Check 1
Which statement best explains artificial intelligence (AI)?
AI uses patterns from data to make predictions, recommendations, or new content.
Lesson 1: Humans Behind AI: How People Shape What AI Does
Step 1.11 of 6
Step 1.1 · Start here
AI systems do not choose their own goals. People decide what problem the system should help solve, what information it may use, how it will be tested, and where it may be used.
In this lesson, you will look at the people behind AI. You will compare four common model types and study the choices people make when they build and use AI.
💡 Big Question and Core Rule
Big Question: How do people's choices shape what AI can do, what it produces, and how the result is used?
Main Rule: AI can sort, predict, suggest, or create. People still set the goal, check the limits, and own the final decision.
🧠 Memory Phrase: People set the purpose. People shape the system. People own the result.
🛡️ Use the Responsible AI Routine
Use this pathway whenever AI may support a task.
When to use this routine
Use it before AI supports a school, workplace, or personal task, especially when information, safety, fairness, or other people may be affected.
Your next move
Study the five steps in the image. Choose one step you think is easiest to forget and explain why it matters.
🎯 Learning Goals
By the end of this lesson, you can:
- Explain at least four ways people shape an AI system.
- Tell the difference between a , , , and system.
- Find human choices that affect an AI-supported workplace task.
- Explain what AI can do and what a trained person must still check or decide.
- Explain how one human choice could make an AI result better or worse.
✅ You will show success by: naming the AI system's purpose and type, explaining one human choice and one limit, identifying who owns the final decision, and suggesting one improvement.
⏱️ Pacing at a Glance
| Lesson Step | Time |
|---|---|
| 1.2 Who Is Really Making the Decisions? | 9 minutes |
| 1.3 The People Behind an AI System | 8 minutes |
| 1.4 Different AI Systems, Different Limits | 7 minutes |
| 1.5 Human Decisions Map Simulation | 14 minutes |
| 1.6 Lesson Assessment | 12 minutes |
| Total | 50 minutes |
📚 Key Words to Know
How people shape AI
: Computer systems built by people to find patterns, make predictions, suggest choices, or create content.
: A choice people make about an AI system's goal, data, design, testing, rules, or use.
: Examples used to help an AI model learn patterns.
Common AI model types
: A model that puts an input into a group.
: A model that estimates something unknown or in the future.
: A model that ranks or suggests choices.
: A model that makes new digital content from patterns and user directions.
Step 1.2 · Learn and check
Connect & Learn: What Can and Cannot Do
Estimated time: 9 minutes
🚗 Start with a workplace decision
A vehicle-diagnostic system flags a possible transmission problem. The technician inspects the vehicle and finds that a damaged wheel-speed sensor caused the unusual data.
Pause and think: What did the AI system do? What choices did people make before the system was built? What must the technician decide now?
▶️ Your next move: Watch for how uses patterns to create new content.
📖 Your next move: Read the short excerpt below. Look for the decisions people make before an AI system ever produces an answer.
AI is a broad group of computer systems built to do specific tasks. AI can find patterns, sort information, make predictions, suggest choices, or create content. It may be part of software, machines, vehicles, medical tools, games, or workplace systems.
AI does not choose its own purpose. People decide what problem it should help solve, what data it may use, and how its output will be used. These choices affect who benefits, what could go wrong, and who is responsible for the final decision.
Key takeaway: AI performs a designed task. A person must decide whether the result fits the real situation and what action should follow.
Check 2
An AI system gives a workplace result. What must a person do before using it?
AI can flag a pattern, but a person must compare it with the real situation and make the decision.
Check 3
A vehicle-diagnostic system flags a possible transmission problem. What should the technician do next?
The system may flag a pattern, but a qualified technician must inspect the evidence and make the final diagnosis.
Step 1.3 · Learn and check
Learn: How Learns from Data
Estimated time: 8 minutes
▶️ Your next move: Watch for one way people help train a model.
📖 Your next move: Read the six stages below. At each stage, ask what could improve the system and what could introduce a problem.
1. Set the purpose
People decide what problem the system should help solve and whether AI is the right tool.
2. Choose and label data
People select examples, organize information, create categories, and decide what the data means.
3. Choose and design the model
People select the features, rules, tools, and model type that fit the goal.
4. Define success
People decide what a useful, accurate, fair, and safe result should look like.
5. Test and improve
People compare results with evidence, look for mistakes or uneven outcomes, and make changes.
6. Use and monitor
People decide where the system may be used, who may rely on it, what checks are required, and when it should be changed or stopped.
Data matters because AI learns patterns from the examples people provide. Results may be weaker when data is missing, wrong, unreliable, private, or does not include the situations the system will face.
👀 Your next move: Study the diagram below. Follow the process from training examples to model refinement, and notice where human choices shape the result.
Key takeaway: AI development is not an automatic process. People shape the purpose, data, model, tests, safeguards, and final use.
Check 1
Which action is human data curation?
Data curation includes selecting, preparing, labeling, and reviewing the examples used to train a system.
Check 2
What should a team measure before it approves an AI system? Choose all correct answers.
Useful success measures address accuracy, fairness, safety, and usefulness for the intended task.
Check 3
What can happen when training data misses important situations?
Missing or unrepresentative data can weaken performance in situations or groups the training data does not reflect.
Step 1.4 · Learn and check
Learn: AI Types and Human Responsibility
Estimated time: 7 minutes
▶️ Your next move: Watch for one example of a task that people designed AI to do.
This video may display YouTube ads.
📖 Your next move: Read the model cards below. Compare each model’s purpose, workplace use, strength, and limitation.
Classifier
Purpose: Puts an input into a group.
Workplace example: Labels an equipment image as normal, damaged, or unclear.
Strength: Uses the same learned groups quickly.
Limit: Results depend on the examples and labels people chose.
Predictor
Purpose: Estimates something unknown or in the future.
Workplace example: Estimates the chance that a machine part may fail.
Strength: Finds patterns in many records or measurements.
Limit: A prediction is not a fact. It may be wrong when data is missing or conditions change.
Recommender
Purpose: Ranks or suggests choices.
Workplace example: Suggests training resources, products, repair steps, or menu ideas.
Strength: Helps narrow a large set of choices.
Limit: Suggestions may be based on old data, popularity, or goals that do not fit the person.
Generative AI
Purpose: Makes new digital content from patterns and user directions.
Workplace example: Drafts a handout, design idea, code sample, or project outline.
Strength: Makes drafts and options quickly.
Limit: A result may look good but still be wrong, unsafe, copied, or incomplete.
Professional check: One tool may use more than one model type. Ask what job it was built to do, what facts support the result, and what a person must still decide.
Check 1
Match each AI model type with the workplace output it would most likely produce.
Each model type is named for its main task: classify, predict, recommend, or generate.
Check 2
Categorize each workplace output by the type of AI model that would most likely produce it.
The tasks align with the models’ main purposes: sorting, estimating, ranking, and creating.
Check 3
What can go wrong with generative AI? Choose all correct answers.
Generative AI can produce polished content that is inaccurate, incomplete, unsafe, or inappropriate.
Step 1.5 · Practical application
Apply Your Learning: Human Decisions Map
In this activity, you will decide how AI can support a workplace task without replacing human judgment.
Interactive practice
Make a Responsible AI Decision
Choose a program
Select the program that fits you best. Work through a four-part decision pathway, then compare your explanation with an exemplar.
After you complete your scenario, you may choose another program for optional practice.
Your selected program: Agriscience - Vet Med
Situation: An AI tool flags a dog's symptoms as a possible infection after a student enters notes from a visit.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Use the AI result as one clue, then have a qualified veterinary professional examine the animal and review the evidence. The tool can identify patterns, but a qualified person must check the animal's condition and make the care decision.
Your selected program: Automotive - Medium Heavy Diesel
Situation: A diagnostic tool suggests replacing a truck's fuel injector based on an error code.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Inspect the truck, compare the error code with test results, and have a trained technician decide on the repair. An error code can guide an inspection, but it does not replace testing or a trained technician's decision.
Your selected program: Aviation
Situation: A planning tool recommends a route change because it predicts better weather conditions.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Check approved weather, route, and safety information before a qualified person decides whether to change the plan. A recommendation is not a final flight decision. Approved sources and qualified human review are required.
Your selected program: Computer Science
Situation: An AI coding assistant suggests a change that would give users more personalized results.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Review the code, test it with appropriate examples, and check whether it protects user information before using it. The team must test the suggestion and protect people's information before adding it to a product.
Your selected program: Construction Trades
Situation: A tool predicts that a framing crew can finish a job faster by changing the work sequence.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Compare the suggestion with the plans, job-site conditions, and safety rules before a trained supervisor approves a change. Speed predictions cannot replace safety requirements, site conditions, or qualified supervision.
Your selected program: Cosmetology
Situation: A scheduling tool recommends adding an extra client appointment to a stylist's day.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Have the stylist review the schedule, client needs, and available time before deciding whether to add the appointment. The tool can organize options, but the person responsible for the service must decide what is realistic and appropriate.
Your selected program: Culinary
Situation: A menu-planning tool suggests a new dish for an event based on popular online recipes.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Check ingredients, allergies, kitchen equipment, cost, and event needs before the chef decides whether to use the dish. A popular recipe may not fit the event. A trained person must check safety, resources, and the full situation.
Your selected program: Education
Situation: A writing tool suggests that a student should receive extra reading support after reviewing a short assignment.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Use the suggestion as one piece of information, then have an educator review the student's work, progress, and needs before deciding on support. A short sample cannot show the whole learner. An educator must consider more evidence and make the support decision.
Your selected program: Health Science - Medical Careers
Situation: An AI tool highlights a patient note as needing quick follow-up.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Have a qualified health professional review the patient record, current condition, and approved procedures before deciding what happens next. The alert may help prioritize review, but it cannot make a care decision or replace professional judgment.
Your selected program: Marketing - Graphic Design
Situation: An image tool creates a campaign graphic that includes a familiar-looking logo and a claim about a product.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Check permission to use the visual elements and verify the claim with approved sources before a person approves the graphic. A polished graphic still needs permission checks and accurate claims before it is shared.
Your selected program: Mechatronics
Situation: A maintenance system predicts that a machine part may fail soon and suggests stopping the line immediately.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Inspect the machine, compare the alert with sensor data and safety procedures, and have a trained person decide on the response. The prediction can guide maintenance, but qualified people must check the evidence and balance safety with the full work situation.
Your selected program: Public Safety
Situation: A dispatch support tool ranks calls by predicted urgency.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Use the ranking as one input, then have trained staff review the details and decide how to respond. Predictions can miss important context. Trained people must make the response decision and remain accountable.
Your selected program: US Army JROTC
Situation: A planning tool suggests who should lead parts of a team exercise based on past performance notes.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Have the instructor and team review current skills, goals, and fairness before assigning leadership roles. Past notes do not tell the full story. People must make fair decisions and take responsibility for team roles.
Your selected program: Welding
Situation: A tool recommends welding settings based on the material listed in a work order.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Check the actual material, equipment condition, job requirements, and safety procedures before a trained welder sets up the work. The tool may offer a starting point, but a trained welder must verify the real conditions and follow safety procedures.
Your selected program: Forensic Science
Situation: A software tool marks part of a comparison image as a possible match.
1. Spot the risk
What is the most important concern to check first?
2. Protect information
What information should stay private or be left out of an AI prompt?
3. Choose the human check
Who should review the AI suggestion before a decision is made?
4. Make and explain the decision
Compare your response with this exemplar: Treat the result as a lead, then use approved methods and qualified expert review before drawing any conclusion. A possible match is not proof. Evidence needs careful human review and documented methods.
Your required scenario is complete. You may stop here or choose another program for optional practice.
Step 1.6 · Lesson assessment
Show What You Know: Humans, Models, and Responsibility
Your place in this lesson is saved automatically on this device.