Claims & Evidence Matrix
Connects material propositions to supporting, contradictory and limiting evidence.
Open matrix →Published MH370 case material
Claims and evidence audit example: MH370
If an AI-generated answer, workflow or conclusion will influence a real decision, this example shows the kind of evidence questions that should be tested before reliance.
The documents are here
The MH370 research package includes the underlying public case material used to demonstrate the evidence-assurance process. It is not a future placeholder.
Claims & Evidence Audit
Tests prominent claims, evidentiary support, independent corroboration, contradiction and overstatement.
Open audit →Alternative explanations
Shows how alternative explanations and challenge testing are retained rather than removed because they are inconvenient.
View publication-controlled MH370 analysis →Why this matters for AI-generated research
The same control problem arises when a client receives AI-assisted research: citations may be fabricated, repeated sources may appear independent, unsupported inference may be presented confidently, and omissions may disappear behind polished prose.
The MH370 case itself is not represented as an AI-generated research audit. It is a real public-source evidence-assurance record that demonstrates the underlying controls Aperture can also apply when auditing AI-assisted research: claim tracing, source verification, independent-corroboration testing, contradiction analysis, evidence-gap recording and human review.
See how Aperture reviews AI-assisted researchWhat this solves for a client
- You can see which claims are actually supported by evidence.
- You can identify when ten citations are really one originating source repeated ten times.
- You can keep contradictory evidence visible before a decision is made.
- You can distinguish a persuasive narrative from a defensible evidentiary conclusion.
- You can see what remains unknown and what additional evidence could change the answer.