AI disclosure

How AI is used to produce reports, and what that means for accuracy.

Where AI is used

  • Spam and low-relevance filtering.
  • Theme grouping across comments.
  • Scoring each comment for insight quality against fixed published factors.
  • Drafting the narrative sections of the report.

Guardrails

  • Every finding must cite the specific comments it came from. Findings that cannot be traced to supplied evidence are removed before the report is shown.
  • Inference is labelled as inference and separated from evidence-backed statements.
  • When the evidence is too thin, the report says so rather than producing a confident answer.

Limitations

  • AI can misread sarcasm, cultural context, and domain jargon.
  • Comment volume is not representativeness: a loud minority can dominate a comment section.
  • For medical, legal, or financial questions, verify anything material with a qualified professional.

Last updated 2 April 2026.