Table of Contents
- 1. Assignment Overview
- 2. Recommended Workflow
- 3. Submission Structure
- 4. Templates
- 4.1 Claim-Evidence Verification Table
- 4.2 Source-Evaluation Template
- 4.3 Citation-Tracing Template
- 4.4 Correction-Record Template
- 5. Checklists
- 5.1 Evidence Checklist
- 5.2 Source-Independence Checklist
- 5.3 Academic-Integrity Checklist
- 5.4 Self-Review Checklist
- 6. Illustrative Examples
- 7. Frequently Asked Questions
- 8. Resources & Further Reading
1. Assignment Overview
- Weight: 10% of final grade
- Submission point: End of Module 4
- Format: Structured submission (see Section 3)
Purpose
Assignment 5 assesses your ability to independently verify claims in an AI-generated passage using the TRACE and CLAIM frameworks. You will demonstrate your capacity to:
- Decompose a passage into atomic claims.
- Identify and classify errors (TRACE).
- Locate and evaluate authoritative, relevant, and current sources.
- Align evidence with claims.
- Detect citation laundering and verify citations.
- Produce corrected wording with qualifications.
- Communicate remaining uncertainty.
- Exercise epistemic responsibility.
What you will submit
- Original AI-generated passage and prompt.
- Claim inventory.
- TRACE error analysis.
- Verification-priority ranking.
- Evidence requirements.
- Search and discovery record.
- Source inventory with evaluations.
- Claim-evidence verification tables.
- Corrected wording and uncertainty statement.
- CLAIM analysis.
- AI-use disclosure.
- Final human assessment.
2. Recommended Workflow
Follow these steps in order. Each step builds on the previous one.
-
Step 1: Preserve the original. Save the exact AI-generated passage and the original
prompt you used. If you are using a provided passage, save it as-is. Do not edit the original text.
Why: Your submission must include the original passage for comparison with your corrected version. - Step 2: Identify purpose and audience. What is the purpose of the passage? Who is the intended audience? How might the passage be used? This context informs your verification priorities.
- Step 3: Build an atomic claim inventory. Read the passage carefully and decompose every claim into atomic propositions. Label each claim with a unique ID (C1, C2, etc.). Include implicit claims.
- Step 4: Apply TRACE. For each claim, identify possible errors (hallucination, fabrication, factual error, unsupported claim, omission, contextual mismatch, invalid inference, false precision). Assess the consequence of each error. Prioritise claims for verification.
- Step 5: Identify evidence required. For each claim, state what kind of evidence would be needed to support or refute it.
- Step 6: Locate candidate sources independently. Search for sources using library databases, public search engines, or open-access resources. Do not rely on the AI’s citations.
- Step 7: Verify source metadata. For every cited source, verify author, title, journal, year, volume, pages, DOI, and other metadata.
- Step 8: Assess authority, relevance, currency, and independence. For each source, evaluate its authority relative to the claim, its relevance, its currency, and its independence.
- Step 9: Align evidence with claims. For each atomic claim, determine whether the evidence directly supports, partially supports, supports with qualification, contradicts, or does not address the claim.
- Step 10: Trace quotations, statistics, and repeated claims. Trace back to the original source. Verify the exact wording, measure, population, and context.
- Step 11: Identify citation laundering. Examine the citation chain for each claim. Are there multiple sources that all depend on the same original? Is the original credible?
- Step 12: Preserve conflicting evidence. If credible sources disagree, document the disagreement and explain the differences.
- Step 13: Correct or qualify the passage. Draft a corrected version of the passage that accurately reflects the evidence.
- Step 14: State remaining uncertainty. Identify what remains uncertain after verification.
- Step 15: Apply CLAIM. For each claim, document the complete CLAIM analysis: Identify → Locate → Assess → Interpret → Modify.
- Step 16: Document AI assistance. If you used AI for discovery, drafting, or any other aspect, disclose this clearly.
- Step 17: Conduct final human review. Read through your entire submission. Check for completeness, accuracy, and clarity.
3. Submission Structure
Organise your submission as follows. Use the templates in Section 4 to format your data.
- Original AI-generated passage. Include the exact text of the passage you are verifying. If the passage was generated using a prompt, include the prompt as well.
- Human purpose and intended audience. Briefly explain the purpose of the passage and the intended audience.
- Atomic claim inventory. List every atomic claim with a unique ID (e.g., C1, C2, C3). Include implicit claims.
- TRACE error analysis. For each claim, identify possible errors. Assess consequence. Prioritise claims for verification.
- Verification-priority ranking. List claims in order of verification priority, with the most consequential claims first.
- Evidence requirements. For each claim, state the type of evidence that would be needed.
- Search and discovery record. Document your search strategy: databases used, search terms, and key results.
- Source inventory. List every source you considered, with full bibliographic information.
- Source-type classification. For each source, classify it as primary, secondary, or tertiary relative to the claim.
- Source-authority evaluation. For each source, evaluate its authority relative to the claim.
- Source-relevance evaluation. For each source, evaluate its relevance.
- Source-currency evaluation. For each source, evaluate whether it is sufficiently current for the claim.
- Source-independence analysis. For each claim, assess whether the sources are independent or share a common origin.
- Claim-evidence alignment. For each atomic claim, determine alignment: direct support, partial support, support with qualification, contradiction, or not addressing.
- Corroboration and triangulation. Summarise the evidence for corroboration or triangulation. Identify any independent sources that converge on the same finding.
- Citation and metadata verification. For each citation, document your verification of author, title, journal, year, volume, pages, DOI, and other metadata.
- Original-source tracing. For any quotation, statistic, or repeated claim, document the chain back to the original source.
- Corrected wording. Provide a fully corrected and qualified version of the passage.
- Uncertainty statement. State what remains uncertain after verification.
- CLAIM analysis. For each claim, document the complete CLAIM analysis.
- AI-use disclosure. State which AI systems you used and how you used them.
- Final human assessment. Provide your overall assessment of the passage and the verification process.
4. Templates
Use these templates to structure your data. Copy and adapt them for your submission.
4.1 Claim-Evidence Verification Table
Use this table to document the verification of each atomic claim.
| Claim ID | Original Claim | Claim Type | Consequence | Evidence Required | Source(s) | Authority | Relevance | Currency | Independence | Alignment | Assessment | Correction | Uncertainty | Status |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| C1 | [State the atomic claim] | [e.g., Quantitative, Causal, Policy] | [High/Medium/Low] | [What evidence is needed?] | [List source IDs] | [High/Medium/Low] | [High/Medium/Low] | [High/Medium/Low] | [Independent/Dependent] | [Direct/Partial/Qualified/Contradicts/Not Address] | [Verified/Unverified/Fabricated/etc.] | [Corrected wording] | [What remains uncertain?] | [Status] |
| C2 | [State the atomic claim] | [Type] | [Consequence] | [Evidence needed] | [Source IDs] | [Authority] | [Relevance] | [Currency] | [Independence] | [Alignment] | [Assessment] | [Correction] | [Uncertainty] | [Status] |
4.2 Source-Evaluation Template
Use this table to evaluate each source you considered.
| Source ID | Citation | Source Type | Authority | Relevance | Currency | Provenance | Limitations | Conflict of Interest? |
|---|---|---|---|---|---|---|---|---|
| S1 | [Full citation] | [Primary/Secondary/Tertiary] | [Evaluation] | [Evaluation] | [Evaluation] | [Trace origin] | [Limitations] | [Yes/No/Unknown] |
| S2 | [Full citation] | [Type] | [Evaluation] | [Evaluation] | [Evaluation] | [Trace origin] | [Limitations] | [Yes/No/Unknown] |
4.3 Citation-Tracing Template
Use this table to document the chain of citations for each claim.
| Claim | Original Source | Intermediate Sources | Transformations | Final Assessment |
|---|---|---|---|---|
| [Claim text] | [Original source] | [List intermediate sources] | [Changes detected] | [Verified / Laundered / Fabricated / Misattributed] |
| [Claim text] | [Original source] | [List intermediate sources] | [Changes detected] | [Assessment] |
4.4 Correction-Record Template
Use this table to document your corrections for each claim.
| Claim ID | Original Wording | Issue Identified | Corrected Wording | Justification |
|---|---|---|---|---|
| C1 | [Original text] | [Error / overstatement / misattribution] | [Corrected text] | [Why this correction?] |
| C2 | [Original text] | [Issue] | [Corrected text] | [Justification] |
5. Checklists
5.1 Evidence Checklist
- I have located evidence for every material claim in the passage.
- I have used primary or authoritative secondary sources for consequential claims.
- I have verified all quantitative claims against their original sources.
- I have traced all quotations to their original context.
- I have documented evidence gaps where no suitable source was found.
- I have not relied solely on AI-generated summaries as evidence.
- I have distinguished between evidence that supports, qualifies, contradicts, or does not address each claim.
5.2 Source-Independence Checklist
- I have assessed whether sources are independent or dependent for each claim.
- I have identified any common-source dependencies (e.g., all sources cite the same original study).
- I have distinguished corroboration from repetition.
- I have identified any citation laundering or circular reporting.
- I have not counted multiple derivative sources as independent corroboration.
5.3 Academic-Integrity Checklist
- I have documented all AI assistance used in this assignment.
- I have independently verified every claim and cited source.
- I have not fabricated any citations or evidence.
- I have disclosed all sources used.
- I have not submitted work that was completed by another person or AI.
- I have complied with the institution’s academic-integrity policies.
5.4 Self-Review Checklist
- I have completed the full assignment workflow (Section 2).
- I have included all required submission sections (Section 3).
- I have verified every claim and cited source.
- I have produced corrected wording that accurately reflects the evidence.
- I have communicated remaining uncertainty and qualifications.
- I have documented my verification process using the templates.
- I have disclosed AI assistance.
- I have conducted a final human review of my submission.
- My submission is well-organised, clear, and professionally presented.
6. Illustrative Examples
The following examples show how to apply the workflow and templates to sample claims. These are illustrative only; your assignment will be based on a different passage.
Example 1: Claim-Evidence Verification Table
Claim: “A 2020 study by the University of Toronto found that green spaces reduce stress by 40%.”
| Claim ID | Original Claim | Claim Type | Consequence | Evidence Required | Source(s) | Authority | Relevance | Currency | Independence | Alignment | Assessment | Correction | Uncertainty | Status |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| C1 | A 2020 study by the University of Toronto found that green spaces reduce stress by 40%. | Quantitative, Causal | High (public health) | Original study, peer review, replication | University of Toronto, 2020 | High (if peer-reviewed) | High (if on green spaces and stress) | 2020 (needs updating) | Single source | Check exact wording | Unverified | Pending verification | Study may have been small or specific | Unverifiable |
Example 2: Source Evaluation
Source: “Johnson, A. (2020). Green spaces and mental health. Journal of Urban Health, 35(2), 112-130.”
| Source ID | Citation | Source Type | Authority | Relevance | Currency | Provenance | Limitations | Conflict of Interest? |
|---|---|---|---|---|---|---|---|---|
| S1 | Johnson, A. (2020). Green spaces and mental health. Journal of Urban Health, 35(2), 112-130. | Primary | High — peer-reviewed journal | High — directly on green spaces and mental health | 2020 — needs updating but relevant | Original research article | Sample may be specific; check population | Unknown |
Example 3: Citation Tracing
Claim: “A 2021 meta-analysis by the WHO found that green spaces reduce depression.”
| Claim | Original Source | Intermediate Sources | Transformations | Final Assessment |
|---|---|---|---|---|
| WHO meta-analysis found green spaces reduce depression | WHO (2021) Global Report on Urban Health | News article → Blog post → AI output | The WHO report mentioned green spaces as one factor; the claim “meta-analysis” and “reduce depression” are not in the report | Misattributed — reject |
Example 4: Corrected Wording
Original passage: “The WHO has proven that green spaces cure depression.”
Corrected passage: “A 2021 WHO report identified green spaces as one of several factors that may contribute to mental health; however, the report does not claim that green spaces cure depression, and causation has not been established.”
Justification: The original claim overstates the evidence and misrepresents the source. The corrected version accurately reflects the WHO report and includes necessary qualifications.
7. Frequently Asked Questions
Q: How long should my submission be?
There is no strict word count, but a thorough submission typically ranges from 2,000 to 3,500 words (or equivalent structured data). Focus on completeness and accuracy rather than length.
Q: Can I use AI to help with this assignment?
You may use AI for discovery support (e.g., suggesting search terms) or drafting assistance, but you must independently verify every claim. You must disclose any AI use in your submission.
Q: What if I cannot access a source because it is paywalled?
Document the access issue and mark the claim as Unverifiable with available evidence. Use the supplied-source alternative if available, or escalate to a librarian.
Q: What if I find that a claim is fabricated?
Document the fabrication clearly. Mark the claim as Unsupported — reject. Do not use the claim in your corrected passage. Escalate if the claim is consequential.
Q: Do I need to verify every single claim?
Yes, every material claim should be addressed. Use the claim inventory to ensure you have not missed any. For low-consequence claims, you may note that they are low priority, but you should still document your verification.
Q: What citation style should I use?
[INSERT REQUIRED CITATION STYLE FROM COURSE MATERIALS]
Q: Where can I find help?
Contact your course coordinator, librarian, or writing centre for assistance with research, writing, or citation questions.
8. Resources & Further Reading
- Primary textbook: Chapter 8: Verification, Evidence, and Epistemic Responsibility (COMP 3270).
- Appendix C: Verification Toolkit (COMP 3270).
- Appendix D: Responsible-Use Toolkit (COMP 3270).
- Caulfield, M. (2017). Web Literacy for Student Fact-Checkers.
- Stanford History Education Group. (2016). “Evaluating Information: The Cornerstone of Civic Online Reasoning.”
- University of Toronto Libraries. (2024). “Evidence-Based Practice.”
- American Library Association. (2023). “Information Literacy Competency Standards.”