Lv.1 0 XP

Source Attribution — Tracking Claims to Their Sources

Core 7 min +30 XP
💡
THE ANALOGY

Academic citations in a research paper. Not just 'studies show X' — but 'Smith et al. (2023) in the Journal of Y found X with a sample size of N.' The citation lets any reader verify the claim independently. Source attribution in AI systems does the same.

⚠️ EXAM TRAP — The Wrong Answer People Choose

Treating source attribution as a nice-to-have for user experience. It's a reliability requirement: when Claude synthesizes from multiple sources, claims must be traceable back to specific source documents — especially when sources conflict.

KEY POINTS
1 Every factual claim in synthesized output must be traceable to a specific source document.
2 Claim-source mapping survives synthesis: attribution must be maintained through summarization and combination steps.
3 Conflicting sources: when two sources disagree, both must be cited and the conflict noted — not one arbitrarily chosen.
4 Temporal data requires publication date: 'adoption rate is 45%' is meaningless without 'per Gartner 2023 survey'.
5 Source quality affects claim confidence: peer-reviewed > industry report > blog post.