The 5-Minute Source-Checking Workflow
"Verify every citation" only works if it's fast.
Lesson 5.1 ended with the rule: every citation in your bibliography is one you've personally opened. Great. The honest follow-up question is: how long does that actually take?
So this lesson is the workflow. Done correctly, source-checking is a quick per-source habit. It uses three tools you already have free access to (Google Scholar, your school library, and the journal site itself), in a specific order. By the end of this lesson you'll know the order, you'll know what kills a source at each step, and you'll know the prompt that makes Claude triage your list before you spend any verification time.
What "checking" actually has to confirm
For a source to make it into your bibliography, four things have to be true: (1) the article exists, (2) the author is correctly attributed, (3) it's actually about what you think it's about, and (4) any quote or specific claim you're going to use is actually in the text. The checklist below tests all four.
The Source-Checking Workflow.
Run this on every AI-surfaced source before it counts as part of your paper.
The five steps
- 1 · Google Scholar search. Open scholar.google.com. Paste the article title in quotes. If nothing comes up, drop the quotes and try the author + 3–4 distinctive words from the title. Pass: the article shows up. Fail: nothing matching shows up. The source is likely invented; drop it from your list.
- 2 · Author cross-check. Click the author's name in Google Scholar. Look at their other publications. Does this article appear in their list? Does the author actually publish in the field your topic is in? Fail-mode flag: the author is real but their publication list doesn't include this paper or anything close to its topic. That's the "real author, fake paper" flavor from 5.1. Drop it.
- 3 · Open the actual paper. Click through to the PDF, either via the Google Scholar link, your library's database, or the journal's site. Fail: can't access it anywhere. Note it as "found but unavailable" and decide if you can replace it; don't cite a paper you couldn't read.
- 4 · Skim the abstract + the section your claim came from. Confirm the article is actually about what Claude said it's about. Find the section / page where your claim or quote is supposed to be. Fail: the article is about a different topic, the quote isn't there, or the page numbers don't match. Drop the source or, if the article itself is still useful, fix the citation details and rewrite the claim from what's actually there.
- 5 · Lock in the citation details from the actual source. Pull author name, full title, journal, year, volume, issue, pages, DOI directly from the article you just opened. Paste them into your bibliography.
The bypass move when you're already in your library database
If your school's library has a good search (most do), you can collapse steps 1–3 by skipping Google Scholar and going straight to your library's catalog. Title-in-quotes search → result → PDF.
When you find a fake: what happens next.
You ran the workflow on a Claude-suggested source and it fails one of the steps. The instinct is to ask Claude to find a real version. That instinct is wrong; you're back in the same loop. Three moves work better.
Don't: ask Claude to "find a real version"
If you ask "this one was fake, can you find a real one on the same topic?", Claude will produce another plausible-sounding citation that may or may not be real. The fix isn't a more careful AI ask. It's an AI-free move.
Do: search the topic yourself
Take the angle Claude was trying to fill (e.g., "a sociology paper on the cost of municipal water privatization in mid-sized cities"), open Google Scholar or your library, and search the angle directly. You'll find real papers in 3–5 minutes. The angle was the useful part. The made-up citation was the dangerous part. Keep the angle, drop the citation, build from real results.
Do: ask the librarian
Every academic library has research librarians who specialize in helping students find sources. They are excellent at it.
Building the "search the angle" muscle
Source-finding is something you do (in databases that index real papers) and AI helps you map angles, summarize what you find, and pressure-test your synthesis. Internalizing that order of operations is the deeper lesson of this module.
The verification habit IS the integrity habit, applied to research
Honest Work Code · Rule 2: your work survives scrutiny. Module 2.4 made the case for verifying any factual claim. This lesson applies the same logic to the most-fragile single artifact in academic work: your citations. A bibliography is a promise. Each entry says "I read this, this exists, this says what I claim it says." Every promise gets tested against a real database before you sign your name to the paper. The students who run the workflow never get the office-hours email about a fake citation. The students who skip it find out the hard way that the click-the-link test is the easiest test a professor can run, and the fastest one to fail. Run the workflow. Every paper. Every source.
Up next: once your sources are real, how do you read 12 PDFs in one weekend?
Once your sources are real, the next problem is the volume. How do you handle 12 sources at 25 pages each when the paper's due Friday? Lesson 5.3 is the speed-reading workflow: the prompts that produce the 4-bullet summary, the 5 best quotes with verifiable page numbers, and the "what's not in this paper" map that tells you which 10% to read closely.
Continue to 5.3 → Speed-Reading Dense PDFs