AI for Everyone study guide · Lesson 7 of 14

Research, Learning, and Decisions

Plan Research Before Answering · Use AI as a Practice Partner · Compare Options without Outsourcing Choice · Six Thinking Moves

This is the lesson's study guide from the FlashFeed app, plus its key facts and a self-test.

What you'll be able to do

  • Use AI to structure inquiry, practice learning, compare options, and perform distinct thinking moves without outsourcing judgment.

Plan Research Before Answering

AI can help convert a broad question into subquestions, search terms, inclusion criteria, and an evidence table. Start by asking what would count as an answer and which claims require primary, current, or local sources. This creates a plan you can inspect before reading polished prose.

Generated citations may be invented, misquoted, or merely adjacent to the claim. Open every material source, confirm authorship and date, and check that the cited passage actually supports the sentence. Treat summaries of sources as leads until verified.

Three ideas to retain:

  • Research Starts with Claim Types: A historical date, medical recommendation, product price, and personal preference require different sources and levels of verification.
  • Search Terms Are Hypotheses: Suggested queries help explore a topic, but wording can bias what appears. Try synonyms, counterarguments, and source-specific searches.
  • A Citation Must Support the Claim: A real link is not sufficient if its content does not entail, qualify, or directly document the sentence it is attached to.

Key term: Primary source means “original evidence or an authoritative first-party record.”

Use AI as a Practice Partner

A tutor is most useful when it makes the learner do cognitive work. Ask for one question at a time, wait for your attempt, request a hint before a solution, and have the system diagnose the misconception. Use retrieval practice and varied examples rather than rereading a generated explanation.

The learner still needs trusted source material and human support where appropriate. AI feedback may be persuasive and wrong. Compare explanations with the course text, teacher guidance, worked solutions, or another authoritative source, and do not upload protected student data casually.

Three ideas to retain:

  • Practice Beats Passive Answering: Questions, retrieval, explanation, and correction require the learner to produce knowledge instead of only reading a generated solution.
  • Hints Preserve Productive Struggle: A staged hint can unblock progress without immediately replacing the learner's reasoning.
  • Tutoring Output Needs a Reference: A model can give confident but incorrect feedback, so learning sessions should connect to a trusted text, teacher, or answer key.

Key term: Retrieval practice means “recalling knowledge without looking at the answer.”

Compare Options without Outsourcing Choice

AI can organize options, suggest criteria, surface tradeoffs, and run scenarios. It should not quietly choose your values. State which outcomes matter, distinguish must-haves from preferences, and provide evidence for scores instead of accepting invented numbers.

Use sensitivity analysis: change a weight or assumption and see whether the ranking changes. A fragile recommendation deserves more investigation. Keep irreversible, ethical, legal, medical, and financial decisions with appropriately qualified humans.

Three ideas to retain:

  • Criteria Encode Values: A decision table reflects what the chooser counts, how it is weighted, and which tradeoffs are acceptable.
  • Scores Need Evidence: A neat matrix can manufacture precision if ratings are guesses. Link each material score to data, experience, or an explicit assumption.
  • Sensitivity Reveals Fragility: If a small change in one weight reverses the recommendation, the decision depends heavily on that judgment.

Key term: Must-have criterion means “a requirement an option cannot trade away.”

Six Thinking Moves

“Help me think” hides several distinct jobs. You may need expansion, compression, comparison, critique, reframing, or sequencing. Naming the move gives the model a clearer role and helps you judge whether it was useful.

Move deliberately between modes. Brainstorm without judging, then cluster and compare; critique a chosen option, then plan a test. Do not let a long list impersonate a decision. The human supplies purpose, taste, responsibility, and the final commitment.

Three ideas to retain:

  • Thinking Has Modes: Generating possibilities, reducing information, comparing alternatives, challenging claims, reframing problems, and sequencing action are different tasks.
  • Separate Expansion from Selection: Brainstorming and judging at the same time can narrow the search too early. Generate first, then apply criteria.
  • Lists Are Not Decisions: AI can widen and organize a choice set, but a person remains responsible for values, evidence, and final commitment.

Practice by sorting the six named moves in this section, then choose one real task and explain which move fits it. For consequential work, keep the evidence, permissions, and final judgment with an accountable person.

Sources

Key terms

Evidence table
A structured record linking claims to sources, passages, dates, and confidence notes.
Formative feedback
Information intended to improve learning during practice rather than only score the final result.
Decision matrix
A table that compares options across explicitly chosen criteria.

Key concepts

Brainstorming

  • Expands the option space before judgment.
  • Quantity and variety matter at first.
  • It generates possibilities.

Summarizing

  • Compresses a larger body of material.
  • It selects what matters for an audience.
  • It creates a shorter representation.

Comparing

  • Places alternatives against shared criteria.
  • It makes similarities and differences visible.
  • It examines options side by side.

Critiquing

  • Searches for weaknesses, omissions, and counterarguments.
  • It pressure-tests an idea.
  • Its job is constructive challenge.

Reframing

  • Changes the question or point of view.
  • It can reveal a hidden assumption.
  • It offers a different way to define the problem.

Sequencing

  • Turns a goal into an ordered path.
  • Dependencies and checkpoints matter.
  • It arranges actions over time.

Key facts

Search Terms Are Hypotheses

Suggested queries help explore a topic, but wording can bias what appears. Try synonyms, counterarguments, and source-specific searches.

A Citation Must Support the Claim

A real link is not sufficient if its content does not entail, qualify, or directly document the sentence it is attached to.

Hints Preserve Productive Struggle

A staged hint can unblock progress without immediately replacing the learner's reasoning.

Tutoring Output Needs a Reference

A model can give confident but incorrect feedback, so learning sessions should connect to a trusted text, teacher, or answer key.

Scores Need Evidence

A neat matrix can manufacture precision if ratings are guesses. Link each material score to data, experience, or an explicit assumption.

Sensitivity Reveals Fragility

If a small change in one weight reverses the recommendation, the decision depends heavily on that judgment.

Separate Expansion from Selection

Brainstorming and judging at the same time can narrow the search too early. Generate first, then apply criteria.

Lists Are Not Decisions

AI can widen and organize a choice set, but a person remains responsible for values, evidence, and final commitment.

Matched pairs

Each item on the left goes with the one on the right.

Primary source
Original evidence or an authoritative first-party record
Secondary source
Analysis or synthesis based on other sources
Claim-source alignment
The source directly supports the attached statement
Retrieval practice
Recalling knowledge without looking at the answer
Scaffolded hint
A hint that preserves the learner's next step
Socratic question
A prompt that advances thinking without supplying the conclusion
Must-have criterion
A requirement an option cannot trade away
Sensitivity analysis
Testing how conclusions change when assumptions change
Decision owner
The person accountable for the final choice

Self-test

Answer each question from memory, then check your work against the answer key.

  1. 1. What should happen to an AI citation?

    1. ACount the URL as proof
    2. BAssume the quote exists
    3. COpen and verify it
    4. DTrust the title alone
  1. 2. Why search counterarguments?

    1. AGuarantee consensus
    2. BReduce one-sided evidence
    3. CAvoid primary sources
    4. DMake every answer longer
  2. 3. Which tutor prompt keeps the learner active?

    1. AWait for my attempt first
    2. BSkip feedback
    3. CRemove all questions
    4. DGive every answer immediately
  3. 4. How should AI feedback be treated?

    1. AAssume confidence means truth
    2. BReplace the course source
    3. CHide disagreements
    4. DCheck it against trusted material
  4. 5. What makes a decision matrix deceptive?

    1. AVisible tradeoffs
    2. BSource links
    3. CUnsupported precise scores
    4. DNamed criteria
  5. 6. Why name the specific thinking move (e.g., critique, reframe) you want AI to do?

    1. AGuarantees the AI never makes a factual error
    2. BLets the AI decide your final values for you
    3. CRemoves the need to check citations afterward
    4. DGives the model a clearer role you can judge
  6. 7. Why keep brainstorming separate from judging ideas early on?

    1. AEarly judging can narrow the search too soon
    2. BSelection criteria are invented after listing ends
    3. CBrainstorming alone produces a final decision
    4. DExpansion and selection are one identical task

Go deeper

Answer key

  1. 1. C Open and verify itThe source and its support for the claim both need checking.
  2. 2. B Reduce one-sided evidenceDeliberately testing alternatives helps reveal missing evidence.
  3. 3. A Wait for my attempt firstAn attempt creates evidence of what the learner understands.
  4. 4. D Check it against trusted materialGenerated feedback can be useful and mistaken.
  5. 5. C Unsupported precise scoresNumbers can create false precision when they lack evidence.
  6. 6. D Gives the model a clearer role you can judgeNaming the move gives the model a clearer role and helps you judge whether it was useful.
  7. 7. A Early judging can narrow the search too soonJudging while brainstorming can narrow the search too early; generate first, then apply criteria.

Sources

Download all 14 lessons: PDF · EPUB