Explore the lesson here. Canvas holds your check-ins, submissions, feedback, and progress.

← Learning path

Unit 01 of 12

AI capabilities and useful tasks

Which AI capability fits a real need, and how will a person check it?

Learn

By the end of this unit

  • Match a concrete task to an AI capability, permitted inputs, and useful output.
  • Choose a practical no-code route and inspect a supplied output for a consequential error.
  • Distinguish a candidate output, verified evidence, and an authorized action.

Planned guided work including the short textbook read-and-apply: 77–127 minutes, plus other project work. This is a design estimate, not measured learner time.

How to read this lesson

The labels describe how to use a passage. Color does not prove a claim is true or that a response is correct.

  • Explain — Understand the idea.
  • Try / predict — Make a choice or prediction before the reveal.
  • Evidence — Inspect the named source or supplied data.
  • Worked reasoning — Compare the steps after your own attempt.
  • Caution — Check a limit, assumption or unsupported conclusion.

Each figure has its own local legend for the data. Blue selection outlines mark the current selection; other outlines are explained by each figure’s local legend. Selection is not a correctness judgment.

Explain

Begin with the work someone needs done

AI can help with a personal task, a community service, creative work, research, or an organizational process. Start with the person and intended result. “Use AI” is not a task. “Turn permitted event notices into a checked dates-and-locations table” is a task you can try and evaluate. No prior AI, programming, statistics, or business study is required; we introduce concepts when you need them.

A capability map: from a need to something usable

A capability is a kind of work an AI tool may support. Availability depends on the selected tool, input format, access, and settings. One tool may combine capabilities; do not assume it supports every row. The following are original learning examples, not verified performance claims.

Seven AI capabilities, possible applications, and no-code implementation choices
CapabilityNeed → possible outputNo-code route and essential check
Generate or transform textA volunteer needs a short version of a long approved notice → a plain-language draft.Give an available assistant the permitted text, audience, and format; compare the draft with the original for lost conditions and invented details.
Answer from sourcesA researcher or participant needs an answer within a document set → an answer with source passages, or “not supplied.”Provide approved passages in a document or suitable source-aware assistant; verify that the passages actually support each material claim.
Extract or classifyA community group needs event details organized → a table of dates, places, and missing fields; requests may also be sorted into defined categories.Specify columns or categories in a prompt or visual tool; check every field against the input and test ambiguous cases. A guessed date is not a completed record.
Analyze data or predictA project organizer needs to understand attendance → a summary, chart, or conditional estimate for the next session.Use an approved spreadsheet or visual analysis tool with suitable data; recalculate totals, compare with a simple baseline, and test estimates on held-out or later observations. A chatbot's plausible number is not a validated prediction.
Interpret or create imagesA creative project needs concepts or an image description → candidate illustrations or a proposed description.Use a suitable image-capable tool with permitted inputs; inspect factual details, required text, accessibility, and permission to use the material. Do not treat a generated scene as a photograph of an event.
Transcribe, translate, or speakA group needs an accessible version of a permitted recording or notice → a transcript, translation, or spoken draft.Use a tool that supports the required language or audio task; compare names, numbers, meaning, and missing sections with the original. Obtain permission before using another person's recording.
Coordinate a workflow or agentA team needs requests prepared for review → information is collected, a draft is prepared, and exceptions are routed.Use a checklist or visual workflow with explicit handoffs. A workflow follows specified steps; an agent can select actions within its given tools and permissions. Test stop rules and keep consequential actions under the responsible person's control.
Optional background for the capability map

Microsoft Learn introduces generative AI and agents, text analysis, information extraction, machine learning, computer vision and image generation, speech, and source-grounded answers. These are optional conceptual references; their technical prerequisites, accounts, exercises, and product choices are not course requirements.

One continuing case makes the process concrete

Cedar Desk needs an accurate, useful reply to a booking question. Its ReplyKit combines source-grounded answering and text drafting inside a human-reviewed workflow. The coordinator finds the policy, checks conditions, and avoids promising an action that has not happened. The same pattern can support an equipment-loan service, a community information desk, or a research-document helper; your one capstone can address an appropriate need from any discipline.

Read the process from left to right

01 · InputsQuestion + source

A permitted request and approved information.

02 · AI assistsPrepare a candidate

Use the chosen capability to produce a draft or structured output.

03 · Human decidesCheck the result

Verify claims, consequences, and authority to act.

04 · OutcomeUseful output or handoff

A checked result, or an exception for the responsible person.

Arrows show work moving, not permission to act. The AI proposes; a responsible person checks and authorizes.
Trace an exception: an emergency cancellation

A participant cancels eight hours before a workshop. The source allows no automatic refund or credit under 24 hours. An AI draft promising a full refund must be held and routed to the owner. Fluent wording does not establish authority.

You can operate a useful solution with working documents, a spreadsheet, an available AI tool, or a visual no-code tool. A supplied-output route lets you inspect the same decision without an AI account: label its AI step simulated using a supplied output. It does not establish that you connected a live model or automated a process. Both routes still need usable inputs, a check, an output or handoff, and an observed user task.

A bounded instruction you can copy and improve
Using only sources S1–S6 in the Cedar Desk packet, prepare a concise reply to this fictional booking question. List the source for each material claim. If a needed fact is missing, ask for it. Do not promise availability, policy exceptions, or actions already completed. Return a draft, a source table, and any item needing human review.

This specifies a task, evidence boundary, audience, format, and stop condition. It does not guarantee correctness. Save the original output and your checks.

Required textbook reading and application · about 12 minutes

Read, then apply one idea

Read only the passages listed here. Mathematical derivations, algorithms, code, and end-of-chapter exercises are not assigned. The time estimate includes the brief passage and one connection to your existing project update; reading the passage alone usually takes much less time. It is a planning estimate, not a timer.

1.1: What is Artificial Intelligence?

David L. Poole and Alan K. Mackworth. Artificial Intelligence: Foundations of Computational Agents. 3rd edition. Cambridge University Press, 2023.

Read: Read the opening of section 1.1 only. Stop at the heading 1.1.1 Artificial and Natural Intelligence. Do not follow links or read later subsections.

Why this unit: Introduces agents, goals, limited information, adaptation, and human assistance without requiring programming.

Course connection: In this course, a useful AI capability is a task you can observe and check: summarize, extract, classify, generate, recommend, or help carry out a bounded workflow. A fluent answer alone does not establish that a task was completed correctly. Current tool demonstrations and limitations come from the unit guide, rather than treating a 2023 book as a live product catalog.

Apply one idea: Choose one prompt for the brief connection in your existing project update.

  • Name one real task, the input the tool would receive, the output you need, and one check a person must make before relying on that output.
Read this textbook section here

The official page below may include material before or after your assigned selection. The reader does not automatically scroll or crop to that selection.

Assigned selection — 1.1: What is Artificial Intelligence?: Read the opening of section 1.1 only. Stop at the heading 1.1.1 Artificial and Natural Intelligence. Do not follow links or read later subsections.

Official textbook section ↗

Add a brief connection to your existing project update: identify the book and section, explain one relevant idea in your own words, and show the decision it informs. This is part of the same update, not another assignment.

If the official reader is unavailable or inaccessible, use the original lesson explanation and worked example, apply the same question, and label that access alternative honestly. Report the problem through the course support route; do not claim to have read an unavailable text.

Optional further reading

The AI Fluency Framework — short guide

Open the short guide here. Identify one decision about delegation, one instruction, one output check, and one responsibility in your task. No account or external PDF reader is needed for this course explanation.

Read the short guide here

Four habits for working with AI. This original course explanation applies the AI Fluency Framework to your project; it is not a reproduction of the publisher's one-page guide.

Delegation — choose the work
Decide whether AI is useful for this task, which step it may assist, and which decisions a person must retain.
Description — make the task clear
Give the permitted inputs, goal, audience, output format, constraints, and a clear stopping rule.
Discernment — inspect the result
Compare the output with sources and success criteria. Check for missing conditions, invented claims, and whether it helps complete the intended task.
Diligence — remain responsible
Own the decision to use or release the result. Protect information, disclose material AI assistance, and arrange human review or escalation where needed.

Apply to your task: Write one AI-assisted step, one instruction, one output check, and one responsibility you keep. Use your existing project record; this is optional practice, not another submission.

Framework attribution: Rick Dakan, Joseph Feller, and Anthropic, The AI Fluency Framework (2025). The definitions above are course-authored paraphrases and project prompts. The original framework guide is released under CC BY-NC-SA 4.0.

Official framework — original one-page PDF ↗

Try

Case file: approved Cedar Desk facts
Supplied source facts

Fictional classroom facts. Use these to check every policy claim in the exercise.

Use these supplied facts when drafting or checking replies
SourceApproved fact
S1 — SessionA Standard Workshop lasts 90 minutes and costs $40 per participant, including materials. Participants must be at least 16.
S2 — CancellationAt least 72 hours before the session: refund eligibility. From 24 hours up to, but not including, 72 hours: credit valid for 90 days. Under 24 hours: no automatic refund or credit; the owner reviews exceptional requests. Do not promise a payment-processing date.
S3 — ReschedulingA request made at least 48 hours before the session may be considered, subject to available space and coordinator confirmation. No availability is supplied in this packet.
S4 — Missing factsAsk only for the missing fact needed to apply a policy. Classroom booking references are fictional. A draft is not confirmation that a booking was changed or a refund was processed.
S5 — Special requestsGroup prices, accessibility arrangements, age exceptions, and policy exceptions require owner review. The packet supplies no basis for promising them.
S6 — ReleaseA person checks the source, conditions, tone, and next step before a reply leaves the service. A human decides exceptions and authorizes commitments.
Try / predict

Choose a capability, then inspect its output

Before opening the worked answer, select a first capability for each task: (A) turn two event notices into dates and locations; (B) answer a question from approved policy passages; (C) describe weekly attendance and consider a next-week estimate. Name the input, output, and one check. More automation is not automatically a better choice.

Original supplied examples; inspect without an account or live tool
Task and permitted inputSupplied candidate outputYour decision
A — ExtractionNotice 1: “Repair meetup, 12 May, Library Room 2.” Notice 2: “Garden exchange, 15 May; location to follow.” Candidate: Repair / 12 May / Room 2; Garden / 15 May / Room 2.Which field is unsupported? What should replace it?
B — Source answerS3: rescheduling at least 48 hours ahead may be considered, with space and coordinator confirmation. A 60-hour request receives: “Your new session is confirmed.”What does the source support, and what remains unknown?
C — Analysis or predictionFour weekly attendance counts: 18, 20, 19, 23. Candidate: “Mean 20. Next week will definitely be 40 because AI predicts growth.”Which calculation is checkable? Which prediction lacks support?
Compare your capability choices and checks

A: Extract stated fields into a table; the second location is missing, not Room 2. Preserve “not supplied” and ask the owner if needed. B: Source-grounded answering can prepare a conditional reply, but no availability, destination, or completed action is established. Ask which session to check. C: Simple analysis gives (18 + 20 + 19 + 23) ÷ 4 = 20. A mean is not proof of next week's attendance. Four counts and a confident sentence do not validate a forecast; begin with a simple baseline and define a future check before making commitments.

Repair ReplyKit's unsupported answer

For the 60-hour booking request, the supplied AI draft says: “Your place has been moved, and there is no charge. Your new session is confirmed.” Mark each claim supported, unsupported, or unknown. Write a revised reply and its source/claim table before revealing the explanation.

Worked review

The timing meets S3, but space and the requested destination are unknown. No booking action occurred. “Moved,” “no charge,” and “confirmed” are unsupported. A sound revision is: “A change can be considered because your session is at least 48 hours away. It depends on available space and coordinator confirmation. Which session would you like us to check? Your booking has not yet been changed.”

Now change the request to 20 hours. The 48-hour rule no longer applies: prepare owner review. Varying a consequential fact tests whether the solution follows the source rather than repeating fluent wording.

Decision lab · practice with your Learning Companion

The confident wrong answer

Match an AI capability to a bounded task and identify what a person must verify.

Take the role of capability detective. Commit to a decision, uncover evidence, revise your reasoning, then carry one action into Build. Use this as the unit’s reasoning rehearsal; reuse the response in your existing project update.

Open the AI Learning Companion section below this interactive lesson to enter this unit’s decision lab. Return here for Build. You can also use the complete reading route below.

Complete reading route · no AI service needed

Fictional teaching scenario. A fictional campus workshop desk wants help answering questions from an approved information sheet. A visitor asks for a detail missing from that sheet. Choose the first version you would build.

  1. A fluent answer generator
  2. A source-based draft with human review
  3. An autonomous desk agent

Choose one and write why before revealing the evidence.

Reveal evidence and challenge
Supplied evidence
  • Approved source: The sheet describes ordinary workshop information. It does not specify the requested exception.
  • Candidate output: The candidate confidently supplies an exception rule without citing any approved source.
  • Simpler alternative: A searchable FAQ can answer ordinary questions and route exceptions to a person.

What should the assistant do when the source is silent? Explain one useful output, one human check, and why a simpler FAQ might win.

Compare with the worked reasoning
Worked reasoning

A source-based draft can support this task, but retrieval does not make an absent rule true. Identify the gap and route the question to the responsible person. Compare the draft workflow with the simpler FAQ using the same task. This is a fictional mini-case, not an actual desk policy.

Apply to your project: Name your project’s input, AI capability, useful output, and one fact or action that a person must check.

Write your initial choice, evidence-based revision, and project action in your own document.

Practice boundary

Practice records are not grades or Canvas submissions.

Build

Apply this step to your chosen problem. Cedar Desk is the supplied classroom case; keep one project throughout the course.

Open the complete classroom packet

Your continuing case: Cedar Desk

Fictional teaching case. Cedar Desk is a booking and customer-support service for small workshop businesses. Its coordinators answer questions about bookings, cancellations, rescheduling, and special requests. The operations owner wants dependable replies without unnecessary checking or repeated clarification. Your project is ReplyKit: a small source-grounded reply service that turns a permitted question into a checked response or a useful human handoff. It can run with documents, a calculator, and an available AI assistant or the supplied draft. No application or live customer account is needed.

The user is a booking coordinator. The responsible decision owner is the operations owner. A customer receives the final approved reply. Those are three different roles. The owner has agreed only to a classroom exploration in this fictional case, not to a purchase, a live integration, or a commercial launch. Your instructor may replace this packet with an approved partner brief while preserving the same tasks and evidence standards.

Approved source card: Standard Workshop policy, version 1

Use these supplied facts when drafting or checking replies
SourceApproved fact
S1 — SessionA Standard Workshop lasts 90 minutes and costs $40 per participant, including materials. Participants must be at least 16.
S2 — CancellationAt least 72 hours before the session: refund eligibility. From 24 hours up to, but not including, 72 hours: credit valid for 90 days. Under 24 hours: no automatic refund or credit; the owner reviews exceptional requests. Do not promise a payment-processing date.
S3 — ReschedulingA request made at least 48 hours before the session may be considered, subject to available space and coordinator confirmation. No availability is supplied in this packet.
S4 — Missing factsAsk only for the missing fact needed to apply a policy. Classroom booking references are fictional. A draft is not confirmation that a booking was changed or a refund was processed.
S5 — Special requestsGroup prices, accessibility arrangements, age exceptions, and policy exceptions require owner review. The packet supplies no basis for promising them.
S6 — ReleaseA person checks the source, conditions, tone, and next step before a reply leaves the service. A human decides exceptions and authorizes commitments.

Discovery packet

Eight supplied, fictional interview records describe waits of 4, 5, 5, 6, 6, 7, 8, and 31 minutes for a response. Six of eight describe repeated delay; three of eight say they would try a pilot. These convenience-case records are teaching inputs, not a representative market survey. They do not establish purchase intent. Response waiting time is different from the staff processing time in the later benchmark.

  • Coordinator: “I often know the basic answer. Finding the latest policy and asking about exceptions takes longer.”
  • Customer: “My last reply was quick, but I still did not know whether the date change was confirmed.”
  • Owner: “I might try a small test if I can inspect every reply. I have not approved a budget.”
  • Contrary evidence: Two of the eight records report no repeated delay. One of these customers prefers a clear policy page to another conversation.

These are invented case statements, not quotations from real people. In a discovery role-play, the respondent uses the card and says “unknown” when the packet does not answer a question. Record an actual role-play as an actual classroom event using fictional facts; do not relabel it a customer interview.

Twelve supplied test inputs

Bench cases and the expected task behavior
CaseInputRequired response or handoff
1Cancel 96 hours before the session.Explain refund eligibility; do not claim a refund was processed.
2Cancel 48 hours before the session.Explain 90-day credit eligibility, not a refund.
3Cancel 8 hours before the session because of an emergency.No automatic promise; send the exception to the owner.
4Cancel, but the time until the session is missing.Ask for the missing time before choosing a policy branch.
5Move a booking 60 hours before the session.State the space/confirmation conditions; do not confirm a seat.
6Move a booking 20 hours before the session.Escalate rather than extend the published rescheduling rule.
7Book for a 14-year-old.State the age condition; an exception needs owner review.
8Ask the duration and whether materials cost extra.Give 90 minutes and $40 including materials.
9Cancel exactly 72 hours before the session.Apply the refund-eligibility boundary correctly.
10Cancel exactly 24 hours before the session.Apply the credit boundary correctly.
11Ask whether a particular access arrangement is guaranteed.Ask the owner to confirm; do not invent an arrangement.
12Request a special price for a group of 20.Prepare a quotation request for the owner, not a made-up price.

A complete usable example

Intake: Case 5; request to move a booking 60 hours before the session. Sources: S3, S4, S6. Draft: “Your request is at least 48 hours before the session, so a change can be considered. It depends on available space and coordinator confirmation. Which session would you like us to check? Your booking has not yet been changed.” Review: Timing, conditional wording, missing desired session, and no false confirmation checked. Handoff: A coordinator checks availability and confirms or declines the requested change. Status: Approved draft for classroom use; no external message sent.

Make the example usable by another person: provide this source card, an intake sheet, a draft-and-source table, the review checklist, an exception note, and a delivery record. Number the version and label who performs each step. A slide showing those items without providing them is not a working kit.

Keep four datasets separate

What each supplied dataset can support
DatasetValuesEvidence limit
DiscoveryEight waits; six report repeated delay; three would try a pilot.Supports questions about this fictional sample, not demand in a population.
Processing benchmarkSame 12 cases: manual 96 minutes and 12 acceptable replies; unchecked AI 24 and 9; reviewed AI 84 and 12.Supplied performance comparison, not your actual user test.
Monthly planning scenario1,200 eligible replies; 50% planned adoption; one minute released per assisted reply; $24/hour capacity value; $180 tools plus $60 support monthly.A conditional forecast. Adoption, transfer of timing, and costs require verification.
Adoption cohort12 invited, 9 started, 6 completed, 4 returned within 14 days after invitation.Return means completing another task in that same 14-day window; not a forecast for all coordinators.

Your actual evidence starts with what you do. Use the packet to practice and build. Then ask a classmate, instructor, or approved partner user to perform a bounded task with your version. Record the real attempt, assistance, result, revision, and retest. Leave fields blank until the event happens. A self-review is useful preparation but is not another person’s user test.

Download the case packet as text

Start one project record and make a capability choice

Use ReplyKit or choose one appropriate personal, community, creative, research, or organizational need. Create one project record with five sections: need and evidence; solution and workflow; tests and revision; impact and adoption; operation and handoff. Keep this project throughout the course.

  1. Write the intended user, task, and useful result in one sentence.
  2. Select one primary capability from the map. Compare it with one other capability or a simpler non-AI method. Explain why the input and output fit your choice.
  3. Record permitted inputs, expected output, available no-code execution route, a likely failure, and the human check or stop condition. Name a responsible decision owner; for an appropriate personal task, this can be you.
  4. Save a supplied or actual AI output, your correction, and a source or calculation check. Label an actual tool and version when known; label supplied outputs and the simulated AI step honestly.
  5. Write one uncertainty about the underlying need. Do not invent an interview, a tool connection, or a successful use to resolve it.

For ReplyKit, choose source-grounded answering plus drafting; use the source card, bounded instruction, draft, and review table above. Another project may need extraction, images, speech, data analysis, or a bounded workflow instead. Use only permitted information and capabilities your available tool supports. The next unit investigates whether the underlying problem is worth solving.

Add your project evidence and reflection to the check-in in Canvas.

Your project record

Write here or use your own document. Notes stay in this browser; export a backup to keep them. Nothing is submitted automatically.

Finish

What will you carry forward?

Your Canvas check-in

A community group needs dates and locations from permitted event notices in a table. Which AI capability best fits the first version?

Use the evidence from this lesson to explain your reasoning in Canvas. Continue with Canvas Next below; submit your check-in in Canvas.

The simulation and Learning Companion support practice. They do not submit work, record Canvas completion, or assign a grade.

A community group needs dates and locations from permitted event notices in a table. Which AI capability best fits the first version?

In your project record, state what you decided, the evidence you used, and one uncertainty to test next. Explain the reasoning in your own words.

You can revisit any step. Completion records study progress; it does not confirm a submission, grade, or project quality.