focus: Regression: linear functions, mean squared error, residuals, regularization intuition, and feature scaling. Primary live activity or lab: Fit linear regression and compare hand-computed predictions with library output.

Topics and coverage

linear functions

What it means: define linear functions clearly and connect it to the module focus: Regression: linear functions, mean squared error, residuals, regularization intuition, and feature scaling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

mean squared error

What it means: define mean squared error clearly and connect it to the module focus: Regression: linear functions, mean squared error, residuals, regularization intuition, and feature scaling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

residuals

What it means: define residuals clearly and connect it to the module focus: Regression: linear functions, mean squared error, residuals, regularization intuition, and feature scaling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

regularization intuition

What it means: define regularization intuition clearly and connect it to the module focus: Regression: linear functions, mean squared error, residuals, regularization intuition, and feature scaling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

feature scaling

What it means: connect feature scaling to the data lifecycle from source and structure through analysis, interpretation, and decision-making.
What to cover: source reliability, missing or biased data, leakage, assumptions, calculations, and the difference between correlation and decision-ready evidence.
Demonstration: walk through a small dataset or example table and mark the checks required before trusting the result.
Evidence of learning: learners produce a short analysis note that includes assumptions, limitations, and verification steps.

Practice and evidence of learning

Learners complete or discuss: Fit linear regression and compare hand-computed predictions with library output.
Learners produce: Fit linear regression and compare hand-computed predictions with library output.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 4. Classification

Module focus: Classification: logistic regression, probabilities, cross-entropy intuition, thresholds, class imbalance. Primary live activity or lab: Build a binary classifier and choose a threshold for two different business goals.

Topics and coverage

logistic regression

What it means: place logistic regression inside the AI system stack so learners know what problem it solves and what tradeoffs it introduces.
What to cover: inputs, outputs, system boundaries, evaluation criteria, cost or latency implications, and common failure cases.
Demonstration: use a diagram, small code sample, worksheet, or tool trace to make the mechanism visible.
Evidence of learning: learners compare two approaches and explain which one they would choose for a realistic constraint.

probabilities

What it means: define probabilities clearly and connect it to the module focus: Classification: logistic regression, probabilities, cross-entropy intuition, thresholds, class imbalance.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

cross-entropy intuition

What it means: define cross-entropy intuition clearly and connect it to the module focus: Classification: logistic regression, probabilities, cross-entropy intuition, thresholds, class imbalance.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

thresholds

What it means: show where thresholds appears in the learner's real workflow and which parts are judgment-heavy versus draftable.
What to cover: current workflow, pain points, AI-assisted steps, human review checkpoints, quality standard, and ownership of the final decision.
Demonstration: convert one messy real-world input into a structured brief, draft, analysis, checklist, or next action.
Evidence of learning: learners produce a reusable template or playbook entry that can be used after the course.

class imbalance

What it means: define class imbalance clearly and connect it to the module focus: Classification: logistic regression, probabilities, cross-entropy intuition, thresholds, class imbalance.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

Practice and evidence of learning

Learners complete or discuss: Build a binary classifier and choose a threshold for two different business goals.
Learners produce: Build a binary classifier and choose a threshold for two different business goals.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 5. Gradient descent

Module focus: Gradient descent: loss landscape intuition, learning rate, epochs, batches, and why optimization can fail. Primary live activity or lab: Code gradient descent for a one-variable model and visualize the update path.

Topics and coverage

loss landscape intuition

What it means: define loss landscape intuition clearly and connect it to the module focus: Gradient descent: loss landscape intuition, learning rate, epochs, batches, and why optimization can fail.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

learning rate

What it means: define learning rate clearly and connect it to the module focus: Gradient descent: loss landscape intuition, learning rate, epochs, batches, and why optimization can fail.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

epochs

What it means: define epochs clearly and connect it to the module focus: Gradient descent: loss landscape intuition, learning rate, epochs, batches, and why optimization can fail.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

batches

What it means: define batches clearly and connect it to the module focus: Gradient descent: loss landscape intuition, learning rate, epochs, batches, and why optimization can fail.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

why optimization can fail

What it means: place why optimization can fail inside the AI system stack so learners know what problem it solves and what tradeoffs it introduces.
What to cover: inputs, outputs, system boundaries, evaluation criteria, cost or latency implications, and common failure cases.
Demonstration: use a diagram, small code sample, worksheet, or tool trace to make the mechanism visible.
Evidence of learning: learners compare two approaches and explain which one they would choose for a realistic constraint.

Practice and evidence of learning

Learners complete or discuss: Code gradient descent for a one-variable model and visualize the update path.
Learners produce: Code gradient descent for a one-variable model and visualize the update path.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 6. Neural networks

Module focus: Neural networks: matrix multiplication, activations, hidden layers, backpropagation concept, and overfitting. Primary live activity or lab: Train an MLP on a small image or tabular dataset and tune architecture choices.

Topics and coverage

matrix multiplication

What it means: define matrix multiplication clearly and connect it to the module focus: Neural networks: matrix multiplication, activations, hidden layers, backpropagation concept, and overfitting.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

activations

What it means: define activations clearly and connect it to the module focus: Neural networks: matrix multiplication, activations, hidden layers, backpropagation concept, and overfitting.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

hidden layers

What it means: define hidden layers clearly and connect it to the module focus: Neural networks: matrix multiplication, activations, hidden layers, backpropagation concept, and overfitting.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

backpropagation concept

What it means: define backpropagation concept clearly and connect it to the module focus: Neural networks: matrix multiplication, activations, hidden layers, backpropagation concept, and overfitting.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

overfitting

What it means: define overfitting clearly and connect it to the module focus: Neural networks: matrix multiplication, activations, hidden layers, backpropagation concept, and overfitting.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

Practice and evidence of learning

Learners complete or discuss: Train an MLP on a small image or tabular dataset and tune architecture choices.
Learners produce: Train an MLP on a small image or tabular dataset and tune architecture choices.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 7. Computer vision and sequences

Module focus: Computer vision and sequences: CNN intuition, embeddings, recurrence limits, attention, and transformers. Primary live activity or lab: Implement a toy attention mechanism and inspect attention scores.

Topics and coverage

CNN intuition

What it means: define CNN intuition clearly and connect it to the module focus: Computer vision and sequences: CNN intuition, embeddings, recurrence limits, attention, and transformers.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

embeddings

What it means: explain how embeddings changes the interaction between human intent, model behavior, external information, and final output.
What to cover: inputs, constraints, examples, output format, grounding, iteration, failure modes, and when a human must intervene.
Demonstration: show a weak attempt, a stronger structured attempt, and a reviewed final version with explicit checks.
Evidence of learning: learners create a reusable prompt, schema, retrieval note, or workflow pattern and test it on at least two examples.

recurrence limits

What it means: define recurrence limits clearly and connect it to the module focus: Computer vision and sequences: CNN intuition, embeddings, recurrence limits, attention, and transformers.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

attention

What it means: define attention clearly and connect it to the module focus: Computer vision and sequences: CNN intuition, embeddings, recurrence limits, attention, and transformers.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

transformers

What it means: place transformers inside the AI system stack so learners know what problem it solves and what tradeoffs it introduces.
What to cover: inputs, outputs, system boundaries, evaluation criteria, cost or latency implications, and common failure cases.
Demonstration: use a diagram, small code sample, worksheet, or tool trace to make the mechanism visible.
Evidence of learning: learners compare two approaches and explain which one they would choose for a realistic constraint.

Practice and evidence of learning

Learners complete or discuss: Implement a toy attention mechanism and inspect attention scores.
Learners produce: Implement a toy attention mechanism and inspect attention scores.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 8. Generative AI

Module focus: Generative AI: tokenization, pretraining, instruction tuning, RLHF/RLAIF idea, hallucination, and sampling. Primary live activity or lab: Compare temperature, top-p, and prompt structure on a controlled task.

Topics and coverage

tokenization

What it means: define tokenization clearly and connect it to the module focus: Generative AI: tokenization, pretraining, instruction tuning, RLHF/RLAIF idea, hallucination, and sampling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

pretraining

What it means: place pretraining inside the AI system stack so learners know what problem it solves and what tradeoffs it introduces.
What to cover: inputs, outputs, system boundaries, evaluation criteria, cost or latency implications, and common failure cases.
Demonstration: use a diagram, small code sample, worksheet, or tool trace to make the mechanism visible.
Evidence of learning: learners compare two approaches and explain which one they would choose for a realistic constraint.

instruction tuning

What it means: define instruction tuning clearly and connect it to the module focus: Generative AI: tokenization, pretraining, instruction tuning, RLHF/RLAIF idea, hallucination, and sampling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

RLHF/RLAIF idea

What it means: define RLHF/RLAIF idea clearly and connect it to the module focus: Generative AI: tokenization, pretraining, instruction tuning, RLHF/RLAIF idea, hallucination, and sampling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

hallucination

What it means: define hallucination clearly and connect it to the module focus: Generative AI: tokenization, pretraining, instruction tuning, RLHF/RLAIF idea, hallucination, and sampling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

sampling

What it means: define sampling clearly and connect it to the module focus: Generative AI: tokenization, pretraining, instruction tuning, RLHF/RLAIF idea, hallucination, and sampling.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

Practice and evidence of learning

Learners complete or discuss: Compare temperature, top-p, and prompt structure on a controlled task.
Learners produce: Compare temperature, top-p, and prompt structure on a controlled task.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 9. Retrieval and grounding

Module focus: Retrieval and grounding: embeddings, vector search, chunking, RAG, citation, and evaluation. Primary live activity or lab: Build a small document QA system over a provided collection and test answer faithfulness.

Topics and coverage

embeddings

What it means: explain how embeddings changes the interaction between human intent, model behavior, external information, and final output.
What to cover: inputs, constraints, examples, output format, grounding, iteration, failure modes, and when a human must intervene.
Demonstration: show a weak attempt, a stronger structured attempt, and a reviewed final version with explicit checks.
Evidence of learning: learners create a reusable prompt, schema, retrieval note, or workflow pattern and test it on at least two examples.

vector search

What it means: define vector search clearly and connect it to the module focus: Retrieval and grounding: embeddings, vector search, chunking, RAG, citation, and evaluation.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

chunking

What it means: define chunking clearly and connect it to the module focus: Retrieval and grounding: embeddings, vector search, chunking, RAG, citation, and evaluation.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

RAG

What it means: explain how RAG changes the interaction between human intent, model behavior, external information, and final output.
What to cover: inputs, constraints, examples, output format, grounding, iteration, failure modes, and when a human must intervene.
Demonstration: show a weak attempt, a stronger structured attempt, and a reviewed final version with explicit checks.
Evidence of learning: learners create a reusable prompt, schema, retrieval note, or workflow pattern and test it on at least two examples.

citation

What it means: define citation clearly and connect it to the module focus: Retrieval and grounding: embeddings, vector search, chunking, RAG, citation, and evaluation.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

evaluation

What it means: connect evaluation to the data lifecycle from source and structure through analysis, interpretation, and decision-making.
What to cover: source reliability, missing or biased data, leakage, assumptions, calculations, and the difference between correlation and decision-ready evidence.
Demonstration: walk through a small dataset or example table and mark the checks required before trusting the result.
Evidence of learning: learners produce a short analysis note that includes assumptions, limitations, and verification steps.

Practice and evidence of learning

Learners complete or discuss: Build a small document QA system over a provided collection and test answer faithfulness.
Learners produce: Build a small document QA system over a provided collection and test answer faithfulness.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 10. Evaluation and robustness

Module focus: Evaluation and robustness: confusion matrix, calibration, fairness slices, adversarial examples, and model cards. Primary live activity or lab: Write an error analysis report with at least three failure categories.

Topics and coverage

confusion matrix

What it means: define confusion matrix clearly and connect it to the module focus: Evaluation and robustness: confusion matrix, calibration, fairness slices, adversarial examples, and model cards.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

calibration

What it means: define calibration clearly and connect it to the module focus: Evaluation and robustness: confusion matrix, calibration, fairness slices, adversarial examples, and model cards.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

fairness slices

What it means: define fairness slices clearly and connect it to the module focus: Evaluation and robustness: confusion matrix, calibration, fairness slices, adversarial examples, and model cards.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

adversarial examples

What it means: define adversarial examples clearly and connect it to the module focus: Evaluation and robustness: confusion matrix, calibration, fairness slices, adversarial examples, and model cards.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

model cards

What it means: place model cards inside the AI system stack so learners know what problem it solves and what tradeoffs it introduces.
What to cover: inputs, outputs, system boundaries, evaluation criteria, cost or latency implications, and common failure cases.
Demonstration: use a diagram, small code sample, worksheet, or tool trace to make the mechanism visible.
Evidence of learning: learners compare two approaches and explain which one they would choose for a realistic constraint.

Practice and evidence of learning

Learners complete or discuss: Write an error analysis report with at least three failure categories.
Learners produce: Write an error analysis report with at least three failure categories.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 11. Deployment basics

Module focus: Deployment basics: APIs, latency, cost, privacy, monitoring, and human-in-the-loop workflows. Primary live activity or lab: Create a simple web or notebook interface around a model.

Topics and coverage

APIs

What it means: define APIs clearly and connect it to the module focus: Deployment basics: APIs, latency, cost, privacy, monitoring, and human-in-the-loop workflows.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

latency

What it means: place latency inside the AI system stack so learners know what problem it solves and what tradeoffs it introduces.
What to cover: inputs, outputs, system boundaries, evaluation criteria, cost or latency implications, and common failure cases.
Demonstration: use a diagram, small code sample, worksheet, or tool trace to make the mechanism visible.
Evidence of learning: learners compare two approaches and explain which one they would choose for a realistic constraint.

cost

What it means: place cost inside the AI system stack so learners know what problem it solves and what tradeoffs it introduces.
What to cover: inputs, outputs, system boundaries, evaluation criteria, cost or latency implications, and common failure cases.
Demonstration: use a diagram, small code sample, worksheet, or tool trace to make the mechanism visible.
Evidence of learning: learners compare two approaches and explain which one they would choose for a realistic constraint.

privacy

What it means in this course: define privacy in operational terms, not as an abstract principle.
What to cover: sensitive data boundaries, affected stakeholders, approval paths, documentation, and what Students in grades 11-12, ideally with basic Python exposure and school-level mathematics must never delegate blindly to AI.
Use case: present one acceptable use, one borderline use, and one prohibited use, then ask learners to justify the classification.
Evidence of learning: learners add a risk control, review step, or escalation rule to their course project.

monitoring

What it means: define monitoring clearly and connect it to the module focus: Deployment basics: APIs, latency, cost, privacy, monitoring, and human-in-the-loop workflows.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

human-in-the-loop workflows

What it means: show where human-in-the-loop workflows appears in the learner's real workflow and which parts are judgment-heavy versus draftable.
What to cover: current workflow, pain points, AI-assisted steps, human review checkpoints, quality standard, and ownership of the final decision.
Demonstration: convert one messy real-world input into a structured brief, draft, analysis, checklist, or next action.
Evidence of learning: learners produce a reusable template or playbook entry that can be used after the course.

Practice and evidence of learning

Learners complete or discuss: Create a simple web or notebook interface around a model.
Learners produce: Create a simple web or notebook interface around a model.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Module 12. Research and capstone studio

Module focus: Research and capstone studio: literature reading, experimental design, ablations, and final presentation. Primary live activity or lab: Present a capstone with data, method, results, limitations, and next steps.

Topics and coverage

literature reading

What it means: define literature reading clearly and connect it to the module focus: Research and capstone studio: literature reading, experimental design, ablations, and final presentation.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

experimental design

What it means: show where experimental design appears in the learner's real workflow and which parts are judgment-heavy versus draftable.
What to cover: current workflow, pain points, AI-assisted steps, human review checkpoints, quality standard, and ownership of the final decision.
Demonstration: convert one messy real-world input into a structured brief, draft, analysis, checklist, or next action.
Evidence of learning: learners produce a reusable template or playbook entry that can be used after the course.

ablations

What it means: define ablations clearly and connect it to the module focus: Research and capstone studio: literature reading, experimental design, ablations, and final presentation.
What to cover: the core concept, why it matters, what good usage looks like, and where learners are likely to misunderstand it.
Demonstration: give one simple example, one realistic example, and one failure or limitation example.
Evidence of learning: learners explain the topic in their own words and apply it to a small artifact or decision.

final presentation

What it means: show where final presentation appears in the learner's real workflow and which parts are judgment-heavy versus draftable.
What to cover: current workflow, pain points, AI-assisted steps, human review checkpoints, quality standard, and ownership of the final decision.
Demonstration: convert one messy real-world input into a structured brief, draft, analysis, checklist, or next action.
Evidence of learning: learners produce a reusable template or playbook entry that can be used after the course.

Practice and evidence of learning

Learners complete or discuss: Present a capstone with data, method, results, limitations, and next steps.
Learners produce: Present a capstone with data, method, results, limitations, and next steps.
Instructor checks for accuracy, practical usefulness, clear assumptions, appropriate human review, and fit with the course audience.
Learners revise once after feedback so the module contributes to the final project, portfolio, or playbook.

Minimum coverage before moving on

Learners can explain the module vocabulary without relying on tool-generated text.
Learners have seen one worked example, one hands-on application, and one limitation or failure case.
Learners know what must be verified, what data must be protected, and who remains accountable for the output.

Core labs and builds

From-scratch gradient descent for a small regression problem.
Neural network classification lab using a small image or tabular dataset.
Mini-RAG system with embeddings, retrieval evaluation, and answer verification.
Paper reading lab using one accessible paper on transformers, diffusion, or reinforcement learning.

Capstone

Students choose one track: predictive ML project, generative AI assistant with retrieval, computer vision prototype, or AI for science mini-investigation. The final artifact includes code, notebook, model evaluation, model card, and presentation.

Assessment design

Technical notebooks with clear code and interpretation.
Short math and concept checks after each unit.
Mid-course practical exam: implement and evaluate a supervised ML model.
Final capstone scored on framing, implementation, evaluation, communication, and safety.

Recommended tools and datasets

Python, Google Colab, NumPy, pandas, scikit-learn, PyTorch or TensorFlow, Hugging Face demos, FAISS or Chroma for vector search, GitHub Classroom if available.

Instructor notes

The key upgrade from a usage course is that students should implement parts of the learning process, not only call AI tools. The emphasis should be conceptual depth plus visible working code.

Instructor Build Checklist

Prepare one short demo for each module and one learner activity that creates a saved artifact.
Prepare examples that match the audience, local context, and likely tools learners can access.
Add a verification step to every AI-generated output: factual check, source check, data sensitivity check, and quality review.
Keep a running portfolio folder so each module contributes to the final project or learner playbook.
Reserve time for reflection on what the learner did, what AI did, what was checked, and what remains uncertain.