Skill This
AI Skill Report Card

Distinguishing Truth Claims

B+78·Oct 2, 2026·Source: Web
Quick Start12 / 15

When presented with a claim, ask: "Does this describe reality as it actually is, independent of anyone's belief about it?"

  • "The Earth has one moon" → Truth claim (verifiable fact about reality)
  • "I think the Earth has one moon" → Belief claim (about mental state)
  • "Many people believe the Earth is flat" → Fact about beliefs, not a truth claim about Earth's shape
Recommendation▾
Add a slightly more complex/ambiguous example (e.g., a contested scientific claim or a mixed fact-opinion statement) to show edge-case handling
Workflow14 / 15

Progress:

  • Identify the claim being made
  • Classify its type: Truth / Belief / Opinion / Assertion / Interpretation
  • Check if it's falsifiable or verifiable against reality
  • Separate the claim from the claimant's certainty or motive
  • State the epistemic status explicitly

Step 1: Extract the core claim Strip away hedging language, rhetoric, and emotional framing to find the bare assertion.

Step 2: Classify the claim type

  • Truth: Describes facts/reality as they actually exist, independent of observer (intangible concept, not itself a physical thing)
  • Belief: A mental state someone holds about what is true
  • Opinion: A subjective preference or judgment, often not fact-apt
  • Assertion: Something stated, regardless of its truth value
  • Interpretation: A reading or framing of facts, which may vary while facts stay fixed

Step 3: Test against reality Ask whether the claim could in principle be checked against how things actually are. If it can't be checked even in principle, it's likely opinion or metaphysical commitment, not a truth claim.

Step 4: Report status clearly State plainly: "This is presented as fact," "This is a belief," "This is an opinion," etc. Avoid blending categories.

Recommendation▾
Include an example of a bad/incorrect classification to illustrate common errors more concretely
Examples16 / 20

Example 1: Input: "Vaccines cause autism." Output: This is a truth claim (makes an assertion about causal reality) that has been empirically tested and found false. Distinguish from "Some people believe vaccines cause autism" (a true fact about belief prevalence, unrelated to the causal claim's truth value).

Example 2: Input: "Pizza is the best food." Output: This is an opinion, not a truth claim — "best" is a subjective value judgment with no mind-independent fact of the matter.

Example 3: Input: "She said the meeting is at 3pm." Output: This is an assertion. Its truth status is separate from the fact that it was stated — the meeting may or may not actually be at 3pm regardless of her saying so.

Recommendation▾
Consider a brief template or output format for reporting epistemic status across multiple claims in a passage, since real-world text often contains several intermingled claim types
Best Practices
  • Keep "truth" (the state of matching reality) distinct from "truth claims" (statements asserting that match) and from "belief in truth" (psychological state)
  • Remember truth itself is an intangible concept — don't demand it be "pointed to" physically; instead point to evidence that bears on it
  • When uncertain, say "this appears true based on available evidence" rather than "this is true" — most real-world claims are probabilistic, not certain
  • Distinguish facts (verifiable) from values (normative, not verifiable the same way)
Common Pitfalls
  • Treating widespread belief as evidence of truth ("everyone knows X" ≠ X is true)
  • Treating sincerity or confidence of the speaker as evidence of truth
  • Collapsing "true for me" into the same category as objective truth — personal relevance or meaning is not the same as fact
  • Assuming unfalsifiable claims are automatically false — they're simply outside the domain of empirical truth-testing
0
Grade B+AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
12/15
Workflow
14/15
Examples
16/20
Completeness
16/20
Format
15/15
Conciseness
13/15
Save to My Skills