# Responsible AI project workbook

Universal AI Institute — Research. Education. Equity.

Use fictional or shareable public data. Keep the prompts, outputs, and your own review notes.

## Protect the people behind the data

redesign an input
1. Invent a six-field record for a scheduling task. Include two fields that are clearly unnecessary.
2. Identify the task’s purpose and the minimum useful input.
3. Remove unnecessary fields and replace the remaining values with fictional examples.
4. Ask whether any combination of remaining details could still identify a real person if this were actual data.
5. Write the condition under which you would stop and seek an approved process instead.

**Your finished artifact:** An original fictional record, a minimized version, and a five-line sharing decision. Do not use actual personal records for the exercise.


### Your work

Input/source:

Prompt:

Output:

What I checked:

What I changed and why:

What I will try next:

## Spot bias and missing perspectives

redesign the exercise
1. Identify three assumptions in the activity above.
2. Explain the practical barrier created by each assumption.
3. Design an alternative that does not require paid access or private documents.
4. Check whether the alternative still teaches source verification.
5. Ask a peer to review the instructions for clarity. If no peer is available, describe the feedback you would want before running the activity.

**Your finished artifact:** A revised activity and a brief explanation connecting each change to a participation barrier and the learning goal.


### Your work

Input/source:

Prompt:

Output:

What I checked:

What I changed and why:

What I will try next:

## Documents, tools, and agents—without the jargon

write a permission card
1. Define one information task for the assistant.
2. List the exact sources it may read.
3. List the actions it may take and the actions it may not take.
4. Define what evidence it must return with an answer.
5. Define when it must stop and ask for human help.
6. Test your design with a missing policy and an unrelated instruction embedded in a fictional source.

**Your finished artifact:** A one-page permission card and two test scenarios. You do not need to build or connect a live agent to complete this exercise.


### Your work

Input/source:

Prompt:

Output:

What I checked:

What I changed and why:

What I will try next:

## Your first reviewed AI project

build, test, and revise
1. Write your brief and three quality criteria. Include factual accuracy and audience fit.
2. Prepare a source packet that you are allowed to share. Label fictional material clearly.
3. Create a first draft with AI assistance and preserve the prompt.
4. Test normal, missing, and conflicting inputs. Record the outputs and failures.
5. Correct the draft and revise the instructions where useful. Rerun all three tests.
6. Review privacy, unsupported claims, and barriers to participation.
7. Produce a final artifact and a short project note explaining what changed.

**Your finished artifact:** The final deliverable, source packet, prompt, test log, and a 200-word project note. Another person should be able to understand how you reached the result without trusting the assistant’s confidence.


### Your work

Input/source:

Prompt:

Output:

What I checked:

What I changed and why:

What I will try next: