Source Verification and Literacy

AI Summaries Are Starting Points, Not Primary Sources

The most important discipline in AI-assisted market research is treating every AI-generated summary as a draft that requires verification against primary sources before it influences any research conclusion or investment-related decision. AI condenses text conceptually — it extracts patterns from what it is given. If the source text contains an error, the summary inherits that error with the appearance of having been organized and verified. If the source text is accurate, the summary may still omit context that matters for your specific research question. In both cases, manually cross-checking the exact source text passage before acting on any AI-generated research note is the professional standard.

What Requires Mandatory Ticker and Source Verification

Four categories of AI-generated research content require explicit manual verification before any use: ticker symbols (AI may typo or confuse similar symbols — verifying HTGC from a loose typo like HGTC matters when the research is live); dates and timelines (dividend announcement schedules, regulatory milestones, and earnings dates must match official company communications); numerical claims (any figures in an AI summary must trace to a primary source — AI does not calculate, it pattern-matches text); and regulatory statements (SEC filing descriptions, compliance language, and corporate governance claims require the original document for confirmation).

Building Source-Check Habits Into Every Research Session

Build a source verification step into every AI-assisted research workflow as a named gate, not as an optional afterthought. Before any AI-generated summary is filed in your research notes, acted upon in a portfolio review conversation, or shared with another investor, the reviewer confirms: what primary source was this summary based on, has that source been checked, and does the summary accurately reflect the source without adding outside assumptions? This three-part check takes two minutes and prevents the compounding of errors through AI-assisted research cycles.

The Pre-Flight Quality Gate Before Filing Any Research Document

The Pre-Flight Market Report Quality Sign-off prompt in the pack structures a five-gate final verification for AI-assisted research documents: Privacy and MNPI checkpoints (no unmasked ledger entries or identity strings); Fact-Grounding Verification (every ticker, deadline, and process step matches primary files); No Value or Return Hallucinations (manual validation for any numerical return metrics); Voice and Authenticity Audit (stripping AI-generated buzzwords); and Final Human Sign-Off and Archival Location. Apply this gate before filing any AI-assisted research document in your tracking system.