Pathfinder Guidebook · FYE 100 Module 4 · Your Academic Toolkit
10 Essential Skills in this module
  • 6Professionalism
  • 10Information Literacy
Lesson 4.3

SIFT: Stop Before You Scroll

Four moves for evaluating any information, including what AI tells you.

About 9 minutes 4 of 8 in this module

Every day you run into dozens of pieces of information: articles, posts, search results, AI answers, videos, and claims from people you trust.

Most of it looks credible, but some of it isn't, and the gap between "looks credible" and "is credible" has never been harder to spot.

SIFT is a practical method for closing that gap. It was developed by information literacy researcher Mike Caulfield, and it gives you four moves you can make quickly, in the moment, before you believe, share, or act on something. It works on news articles, social media, and AI-generated content, which may be where it matters most right now.

Before you start: consider this

You ask an AI tool, "What are the best study strategies for college students?" The answer is detailed, confident, and well organized. It mentions "research" and uses phrases like "studies show" and "experts recommend."

It sounds authoritative. It reads like something from a credible source. You take notes and plan to follow the advice.

The problem: AI tools sound authoritative whether or not they're accurate. There may be no actual study behind "studies show." The "expert" may not exist. The advice may be useful, or it may be a believable mix of things the AI was trained on. Without checking, you can't tell which. That's what SIFT is for.

The four moves

Stop

Before you read the article, stop. Before you share the video, stop. Before you act on a strong reaction to a headline or a convincing AI summary, stop.

This is the hardest move, not because it's complicated, but because the information around you is designed to keep you moving. Headlines are written to provoke a reaction. AI answers are formatted to feel complete and final. Social media rewards the fast share.

Stopping means pausing before your first reaction becomes your conclusion, not doubting everything. Ask yourself: Do I know this source? Do I know its reputation? Am I reacting to how confident this sounds instead of whether it's accurate?

Applied to AIWhen an AI answer sounds complete and authoritative, stop. That confidence comes from how AI writes, not from evidence that the content is right, so pause before you use it.

Watch: Stop, with Mike Caulfield (YouTube) (opens in a new tab)

Investigate the source

Before you read deeply, take a minute to find out what you're reading. Who made this? What's their purpose? What's the reputation of this person or organization? You're looking for context before you invest your attention, not reasons to dismiss it.

The key technique is lateral reading: instead of reading deeper into the source itself, open a new tab and search for what others say about it. Professional fact-checkers work this way, and it's far more effective than judging credibility from the inside.

  • Search the name of the publication, website, or author along with words like "bias," "credibility," or "fact-check."
  • Check Wikipedia for background on news organizations, think tanks, or advocacy groups, as a quick starting point, not a final authority.
  • Look at the "About" page, but don't stop there. Anyone can write an impressive About page.

Applied to AIYou can't look up an AI's reputation, but you can check its claims. If an AI answer names a specific study, expert, or statistic, investigate that claim directly. Does the study exist? Does the expert? Where does the number come from?

Watch: Investigate the Source, with Mike Caulfield (YouTube) (opens in a new tab)

Key Term

Lateral reading: checking a source by leaving it and searching what other sources say about it.

Pathfinder Tip

Lateral reading takes less than two minutes and catches things deep reading misses. Open a new tab before you commit to a source. It's the single most effective habit in the whole method.

Find better coverage

Sometimes the best move is to stop judging the source you have and find a better source for the same information. If a claim matters enough to act on, it's worth seeing whether other credible sources report it the same way.

Look for sources with a track record of accuracy, not sources that agree with you, and see how they handle the same claim.

Applied to AIWhen AI gives you information you plan to use for an assignment or a decision, find better coverage from a credible source before you rely on it. The AI answer is a starting point, not a destination.

Watch: Find Better Coverage, with Mike Caulfield (YouTube) (opens in a new tab)

Trace claims, quotes, and media to the original

A lot of misinformation starts with something real, taken out of context: a real statistic used to support a false conclusion, a real quote trimmed to mean something the speaker didn't say, a real study cited for a finding it never showed.

Tracing means following a claim back to where it started and asking: does the original actually say what this source claims it says?

  • Click through links. Don't assume a linked source says what the article claims.
  • When a statistic is cited, find the original study or report, not just the article about it.
  • When a quote is used, find the full quote in its original setting. Speeches, interviews, and documents are usually findable.
  • For images or videos, use a reverse image search to find where they first appeared.

Applied to AIAI tools sometimes produce citations that look real but don't exist. This is called a hallucination. If an AI answer cites a source, trace it. If you can't find it, don't use it. If you find it and it says something different from what the AI claimed, that's your answer.

Watch: Trace Claims, with Mike Caulfield (YouTube) (opens in a new tab)

Pathfinder Tip

Most misinformation uses real information in misleading ways rather than telling an outright lie. Tracing takes longer than the other moves, but it's the one that finds the biggest problems.

SIFT and AI: why it matters more than ever

AI Literacy BE Critical

Traditional misinformation often has tells: obvious bias, unknown sources, claims that seem too good to be true. AI output often has none of them. It's fluent, organized, confident, and often mostly correct, which makes the errors harder to spot and easier to act on.

  • Stop. Don't use an answer just because it sounds complete. Sounding complete is a style, not a sign of reliability.
  • Investigate. You can't investigate the AI, but you can investigate its claims. Every specific fact, number, or citation can be checked on its own.
  • Find better coverage. Use the AI answer as a starting point, then find credible sources on the same topic. The AI may have pointed you somewhere useful even if its details need checking.
  • Trace. If an AI cites a source, find it. AI can invent references that don't exist, or that say something different from what it claimed.

The habit: treat AI output like a smart but unverified tip from a well-informed friend. It's worth listening to and following up on, but not worth acting on until you've checked.

See also

SIFT is for sources. For checking an answer from AI, Before You Use AI's Answer (opens in a new tab) gives you four quick checks.

Practice it: Is This Legit?

In your course, you'll put SIFT to work. You'll read a short document, then talk it through with Coach Pathfinder, an AI coach who walks you through the four SIFT moves one at a time. Coach Pathfinder asks questions; it doesn't hand you answers. Your job is to do the evaluation, not to describe what you would do.

  • Read the document first, before you open the chat. You'll find it with this module's activities in Blackboard.
  • Keep it open in another tab the whole time. You'll need to refer back to it.
  • Give it 10 to 15 minutes of real attention. This isn't a quick task.
  • Don't rush a step because it feels done. Coach Pathfinder may push you to go deeper.

There are real things to find in this document. Some are obvious and some take digging, so take your time.

Keep handy

Keep this page open for a refresher on the four moves while you chat.

SIFT method developed by Mike Caulfield. Parts of this lesson are adapted from Clark College Libraries (opens in a new tab) and Hapgood (opens in a new tab) under a CC BY 4.0 (opens in a new tab) license.