
Build a source-verification script
You'll have a working script that takes AI-generated research output and automatically checks citations against original sources, flagging unsourced or weak claims.
What you'll make
Source Verification Script
- extract_citations() — pulls URLs and claims from AI output: what to include
- verify_source() — checks each citation against the original: what to include
- flag_problems() — identifies unsourced claims and broken links: what to include
- main() — orchestrates verification and outputs flagged content: what to include
Paste your Python script with example output showing flagged claims from a real AI research report
The 4 parts
Each part ends with a check you can see.
Test AI Research Citations
Generate AI research on a topic you know well and manually check where citations break down
Done when
You have AI-generated research output with at least 5 citations, 3 of which you've verified as accurate/inaccurate by hand
Build Citation Extractor
Write the function that pulls citations and claims from AI research output
Done when
You have a working extract_citations() function that parses AI research text and returns structured citation data
Build Source Verifier
Write the function that checks extracted citations against original sources
Done when
You have a working verify_source() function that fetches URLs and returns match/no-match judgments for claims
Test Complete Pipeline
Wire together the extraction and verification functions into a complete script and test on fresh AI research
Done when
You have a working end-to-end script that processes AI research output and produces a flagged report showing verification results
Capstone: you put together your Source Verification Script.
Where this helps
An elective: a single lesson that sits outside rtwo's job certificates and doesn't count toward one.
About this lesson
What you'll be able to do
- Run a working verification script on AI-generated research output that flags unsourced claims
- Identify the 3-5 citation patterns where AI consistently fails vs where verification catches real errors
- Have a documented workflow for checking AI research against original sources before using it in decisions
Who it's for
- Researchers and analysts who use AI for background research but need to verify claims
- Anyone who's been burned by AI hallucinations in research and wants an automated safety net
- Teams that need to cite sources in reports but want to use AI as a research accelerator
What it won't cover
- Building a full fact-checking database or professional verification service
- Legal-grade source validation or academic plagiarism detection
- Replacing human judgment about source quality — this flags missing citations, not bad sources
Skills you'll practise
- Citation pattern recognition in AI output
- Automated source-checking workflow design
- AI research quality assessment
- Hybrid verification system debugging