COMP 3270 / Module 4 / Unit 11 / Lesson 4

Lesson 11.4 · Weeks 7–8 · Reading: Chapter 8

CLAIM, Epistemic Responsibility, and Corrected Communication

Integrate TRACE with CLAIM. Create claim-evidence verification tables. Produce corrected wording. Communicate uncertainty. Determine the disposition for each claim and exercise epistemic responsibility.

Estimated study time: 5–6 hours Prerequisite: Lessons 11.1–11.3 Module 4 · Weeks 7–8

Table of Contents

1. Lesson Overview

In Lessons 11.1–11.3, you learned to evaluate sources, align evidence with claims, assess independence, read laterally, trace citations, and detect laundering. In this final lesson of Unit 11, you will integrate all of these skills into a complete verification process using the author-approved CLAIM framework.

You will learn to:

This lesson prepares you for the Unit 11 cumulative project and for Assignment 5, where you will apply the complete verification process to an AI-generated passage.

⚖️ Evidence and verification reminder: This lesson is the culmination of Unit 11. You will integrate everything you have learned—from source evaluation to claim-evidence alignment to citation tracing—into a complete verification process using the CLAIM framework. You will also learn to exercise epistemic responsibility: the practice of forming, evaluating, communicating, and acting on beliefs and claims in an accountable way.

2. Learning Outcomes

By the end of this lesson, you should be able to:

  1. Apply every element of the author-approved CLAIM framework to a claim.
  2. Integrate TRACE error analysis with CLAIM verification.
  3. Create a claim-evidence verification table documenting all verification steps.
  4. Draft corrected wording that reflects available evidence and preserves qualifications.
  5. Communicate remaining uncertainty and source limitations.
  6. Exercise epistemic responsibility in forming and communicating claims.
  7. Establish escalation, correction, withdrawal, and stop conditions for problematic claims.
  8. Determine the appropriate disposition for each claim (verified, qualified, contradicted, unsupported, unverifiable, escalate, withdraw, reject).
  9. Complete Assignment 5 using an independent and documented verification process.
  10. Connect Unit 11 learning to Assignment 1 and later research work.

3. Before You Begin

4. Unit 10 Retrieval Practice

The following questions review TRACE and claim analysis from Unit 10. Answer each before revealing the feedback.

Question 1: What are the five steps of the TRACE framework in order?

Show feedback

Answer: TRACE stands for:

  1. T — Identify and classify possible errors (hallucination, fabrication, factual error, unsupported claim, omission, contextual mismatch, invalid inference, false precision).
  2. R — Assess the consequence of each error.
  3. A — Assess the consequence of each error (or prioritise verification based on consequence).
  4. C — Prepare a preliminary correction or qualification.
  5. E — Establish escalation and stop conditions.

Note: Some versions of TRACE combine steps; the key elements are error identification, consequence assessment, prioritisation, correction preparation, and escalation/stop conditions.

Question 2: You identify a claim that appears to be fabricated. According to TRACE, what should you do next after identifying the error?

Show feedback

Answer: After identifying the error, you should assess the consequence of that error. How serious is the fabrication? Does it affect the overall argument? Could it cause harm if used? This assessment informs your priority for verification and your decision about whether to escalate.

Question 3: According to TRACE, what is the purpose of establishing escalation and stop conditions?

Show feedback

Answer: Escalation and stop conditions provide a clear pathway when verification cannot be completed or when a claim is too consequential. They ensure that you do not use or communicate unverified or harmful claims. Escalation means bringing in a subject-matter expert, supervisor, or institutional authority. Stop conditions are the triggers for stopping use or communication of a claim (e.g., fabrication, contradictory evidence, high consequence, inability to verify).

5. Opening Scenario: A Synthetic Interdisciplinary Passage

This scenario contains multiple verification problems. You will return to it throughout the lesson.

📌 Scenario — Community Mental Health Initiative

You are a policy analyst reviewing an AI-generated briefing on a proposed community health initiative. The AI produced the following passage:

Briefing: Community Mental Health Initiative

The proposed Community Mental Health Initiative is supported by robust evidence. A 2021 study published in the Journal of Community Psychology by Dr. Sarah Mitchell and colleagues found that community-based mental health programs reduce depression by 40% (Mitchell et al., 2021). This finding has been confirmed by multiple independent studies (Johnson, 2020; Lee & Park, 2022; Smith, 2023). The Canadian Mental Health Association (CMHA) has endorsed the initiative, stating that “community-based approaches are the single most effective intervention for mental health in Canada” (CMHA, 2022). Furthermore, a 2020 policy document from the Government of British Columbia recommended scaling up community mental health programs across the province. Given this evidence, the initiative should be funded immediately.

Your task: As you work through this lesson, you will analyse this passage using TRACE and CLAIM. For now, consider:

  1. What claims are being made? Decompose the passage into atomic claims.
  2. What possible errors might be present (using TRACE categories)?
  3. What evidence would be required to verify each claim?
  4. What are the consequences if these claims are wrong?
Show guided analysis preview

Atomic claims (preliminary):

  • Claim 1: Mitchell et al. (2021) published a study in the Journal of Community Psychology.
  • Claim 2: That study found a 40% reduction in depression.
  • Claim 3: Multiple independent studies have confirmed this finding (Johnson, 2020; Lee & Park, 2022; Smith, 2023).
  • Claim 4: The CMHA endorsed the initiative and made a specific statement.
  • Claim 5: The BC government recommended scaling up such programs in 2020.
  • Claim 6: The initiative should be funded immediately (recommendation).

Possible errors (TRACE):

  • Fabrication: Mitchell et al. may not exist, or the study may not exist.
  • Unsupported claim: The 40% figure may be exaggerated or misquoted.
  • Contextual mismatch: The CMHA statement may be taken out of context.
  • Omission: Important qualifications about the studies may be missing.
  • Invalid inference: The recommendation to fund immediately may not follow from the evidence.

Consequences: If the claims are false, public funds could be misallocated, and ineffective programs could be implemented, potentially harming communities.

6. Core Content

6.1 The CLAIM Framework

Note: The author-approved CLAIM framework is defined in the course textbook and course materials. If the exact expansion has not been supplied to your instructor, the following operational definition is used.
Operational CLAIM Framework

CLAIM is a structured process for verifying claims and exercising epistemic responsibility. It consists of the following elements:

CLAIM Element 1: Identify and decompose the claim
  • Definition: State the exact claim as a proposition. Decompose compound claims into atomic units.
  • Guiding questions:
    • What is the precise statement being made?
    • Is this one claim or multiple claims?
    • What are the key terms that need to be verified?
  • Example: “Mitchell et al. (2021) found a 40% reduction in depression” → Atomic claim: “A 2021 study by Mitchell et al. reported a 40% reduction in depression.”
  • Common mistake: Verifying a compound claim as if it were a single proposition.
CLAIM Element 2: Locate appropriate evidence
  • Definition: Identify the type of evidence that could support or challenge the claim, and the source types most likely to provide it.
  • Guiding questions:
    • What kind of evidence would be needed to support this claim?
    • Who is most likely to have produced that evidence?
    • What source type (primary, secondary, tertiary) would be most authoritative?
  • Example: For the Mitchell et al. claim, the ideal source is the original journal article (primary) and any subsequent systematic reviews (secondary).
  • Common mistake: Using a tertiary source as the primary evidence when a primary or secondary source is available.
CLAIM Element 3: Assess source authority, relevance, currency, and independence
  • Definition: Evaluate each source on authority, relevance, currency, and independence relative to the claim.
  • Guiding questions:
    • Does the source have relevant expertise?
    • Is the source current enough for the claim?
    • Does the source address the right population, jurisdiction, and context?
    • Is the source independent, or does it share a common origin with other sources?
  • Example: Mitchell et al. may be authoritative if published in a peer-reviewed journal, but the 40% figure may be from a small sample or specific population.
  • Common mistake: Assuming a source is authoritative because it is published, without checking relevance or currency.
CLAIM Element 4: Interpret claim-evidence alignment
  • Definition: For each atomic claim, determine how the evidence relates to it: directly supports, partially supports, supports with qualification, contradicts, or does not address.
  • Guiding questions:
    • Does the source actually support the claim as worded?
    • Are there important qualifications or limitations in the source that are omitted?
    • Does the claim generalise beyond the source’s scope?
  • Example: If Mitchell et al. found a 40% reduction only in a specific subgroup, the claim “reduces depression by 40%” is overgeneralised.
  • Common mistake: Treating partial support as full support.
CLAIM Element 5: Modify, correct, qualify, or reject the claim
  • Definition: Based on the evidence, produce a corrected version of the claim that accurately reflects the evidence, preserves qualifications, and communicates uncertainty.
  • Guiding questions:
    • What should be corrected or qualified?
    • What uncertainty remains?
    • Should the claim be escalated, withdrawn, or rejected?
  • Example: “A 2021 study by Mitchell et al. reported a 40% reduction in depression in a sample of adults in a community program; however, this was a single study with a specific population, and further research is needed to confirm these findings.”
  • Common mistake: Failing to preserve qualifications and uncertainty.

6.2 TRACE-to-CLAIM Integration

TRACE (from Unit 10) and CLAIM (this unit) are complementary frameworks. TRACE helps you identify what might be wrong with a claim; CLAIM helps you verify the claim and correct it.

Table 1. TRACE and CLAIM Integration
TRACE Step CLAIM Connection
T — Identify and classify errors CLAIM Element 1 (Identify the claim) — you cannot classify errors without first knowing what the claim is.
R — Assess consequence Informs the priority and urgency of verification; affects escalation decisions.
A — Assess consequence / Prioritise CLAIM Element 2 (Locate evidence) — high-consequence claims require stronger evidence.
C — Prepare preliminary correction CLAIM Element 5 (Modify/correct) — the final correction is informed by the evidence.
E — Escalation and stop conditions CLAIM Element 5 — determines whether to escalate, withdraw, or reject.

Integration process:

  1. Use TRACE to identify what might be wrong with a claim (fabrication, overstatement, contextual mismatch, etc.).
  2. Use CLAIM to verify the claim by locating evidence, assessing sources, and aligning evidence.
  3. Use TRACE to assess the consequence of the error and prioritise verification.
  4. Use CLAIM to produce corrected wording that reflects the evidence.
  5. Use TRACE to establish escalation and stop conditions based on the consequence of the claim.

6.3 Claim-Evidence Verification Table

A claim-evidence verification table is a structured document that records every step of your verification process for each atomic claim. It is the central tool for Assignment 5.

Table 2. Claim-Evidence Verification Table Structure
Claim ID Original Claim Claim Type Consequence Evidence Required Source(s) Authority Relevance Currency Independence Alignment Assessment Correction Uncertainty Status
C1 Mitchell et al. (2021) found 40% reduction Quantitative High Original study, peer review Mitchell et al. (2021) Medium (if exists) High 2021 N/A (single source) Direct support? Check Not verified Pending High Unverified

When to use the table:

  • For every consequential claim in an AI-generated passage.
  • To document your verification process for Assignment 5.
  • To communicate your findings to others (e.g., supervisors, colleagues).
  • To ensure you have not missed any claims.

6.4 Epistemic Responsibility

Definition — Epistemic Responsibility

Epistemic responsibility concerns responsible practices for forming, evaluating, communicating, and acting on beliefs and claims. It includes:

  • Seeking appropriate evidence — not relying on convenience or hearsay.
  • Checking source authority — verifying that sources are credible and relevant.
  • Distinguishing fact from inference — being clear about what is known and what is inferred.
  • Acknowledging uncertainty — communicating what is not known.
  • Avoiding fabricated support — not inventing evidence or citations.
  • Correcting mistakes — promptly and transparently correcting errors.
  • Resisting pressure to overstate — maintaining accuracy even when it is unpopular.
  • Documenting verification — keeping records of your verification process.
  • Consulting qualified people — escalating when you lack expertise.
  • Declining to make unsupported claims — refusing to communicate claims that cannot be verified.

Why it matters: As a responsible user of AI, you are accountable for the claims you communicate. You cannot outsource epistemic responsibility to an AI system. You must exercise human judgment at every step.

6.5 Correction, Qualification, Escalation, Withdrawal, or Rejection?

For each material claim, you must determine the appropriate disposition. The table below summarises the options.

Table 3. Disposition Options for Claims
Disposition When to Use Example
Verified as written Evidence directly supports the claim without significant qualification. “The study found a 15% reduction in symptoms.” (Verified against the original study.)
Verified with qualification Evidence supports the claim but with important conditions or limitations. “In a sample of 200 adults, the study found a 15% reduction; further research is needed for generalisation.”
Supported only with qualification The claim is not fully supported but can be qualified. “The evidence suggests a modest effect in specific populations.”
Contradicted by stronger evidence Stronger evidence directly contradicts the claim. “A more rigorous study found no significant effect.”
Unsupported No evidence supports the claim, or the evidence is irrelevant. “No study supports the claim that the intervention reduces anxiety.”
Unverifiable with available evidence Evidence is inaccessible, paywalled, or insufficient. “The original study is paywalled and could not be accessed.”
Requires specialist review Claim is consequential and requires expert judgment. “This health claim requires review by a medical professional.”
Withdraw Claim is fabricated, dangerous, or should not be used. “The citation is fabricated; the claim should be withdrawn.”
Do not use / Reject Claim is unsupported and should not be communicated. “The claim is not supported by evidence and should not be used.”

Escalation conditions:

  • A central source cannot be verified.
  • A citation is fabricated.
  • Evidence conflicts materially.
  • The claim affects rights, safety, health, finances, education, employment, or essential services.
  • Jurisdiction is uncertain.
  • Original evidence is unavailable.
  • Specialist expertise is required.
  • The output has already been distributed and correction is needed.
  • Correction authority is unclear.
  • Pressure exists to communicate before verification.

6.6 Verification Limits

Definition — Verification Limits

Verification limits are the boundaries of what can be verified with the available evidence, resources, and expertise.

Verification may be limited by:

  • Inaccessible source (paywall, offline, removed).
  • Missing data (incomplete records).
  • Proprietary information (trade secrets, confidential data).
  • Conflicting evidence (credible sources disagree).
  • Language barrier (source is in a language you do not read).
  • Unclear provenance (the chain of evidence cannot be reconstructed).
  • Insufficient expertise (the claim is in a specialised field).
  • Changing conditions (the evidence is time-sensitive).
  • Unavailable authoritative source (the claim is novel and no authoritative source exists).

Responsible approach: Document these limits rather than conceal them. When verification is impossible, escalate, qualify, or withdraw the claim.

7. CLAIM Walkthrough: Complete Application

This walkthrough applies all CLAIM elements to a single claim from the opening scenario.

Claim

“A 2021 study published in the Journal of Community Psychology by Dr. Sarah Mitchell and colleagues found that community-based mental health programs reduce depression by 40% (Mitchell et al., 2021).”

CLAIM Element 1: Identify and decompose the claim

CLAIM Element 2: Locate appropriate evidence

CLAIM Element 3: Assess source authority, relevance, currency, and independence

CLAIM Element 4: Interpret claim-evidence alignment

Result (supplied search data): The Journal of Community Psychology exists, but no article by Mitchell et al. appears in volume 2021. The 40% figure appears in a blog post citing a conference presentation, not a peer-reviewed article. The citation is fabricated.

CLAIM Element 5: Modify, correct, qualify, or reject the claim

8. Verification-Process Walkthrough: TRACE to CLAIM

This walkthrough demonstrates the full integration of TRACE and CLAIM for the opening scenario.

  1. Step 1: TRACE — Identify errors and classify.
    • Error 1: Mitchell et al. (2021) — likely fabrication (the article does not exist).
    • Error 2: 40% reduction — likely unsupported claim (no evidence supports this figure).
    • Error 3: “Multiple independent studies” — likely omission (the sources may be dependent or non-existent).
    • Error 4: CMHA statement — likely contextual mismatch (the statement may be taken out of context).
    • Error 5: “Funded immediately” — invalid inference (the conclusion does not follow from the evidence).
  2. Step 2: TRACE — Assess consequence and prioritise.
    • Consequence: High — the claim is about public health funding and intervention effectiveness.
    • Priority: The Mitchell et al. claim is highest priority because it is the central evidence for the initiative.
    • Decision: Verify the Mitchell et al. citation first.
  3. Step 3: CLAIM — Verify the claim.
    • Identify: Mitchell et al. (2021) in Journal of Community Psychology.
    • Locate evidence: Search journal database.
    • Assess: No article found. The citation is fabricated.
    • Align: No evidence supports the claim.
    • Modify: The claim is unsupported and should be rejected.
  4. Step 4: CLAIM — Verify other claims.
    • Johnson (2020), Lee & Park (2022), Smith (2023): Verify each citation. Search databases. If they do not exist or do not support the claim, they are fabricated or misattributed.
    • CMHA statement: Search for the CMHA statement in official documents. Check if it is taken out of context.
    • BC government policy: Search for the 2020 policy document. Verify its recommendations.
  5. Step 5: TRACE — Escalation and stop conditions.
    • Escalation: Since the claim affects public health funding, escalate to a qualified health policy expert.
    • Stop condition: Do not use any claim that is fabricated or unsupported. Stop the funding recommendation until the evidence is verified.
    • Correction: Produce a corrected briefing that accurately reflects the available evidence (or the lack thereof).

9. Worked Example: Complete CLAIM Analysis

Claim: “A 2022 meta-analysis by the Centre for Evidence-Based Medicine found that vitamin D supplementation reduces the risk of respiratory infections by 15% (CEBM, 2022).”

Step 1: Identify and decompose

Step 2: Locate evidence

Step 3: Assess sources

Step 4: Interpret alignment

Step 5: Modify and correct

10. Guided Learning Activities

🧩 Activity 11.4.A: CLAIM Walkthrough
  • Purpose: Apply CLAIM to a claim.
  • Learning outcome: Apply every CLAIM element.
  • Estimated time: 20 minutes.

Claim: “A 2020 study in Nature by Dr. Hiroshi Tanaka proved that a new drug cures Alzheimer’s disease.”

Apply each CLAIM element to this claim. What would you need to verify? What evidence would be required? What is the likely disposition?

Show activity feedback

CLAIM Element 1 — Identify: Atomic claims: (1) Tanaka published a study in Nature in 2020; (2) The study proved a cure for Alzheimer’s. Claim type: Scientific, causal. Consequence: Very high (health, lives).

CLAIM Element 2 — Locate evidence: Search Nature for Tanaka, 2020. Look for the original article and any subsequent replication studies.

CLAIM Element 3 — Assess: Nature is authoritative. But “proved” and “cures” are very strong claims. Check if the study was a small animal study or a human trial. Check for replication.

CLAIM Element 4 — Align: If the study is a small animal study, the claim is overgeneralised. If it is a human trial, check the sample size, effect size, and limitations.

CLAIM Element 5 — Modify: Likely disposition: Supported only with qualification (if a study exists but is preliminary) or Unsupported (if no study exists). Corrected wording: “A 2020 study in Nature reported preliminary findings on a potential Alzheimer’s treatment; however, these findings are early-stage and require further research. The claim of a ‘cure’ is not supported.”

🧩 Activity 11.4.B: TRACE-to-CLAIM Integration
  • Purpose: Integrate TRACE and CLAIM.
  • Learning outcome: Integrate TRACE error analysis with CLAIM verification.
  • Estimated time: 20 minutes.

You have a claim: “A 2021 survey by the Health Foundation found that 80% of Canadians support universal pharmacare.” You apply TRACE and identify a possible fabrication (the Health Foundation may not exist). Describe how you would use CLAIM to verify this claim.

Show activity feedback

TRACE — Error: Possible fabrication (the Health Foundation may not be a real organisation, or the survey may not exist).

TRACE — Consequence: Moderate to high — this claim could influence health policy decisions.

CLAIM — Identify: Atomic claim: “The Health Foundation conducted a survey in 2021; 80% of Canadians support universal pharmacare.”

CLAIM — Locate: Search for “Health Foundation” and “survey 2021 universal pharmacare.” Check government databases, polling firms, and academic sources.

CLAIM — Assess: If the Health Foundation exists, assess its authority. If it is a non-profit, check its funding and methodology. If the survey exists, check the sample size, margin of error, and question wording.

CLAIM — Align: If the survey exists but uses a small sample, the claim is partial support. If the survey does not exist, the claim is unsupported.

CLAIM — Modify: Corrected wording: “A 2021 survey by the Health Foundation reported that 80% of respondents supported universal pharmacare; however, the survey methodology and sample size were not provided.”

🧩 Activity 11.4.C: Claim-Evidence Verification Table
  • Purpose: Create a verification table.
  • Learning outcome: Create a claim-evidence verification table.
  • Estimated time: 25 minutes.

Using the opening scenario claims, create a claim-evidence verification table for at least three atomic claims. Include: Claim ID, Original Claim, Claim Type, Consequence, Evidence Required, Sources, Authority, Relevance, Currency, Independence, Alignment, Assessment, Correction, Uncertainty, Status.

Show activity feedback

Sample table (partial):

Sample claim-evidence verification table
Claim ID Original Claim Claim Type Consequence Evidence Required Sources Authority Relevance Currency Independence Alignment Assessment Correction Uncertainty Status
C1 Mitchell et al. (2021) found 40% reduction Quantitative High Original article Mitchell et al. None (fabricated) N/A N/A N/A No support Fabricated Unsupported No evidence Unsupported
C2 Multiple independent studies confirm Claim about evidence High Check each citation Johnson, Lee, Smith Unknown Unknown Unknown Unknown Unknown Unverified Pending High Unverifiable
C3 CMHA endorsed and made statement Claim of endorsement Moderate CMHA official documents CMHA High High 2022 N/A Check context Unverified Pending Context needed Unverified
🧩 Activity 11.4.D: Epistemic Responsibility Review
  • Purpose: Reflect on epistemic responsibility.
  • Learning outcome: Exercise epistemic responsibility.
  • Estimated time: 15 minutes.

You are asked to present findings from an AI-generated briefing that contains unverified claims. What does epistemic responsibility require you to do? List at least five actions.

Show activity feedback

Epistemic responsibility actions:

  1. Seek appropriate evidence: Verify the claims before presenting them.
  2. Check source authority: Evaluate the credibility of all cited sources.
  3. Acknowledge uncertainty: Communicate what is not known.
  4. Avoid overstatement: Do not present unverified claims as facts.
  5. Correct mistakes: If errors are found, correct them promptly.
  6. Document verification: Keep records of what was verified and how.
  7. Escalate: If you lack expertise, consult qualified individuals.
  8. Decline to make unsupported claims: Refuse to communicate claims that cannot be verified.
🧩 Activity 11.4.E: Correct, Qualify, Escalate, Withdraw, or Reject?
  • Purpose: Determine the appropriate disposition.
  • Learning outcome: Determine the appropriate disposition for each claim.
  • Estimated time: 15 minutes.

For each scenario, choose the appropriate disposition:

  1. A claim is supported by a single small study with no replication.
  2. A claim is supported by multiple independent studies but with some conflicting results.
  3. A claim is based on a citation that does not exist.
  4. A claim is supported by evidence, but the evidence is paywalled and inaccessible.
  5. A claim is about a life-threatening medical condition, and the evidence is contradictory.
Show activity feedback
  1. Supported only with qualification — the claim should be qualified due to the small sample and lack of replication.
  2. Verified with qualification — the claim is supported but with uncertainty about the conflicting results.
  3. Unsupported — reject — the citation is fabricated; the claim should be rejected.
  4. Unverifiable with available evidence — document the gap and escalate if necessary.
  5. Requires specialist review — due to the high consequence and contradictory evidence, escalate to a medical professional.

11. Exercises

Complete the following exercises. For each, show your reasoning and final answer, then check against the feedback.

📝 Exercise 11.4.1: Apply CLAIM
  • Difficulty: Introductory
  • Estimated time: ~12 min
  • Learning outcome: Apply CLAIM to a claim.

Claim: “A 2021 study by the World Health Organization found that air pollution causes 7 million deaths annually.”

Apply CLAIM Element 1 (Identify and decompose) and Element 2 (Locate appropriate evidence). What are the atomic claims? What evidence would be required?

Show answer
  • Atomic claims: (1) WHO published a study in 2021; (2) The study found 7 million deaths from air pollution; (3) Air pollution causes these deaths (causal claim).
  • Evidence required: The original WHO report (primary source), the methodology for estimating deaths, the definition of “air pollution,” the populations studied, and the causal evidence.
  • Source type: Primary (WHO report) or secondary (systematic review).
📝 Exercise 11.4.2: Integrate TRACE and CLAIM
  • Difficulty: Intermediate
  • Estimated time: ~12 min
  • Learning outcome: Integrate TRACE and CLAIM.

You identify a claim that the AI has generated: “A 2019 study proved that meditation cures depression.” Using TRACE, you identify invalid inference (correlation vs. causation) and unsupported claim (no study “proved” a cure). Describe how you would use CLAIM to verify and correct this claim.

Show answer

TRACE errors: Invalid inference (causation from correlation) and unsupported claim (no proof).

CLAIM — Identify: Atomic claim: “A 2019 study proved that meditation cures depression.”

CLAIM — Locate: Search for the 2019 study. Find the original study.

CLAIM — Assess: Check the study design (RCT, correlational). Assess the effect size and population.

CLAIM — Align: If the study is correlational, the claim “proved” and “cures” are unsupported.

CLAIM — Modify: Corrected wording: “A 2019 study found an association between meditation and reduced depression symptoms; however, the study did not establish causation, and the term ‘cure’ is not supported by the evidence.”

📝 Exercise 11.4.3: Complete a Claim-Evidence Table
  • Difficulty: Intermediate
  • Estimated time: ~15 min
  • Learning outcome: Complete a claim-evidence verification table.

Claim: “A 2022 study in The Lancet found that a new vaccine is 95% effective.”

Complete a verification table with the following columns: Claim ID, Original Claim, Claim Type, Consequence, Evidence Required, Sources, Authority, Relevance, Currency, Independence, Alignment, Assessment, Correction, Uncertainty, Status.

Show answer
Claim-evidence verification table for Exercise 11.4.3
Field Entry
Claim ID C1
Original Claim A 2022 study in The Lancet found a new vaccine is 95% effective.
Claim Type Quantitative, efficacy claim
Consequence Very high (public health)
Evidence Required Original Lancet article, trial protocol, peer review
Sources The Lancet, clinical trial registry
Authority High (The Lancet is prestigious)
Relevance High (if the trial is on the relevant population)
Currency 2022 (check for updates)
Independence Check if the trial was industry-funded
Alignment Check if the study reported 95% efficacy
Assessment Verify against the original article
Correction Pending verification
Uncertainty Check sample size, confidence intervals, population
Status Unverified
📝 Exercise 11.4.4: Evaluate Source Authority and Alignment
  • Difficulty: Intermediate
  • Estimated time: ~10 min
  • Learning outcome: Evaluate authority and alignment.

Claim: “A 2020 government report showed that crime rates fell by 10%.”

Source: A 2020 government report from the Ministry of Justice. The report shows a 10% reduction in property crime, but violent crime increased by 5%.

Evaluate the authority of the source and the alignment of the evidence with the claim.

Show answer
  • Authority: High — a government ministry report is authoritative for crime statistics.
  • Alignment: Partial support — the claim says “crime rates” broadly, but the report shows property crime decreased while violent crime increased. The claim is overgeneralised.
  • Correction: “A 2020 government report showed a 10% reduction in property crime; however, violent crime increased by 5% during the same period. Overall crime rates were mixed.”
📝 Exercise 11.4.5: Preserve Conflicting Evidence
  • Difficulty: Intermediate
  • Estimated time: ~10 min
  • Learning outcome: Preserve conflicting evidence.

You have two sources: Source A (a 2021 study) finds that a new drug reduces pain, while Source B (a 2022 study) finds no significant effect. How would you communicate this? Draft a brief statement.

Show answer

Draft statement: “Evidence on the new drug’s effectiveness is mixed. A 2021 study reported a reduction in pain, but a 2022 study found no significant effect. Differences in methodology and sample populations may explain the discrepancy. Further research is needed before any definitive conclusions can be drawn.”

📝 Exercise 11.4.6: Draft Corrected Wording
  • Difficulty: Intermediate
  • Estimated time: ~10 min
  • Learning outcome: Draft corrected wording.

Original claim: “A 2021 study proved that eating chocolate causes weight loss.”

Evidence: A 2021 correlational study found that people who eat chocolate have lower BMI, but the study did not control for other factors (exercise, diet).

Draft corrected wording that accurately reflects the evidence.

Show answer

Corrected wording: “A 2021 correlational study found an association between chocolate consumption and lower BMI; however, this study does not establish causation. Other factors such as exercise and overall diet were not controlled for. The claim that chocolate causes weight loss is not supported by this evidence.”

📝 Exercise 11.4.7: Communicate Remaining Uncertainty
  • Difficulty: Intermediate
  • Estimated time: ~10 min
  • Learning outcome: Communicate remaining uncertainty.

Claim: “A new policy will reduce traffic accidents by 20%.”

Evidence: A pilot study in one city showed a 20% reduction, but the study was short-term and has not been replicated.

Draft a statement that communicates the uncertainty.

Show answer

Draft statement: “A pilot study in one city found a 20% reduction in traffic accidents following the new policy. However, this was a short-term, single-city study that has not been replicated. The effectiveness of the policy on a larger scale remains uncertain, and further evaluation is needed.”

📝 Exercise 11.4.8: Choose Correction, Qualification, Escalation, Withdrawal, or Rejection
  • Difficulty: Advanced
  • Estimated time: ~10 min
  • Learning outcome: Choose the appropriate disposition.

Claim: “A 2023 study in a major journal found that a new drug reduces the risk of heart attack by 50%.” You find that the study was funded by the drug manufacturer and has been retracted due to data fabrication.

What is the appropriate disposition? Explain.

Show answer

Disposition: Withdraw — do not use.

Explanation: The study has been retracted due to data fabrication. The claim is false and should be withdrawn. If the claim has already been communicated, a correction or retraction is needed. This should be escalated to ensure proper communication of the retraction.

12. Homework Problems

Formative · Ungraded · Recommended preparation for Assignment 5 and the Unit 11 cumulative project.

🧠 Problem 11.4.1: CLAIM Element Matching

Match each CLAIM element to its definition:

  1. Identify and decompose the claim
  2. Locate appropriate evidence
  3. Assess source authority, relevance, currency, and independence
  4. Interpret claim-evidence alignment
  5. Modify, correct, qualify, or reject the claim

Definitions: (A) Determine how the evidence relates to the claim; (B) State the exact claim and break it down; (C) Produce corrected wording based on the evidence; (D) Find the right evidence and source types; (E) Evaluate source quality.

Show answer
  1. B — Identify and decompose the claim
  2. D — Locate appropriate evidence
  3. E — Assess source authority, relevance, currency, and independence
  4. A — Interpret claim-evidence alignment
  5. C — Modify, correct, qualify, or reject the claim
🧠 Problem 11.4.2: Complete CLAIM Analysis

Claim: “A 2021 study by the National Research Council found that 60% of adults are sleep-deprived.”

Apply CLAIM to this claim. What are the atomic claims? What evidence would be required? How would you assess source authority and relevance? What alignment issues might arise? Draft corrected wording.

Show solution design
  1. Identify: (1) NRC published a study in 2021; (2) 60% of adults are sleep-deprived.
  2. Locate: Search for NRC study, check for survey methodology, sample size, definition of “sleep-deprived.”
  3. Assess: NRC is authoritative if it is a government research council. Check if the study is peer-reviewed. Check the definition of “adults” and “sleep-deprived.”
  4. Align: If the study used a specific measure (e.g., 6 hours or less), check if the claim matches. If the sample is from one country, generalisation may be an issue.
  5. Modify: “A 2021 study by the National Research Council reported that 60% of surveyed adults met the criteria for sleep deprivation; however, this was based on self-report in a specific population, and the definition of sleep deprivation may differ from clinical definitions.”
🧠 Problem 11.4.3: TRACE-to-CLAIM Integration

You have a claim: “A 2020 study by the Global Institute for Climate Studies found that climate change will cause a 50% increase in extreme weather events by 2050.” Using TRACE, you identify possible omission (the study may be a projection, not a finding) and false precision (50% is too precise). Describe how you would use CLAIM to verify and correct this claim.

Show answer

TRACE errors: Omission (the study may be a projection with uncertainty) and false precision (50% is too precise).

CLAIM — Identify: Atomic claim: “A 2020 study found a 50% increase in extreme weather by 2050.”

CLAIM — Locate: Search for the Global Institute for Climate Studies and the 2020 study.

CLAIM — Assess: Check if the study is a projection or an empirical finding. Check the model and uncertainty range.

CLAIM — Align: If the study is a projection with a range (e.g., 30-70%), the claim should be qualified.

CLAIM — Modify: “A 2020 study by the Global Institute for Climate Studies projected an increase in extreme weather events of 50% by 2050 under a high-emissions scenario, with a range of 30-70%; however, this is a projection, not a confirmed finding, and uncertainty remains significant.”

🧠 Problem 11.4.4: Claim-Evidence Table Problem

Complete a claim-evidence verification table for the claim: “A 2023 meta-analysis found that online learning improves student outcomes by 15% compared to traditional learning.”

Show solution design
Claim-evidence verification table for Problem 11.4.4
Field Entry
Claim ID C1
Original Claim A 2023 meta-analysis found that online learning improves student outcomes by 15% compared to traditional learning.
Claim Type Quantitative, comparative effectiveness
Consequence Moderate-high (education policy)
Evidence Required Original meta-analysis, included studies, effect size calculation
Sources Meta-analysis article, systematic review
Authority Depends on journal and authors
Relevance Check if the meta-analysis covers the same populations and outcomes
Currency 2023 (recent)
Independence Check if the meta-analysis is independent
Alignment Check if the meta-analysis reported a 15% effect size
Assessment Verify against the original meta-analysis
Correction Pending verification
Uncertainty Check heterogeneity, publication bias, study quality
Status Unverified
🧠 Problem 11.4.5: Source Independence Problem

You have three sources that all support the claim that “a new drug reduces cholesterol.” Source A is the original clinical trial. Source B is a news article about Source A. Source C is a blog post about Source B. Are these sources independent? What does this mean for corroboration?

Show answer

Independence: No, they are not independent. Source B depends on Source A, and Source C depends on Source B. They are all part of the same dependency chain.

Corroboration: They do not provide corroboration. They are repetition. To corroborate the claim, you would need independent trials or systematic reviews that include multiple independent studies.

🧠 Problem 11.4.6: Conflict-Communication Problem

You have two systematic reviews on the same topic. Review A finds a positive effect. Review B finds no effect. Draft a statement that communicates this conflict responsibly.

Show answer

Draft statement: “The evidence base for this intervention is conflicting. Systematic Review A (2021) reported a positive effect, while Systematic Review B (2022) found no significant effect. Differences in inclusion criteria, study quality assessment, and outcome measures may explain these discrepancies. A definitive conclusion cannot be drawn from the current evidence, and further high-quality studies are needed.”

🧠 Problem 11.4.7: Corrected-Wording Problem

Original claim: “A 2020 study by the University of Oxford proved that drinking coffee prevents cancer.”

Evidence: A 2020 observational study found an association between coffee consumption and lower cancer risk, but the study was correlational and did not control for all confounding factors.

Draft corrected wording.

Show answer

Corrected wording: “A 2020 observational study by the University of Oxford found an association between coffee consumption and lower cancer risk; however, the study was correlational and does not establish causation. The claim that coffee prevents cancer is not supported by this evidence.”

🧠 Problem 11.4.8: Epistemic Responsibility Problem

Your supervisor asks you to include a claim in a report that you have not been able to verify. You are under pressure to meet a deadline. What does epistemic responsibility require you to do?

Show answer

Epistemic responsibility actions:

  1. Communicate the verification status: Tell your supervisor that the claim is unverified.
  2. Document the gap: Note in the report that the claim could not be verified.
  3. Qualify or exclude: Either exclude the claim or add a clear qualification.
  4. Escalate: If the claim is consequential, escalate to a subject-matter expert.
  5. Resist pressure: Do not include unsupported claims just to meet a deadline.
  6. Document your efforts: Keep records of your verification attempts.
🧠 Problem 11.4.9: Escalation or Withdrawal Problem

Claim: “A 2022 study found that a new drug cures Alzheimer’s disease.” You search for the study and find that it was retracted due to data fabrication. What should you do? Explain your reasoning.

Show answer

Action: Withdraw — do not use.

Reasoning: The study has been retracted due to data fabrication. The claim is false. It should be withdrawn immediately. If the claim has been communicated, a correction or retraction is required. Escalate to ensure proper handling of the retraction.

🧠 Problem 11.4.10: Assignment 5 Planning Problem

You are preparing to complete Assignment 5. Outline the steps you would take to verify an AI-generated passage, from receiving the passage to submitting the final product.

Show solution design
  1. Preserve: Save the original AI-generated passage and the prompt.
  2. Identify purpose and audience: Understand the context.
  3. Build an atomic claim inventory: Decompose every claim.
  4. Apply TRACE: Identify and classify errors, assess consequence, prioritise.
  5. Identify evidence required: For each claim, determine what evidence is needed.
  6. Locate candidate sources: Search independently for evidence.
  7. Verify source metadata: Confirm citations exist.
  8. Assess authority, relevance, currency, and independence: Evaluate each source.
  9. Align each source with the claim: Determine alignment.
  10. Trace quotations, statistics, and repeated claims: Find original sources.
  11. Identify citation laundering: Check for laundering.
  12. Preserve conflicting evidence: Document disagreements.
  13. Correct or qualify the passage: Draft corrected wording.
  14. State remaining uncertainty: Be transparent.
  15. Apply CLAIM: Complete the full CLAIM process.
  16. Document AI assistance: Disclose any AI use.
  17. Conduct final human review: Review your work.
🧠 Problem 11.4.11: Transfer Problem

Choose a claim from your field or profession that you have recently encountered. Apply the complete CLAIM framework to it. What is the claim? What evidence would be required? What sources would you use? What is the likely disposition? Draft corrected wording.

Show solution design

Solution design:

  1. State the claim clearly.
  2. Decompose the claim into atomic claims.
  3. Identify the type of evidence required.
  4. Identify the source types that would be authoritative.
  5. Apply CLAIM to each atomic claim.
  6. Determine the disposition (verified, qualified, etc.).
  7. Draft corrected wording that reflects the evidence.
  8. Document remaining uncertainty.

13. Mini-Project: Complete CLAIM Verification and Correction Review

📁 Project: Complete CLAIM Verification and Correction Review
  • Purpose: Integrate all Unit 11 learning into a complete verification and correction process for a synthetic AI-generated passage.
  • Learning outcomes: Apply TRACE and CLAIM to a passage; create a claim inventory; evaluate sources; trace citations; produce corrected wording; document uncertainty; establish escalation and stop conditions.
  • Estimated effort: 3–5 hours.

Supplied passage

📌 Passage — Benefits of AI in Healthcare

Artificial intelligence (AI) is revolutionising healthcare. A 2022 study by the Institute for Health Informatics found that AI-powered diagnostic tools reduce diagnostic errors by 30% (IHI, 2022). This finding has been replicated in multiple independent studies (Garcia et al., 2021; Wong & Patel, 2020; Lee, 2021). The World Health Organization (WHO) has endorsed AI in healthcare, stating that “AI has the potential to transform global health care delivery” (WHO, 2021). Furthermore, a 2020 report from the Government of Canada recommended the widespread adoption of AI in Canadian hospitals. Given this evidence, AI should be deployed immediately across all healthcare settings to improve patient outcomes and reduce costs.

Required deliverable — 44 sections

Complete the following sections for the supplied passage. Use the supplied source data or your own independent search.

  1. Passage and human purpose: What is the purpose of this passage?
  2. Intended audience: Who is the audience?
  3. Intended use: How would this passage be used?
  4. Original prompt and context, if available: (Synthetic — not available).
  5. TRACE summary: Identify and classify errors, assess consequence, prioritise.
  6. Atomic claim inventory: List every atomic claim.
  7. Claim type: Classify each claim.
  8. Consequence: Assess the consequence of each claim.
  9. Verification priority: Rank claims by priority.
  10. Evidence required: For each claim, state the evidence needed.
  11. Search and discovery strategy: How would you find the evidence?
  12. Candidate sources: List the sources you would check.
  13. Primary, secondary, and tertiary sources: Classify the sources you find.
  14. Source authority: Evaluate the authority of each source.
  15. Source expertise: Evaluate the expertise.
  16. Source relevance: Evaluate relevance.
  17. Source currency: Evaluate currency.
  18. Jurisdiction and scope: Evaluate jurisdiction and scope.
  19. Source provenance: Trace the origin of the claims.
  20. Source independence: Assess independence of sources.
  21. Claim-evidence alignment: Align each claim with its evidence.
  22. Corroboration: Identify corroboration or repetition.
  23. Triangulation: Consider triangulation.
  24. Conflicting evidence: Identify any conflicts.
  25. Metadata verification: Verify citations.
  26. Original-source tracing: Trace to original sources.
  27. Quotation verification: Verify the WHO quotation.
  28. Quantitative-claim verification: Verify the 30% statistic.
  29. Citation-laundering analysis: Detect any laundering.
  30. Evidence gaps: Identify gaps.
  31. Corrected wording: Rewrite the passage.
  32. Qualifications: State all qualifications.
  33. Remaining uncertainty: State what remains uncertain.
  34. Specialist-review need: Does specialist review apply?
  35. Correction or withdrawal process: What correction process is needed?
  36. Escalation: Should this be escalated?
  37. Stop conditions: What are the stop conditions?
  38. Epistemic-responsibility statement: Acknowledge responsibility.
  39. Complete CLAIM analysis: Apply CLAIM to the entire passage.
  40. Integration with TRACE: Show how TRACE informed CLAIM.
  41. Final disposition: For each claim, state the disposition.
  42. Assignment 5 connection: How does this prepare you for Assignment 5?
  43. Assignment 1 reflection connection: Reflect on your learning.
  44. Preparation for Unit 12: How does this prepare you for Unit 12?

Supplied source data (for verification):

  • Institute for Health Informatics (IHI, 2022): The IHI is a real organisation, but no 2022 study on AI diagnostic errors was found. A 2021 report from a different organisation found a 15% reduction in a specific diagnostic area.
  • Garcia et al. (2021): A study exists, but it is a small pilot study (n=50) with no control group.
  • Wong & Patel (2020): No record found — likely fabricated.
  • Lee (2021): A study exists, but it is a review article that summarises other studies, not an original study.
  • WHO (2021): The WHO statement is accurate, but it is taken from a broader document with many qualifications.
  • Government of Canada (2020): A report exists, but it was a pilot project recommendation in a specific province, not a nationwide recommendation.

Assessment criteria

  • All sections completed.
  • TRACE and CLAIM accurately applied.
  • Errors classified with justification.
  • Corrections are evidence-based.
  • Uncertainty explicitly communicated.
  • Disposition well justified.
Show staged hints

Hint 1: Start with TRACE. Use TRACE to identify possible errors. Focus on fabrication (Wong & Patel), overstatement (30% may be exaggerated), and contextual mismatch (the government report may be provincial, not national).

Hint 2: Build the claim inventory. Break the passage into atomic claims. You should have at least 6-8 claims. Include the recommendation as a claim that requires justification.

Hint 3: Apply CLAIM systematically. For each atomic claim, work through CLAIM: Identify → Locate → Assess → Align → Modify. Use the claim-evidence verification table to document your findings.

Show solution architecture

Approach: Follow the TRACE-to-CLAIM integration process. Produce a structured document with headings for each section. Use the claim-evidence verification table for each claim. Conclude with a corrected version of the passage and a clear statement of remaining uncertainty.

Show illustrative partial example (different passage)

See the worked example in Section 9 for a complete CLAIM analysis. Apply the same process to this passage.

Sample corrected wording (for the opening passage): “Some evidence suggests that AI-powered diagnostic tools may reduce diagnostic errors in certain contexts. A 2021 report found a 15% reduction in a specific diagnostic area, but the frequently cited 30% figure is not supported by peer-reviewed evidence. The WHO has noted the potential of AI but also emphasised the need for careful implementation and oversight. The Government of Canada report was a provincial pilot recommendation, not a nationwide mandate. Given the uncertainty and the absence of robust peer-reviewed evidence, AI should not be deployed immediately across all healthcare settings. Further research and specialist review are required.”

14. Self-Check Quiz

9 questions · Formative · Ungraded. Expand each item to see feedback.

Question 1: What is the first step in the CLAIM framework?

Select one:
Show answer and explanation

Answer: b.

You must first understand the exact claim before you can find evidence or assess alignment.

Question 2: Which of the following is not an element of CLAIM?

Select one:
Show answer and explanation

Answer: d.

CLAIM is about verifying and correcting existing claims, not generating new ones.

Question 3: What is epistemic responsibility?

Select one:
Show answer and explanation

Answer: b.

Epistemic responsibility is a human practice that cannot be outsourced to AI.

Question 4: A claim that is supported by evidence but with important limitations should be:

Select one:
Show answer and explanation

Answer: b.

When evidence supports a claim but with important limitations, you must state those qualifications in the corrected wording.

Question 5: How does TRACE inform CLAIM?

Select one:
Show answer and explanation

Answer: b.

TRACE helps you identify what might be wrong; CLAIM helps you verify and correct the claim.

Question 6: What is a claim-evidence verification table?

Select one:
Show answer and explanation

Answer: b.

The table documents claim ID, original claim, evidence, sources, alignment, correction, uncertainty, and status.

Question 7: If a citation is fabricated, the appropriate disposition is:

Select one:
Show answer and explanation

Answer: c.

A fabricated citation means the claim is unsupported and should be withdrawn.

Question 8: What does it mean to preserve disagreement when evaluating evidence?

Select one:
Show answer and explanation

Answer: c.

Responsible verification acknowledges and explains conflicting evidence.

Question 9: Which of the following is an example of an escalation condition?

Select one:
Show answer and explanation

Answer: c.

When a claim has high consequence and evidence is conflicting, you should escalate to a subject-matter expert.

15. Assignment Connections

📝 Assignment 5: Verification and Fact-Checking Assignment (10% of final grade)

This lesson has provided the complete methodology for Assignment 5. You will now be able to:

  • Decompose an AI-generated passage into atomic claims.
  • Apply TRACE to identify errors and prioritise verification.
  • Use CLAIM to verify each claim.
  • Create a claim-evidence verification table.
  • Trace citations, quotations, and statistics to original sources.
  • Detect citation laundering.
  • Produce corrected wording with qualifications.
  • Communicate remaining uncertainty.
  • Exercise epistemic responsibility.
  • Document your verification process.

Recommended assignment workflow (from this lesson):

  1. Preserve the original AI-generated passage and prompt.
  2. Identify purpose and audience.
  3. Build an atomic claim inventory.
  4. Apply TRACE.
  5. Prioritise claims.
  6. Identify evidence required for each claim.
  7. Locate candidate sources independently.
  8. Prefer original and authoritative sources.
  9. Verify source metadata.
  10. Assess authority, relevance, currency, and independence.
  11. Align each source with the exact claim.
  12. Trace quotations, statistics, and repeated claims.
  13. Identify citation laundering.
  14. Preserve conflicting evidence.
  15. Correct or qualify the passage.
  16. State remaining uncertainty.
  17. Apply CLAIM.
  18. Document AI assistance.
  19. Conduct final human review.

Your work in the mini-project (Section 13) directly prepares you for this assignment.

📓 Assignment 1 journal prompts

Reflect on the following questions in your Assignment 1 journal.

  1. How does verification differ from asking an AI system to check itself? What did you learn about why self-checking is insufficient?
  2. How does source authority depend on the claim? Provide an example.
  3. Why does source existence not prove claim support? What else must you verify?
  4. How did claim-evidence alignment change your evaluation of a source? Provide a specific example.
  5. Why are repeated sources not independent? Provide an example.
  6. What did lateral reading reveal about a source you previously trusted?
  7. How can a citation acquire false authority? Describe a citation-laundering example.
  8. How should conflicting evidence be communicated? Why is it important to preserve disagreement?
  9. What does epistemic responsibility require of you when you encounter unverified claims?
  10. How has the CLAIM framework changed your verification process?
  11. How has your trust in sources and your evaluation practices changed after completing Unit 11?

Write in your own voice; there are no model responses.

Show reflection planning checklist
  • I have addressed at least five prompts in depth.
  • I have used specific examples from the lessons or my own experience.
  • I have connected my reflection to Assignment 5 preparation.
  • I have identified changes in my own verification practice.
  • I have used clear, specific language.

16. Key Takeaways

Remember these points
  • CLAIM is a structured framework for verification: Identify → Locate evidence → Assess sources → Align evidence → Modify/Correct.
  • TRACE and CLAIM are complementary: TRACE identifies errors; CLAIM verifies and corrects.
  • A claim-evidence verification table documents every step of verification and is essential for Assignment 5.
  • Epistemic responsibility is a human practice — you cannot outsource it to AI.
  • For each claim, you must determine the appropriate disposition: verified, qualified, contradicted, unsupported, unverifiable, escalate, withdraw, or reject.
  • Qualification is not a failure — it is an honest reflection of the evidence.
  • Uncertainty must be communicated transparently.
  • Escalation conditions protect against harm from unverified claims.
  • Verification limits are real — document them rather than conceal them.
  • Corrected wording must reflect available evidence, preserve qualifications, and identify remaining uncertainty.

17. Key Terms

CLAIM framework
A structured process for verifying claims: Identify, Locate, Assess, Interpret, Modify.
TRACE-to-CLAIM integration
The use of TRACE error analysis to inform and guide the CLAIM verification process.
Claim-evidence verification table
A structured document that records the verification process for each atomic claim.
Epistemic responsibility
Responsible practices for forming, evaluating, communicating, and acting on beliefs and claims.
Disposition
The final decision about a claim: verified, qualified, contradicted, unsupported, unverifiable, escalate, withdraw, or reject.
Qualification
A statement of conditions or limitations that must accompany a claim to accurately reflect the evidence.
Escalation
Raising a verification issue to a subject-matter expert, supervisor, or institutional authority.
Stop condition
A trigger for stopping the use or communication of a claim (e.g., fabrication, high consequence).
Verification limit
A boundary on what can be verified with the available evidence, resources, or expertise.
Corrected wording
Revised phrasing of a claim that accurately reflects the available evidence and preserves qualifications.
Consequence assessment
Evaluating the potential impact of a claim if it is false or misleading.
Human judgment
The irreducible human role in verification, correction, and epistemic responsibility.

18. Looking Ahead: Unit 12

Congratulations — you have completed the Unit 11 verification methodology! You have learned to:

In Unit 12: Generative AI for Learning, you will build on these verification skills to examine:

The verification skills you have developed will be essential for evaluating AI-generated learning content and for using AI responsibly in your own learning.

Before moving on, complete the Unit 11 Consolidation Exercise and the Cumulative Project to solidify your learning. Then, complete the Assignment 5 Completion Guide to prepare for your final assignment submission.

Continue to Unit 11 Consolidation Return to Unit 11 Home

19. Sources & Further Reading

[VERIFICATION REQUIRED: External links and institutional resources should be reviewed for currency and accessibility.]

⚠️ Note

Additional authoritative sources may be provided by your instructor. If you are unable to locate a source, use the supplied-source alternatives in the activities.

👤