How to learn from an article or web page without just skimming it

To learn from an article instead of skimming it, slow the reading down on purpose and make yourself produce something after each part. Here is the method.

1. Spend a minute finding out who wrote the page before you read it.
2. Decide whether you want to learn the page itself or the topic it is about.
3. Preview the headings and turn them into questions.
4. Read one section, look away, and say what it claimed.
5. Follow side questions one at a time, and come back.
6. Trace the big claims to their original source.

An AI tutor can help with steps 3 to 6. It cannot do them for you, and it should not try.

Step 1. Find out what you are reading

Mike Caulfield, who designed the SIFT method for checking online information, puts it this way. "You want to know what you're reading before you read it." He suggests a quick check, often about a minute, on who is behind a source and what they might want.

The fastest way to do that is to leave the page. In a Stanford History Education Group study, professional fact checkers, historians and Stanford undergraduates evaluated real websites. The fact checkers did best. They scanned a site quickly, then opened "a series of additional browser tabs, seeking context and perspective from other sites." The historians and students mostly stayed on the original page and were often fooled by cues like a professional-looking name or logo, scholarly references or a nonprofit web address.

So before you invest an evening in a page, search the author's name and the publication in a new tab. A specialist writing in their field, a journalist summarizing a study and a company selling the thing it describes are all worth reading. They need different amounts of salt.

Step 2. The page, or the topic?

This decision does more than any wording trick to shape what you get from an AI tutor. Say you found a long article arguing that cities should remove downtown highways.

If you want to understand and judge that argument, ask for the page.

Teach me what this article argues, section by section. Flag anything it claims without evidence, and tell me what someone who disagrees would say.

If the article just made you curious, ask for the topic.

Use this article as a starting point and teach me how urban highways affect traffic and neighborhoods. I know nothing about urban planning.

The first course stays close to the text. The second treats the text as a doorway. Decide which you want before you start, because it changes what you ask for at every step after.

Switch between the two goals. Pick the question that matches why you opened the article.

In a general chatbot you can paste the link with one of those lines. In Learnfold, you paste the link on the New Course screen to start the course, and you see a plan to approve before anything is written. Check the plan against the choice you made. If it is not right, don't approve it and start again with a sharper question. Once you are reading, steer with "Where next?", questions about the page and your own questions. To stay close to the article, ask what each section claims and what evidence it gives. To use it as a doorway, take Go sideways or Zoom out. If the page is behind a login or paywall, a tool may not see what you see. Pasting the text in as notes is the dependable fallback.

Step 3. Preview, then ask before you read

Harvard Library's guide to reading habits starts with previewing, looking "around" the text before reading it. Read the title, the subheads and the last paragraph. Then turn each subhead into a question. "The induced demand problem" becomes "What is induced demand, and why would it matter here?"

Now you are reading to answer something, which is a very different activity from reading to get to the end.

Step 4. Read a section, then produce something

Rereading and highlighting, the two habits many of us reach for with an article we care about, are weak. John Dunlosky's review of study techniques for teachers notes that "rereading has inconsistent effects on student learning" and that highlighting "has been shown to have failed to help students of all sorts." Harvard's guide puts it more bluntly. "Throw away your highlighter."

What works better is making your brain produce the idea. After each section, look away and say what it claimed, in one or two sentences, in your own words. Harvard's version is "The best way to determine that you've really gotten the point is to be able to state it in your own words."

Then ask why it is true. The Learning Scientists describe this as elaboration, asking yourself "how these ideas work and why," and they add the step people skip, which is to double-check your explanation against the material.

With an AI tutor, this loop gets quicker. Learnfold ends a page with optional checks, where you can choose multiple choice, a written answer or a guided discussion, and it gives feedback on what you got right. Choose the written answer. It is harder, which is the point.

Step 5. Follow side questions, one at a time

Good articles throw off questions. Some are worth chasing. In Learnfold, every lesson ends with "Where next?" and four kinds of doors. Go deeper takes you into something the page mentioned briefly. Go sideways gives you a related or opposite idea. Zoom out shows the bigger picture. Or you type your own question. Picking one writes a new page linked to the lesson it came from.

The risk is the classic browser spiral, forty tabs and no memory of why you opened the first one. Caulfield's "Stop" move covers this too. When you notice you are lost in tangents, pause and remember what you set out to learn. Learnfold's course map helps here. It shows every page you opened with a "You are here" marker, and you can pin a page as a goal so the detour has somewhere to come back to. More on that habit in following your curiosity without losing the thread.

Step 6. Verify against the source

When you learn from a page with an AI tutor, you have two layers that can be wrong. The page can be wrong, and the AI's explanation of the page can be wrong. Learnfold's own note says as much. It uses AI to write explanations, AI makes mistakes, and you should treat it as a tutor, not a textbook, and check important facts against a reliable source.

Caulfield's last move is to trace claims, quotes and media back to their original context. If the article says a study found something, find the study. If a lesson summarizes the article, find the sentence in the article. In Learnfold, type your own question about the page.

Which sentence in the original article supports the claim about traffic? Quote it.

Then go look. The full checklist is in how to check an AI explanation for mistakes.

When this goes wrong

The course says things the article never said. You probably asked for the topic, or the tutor filled gaps from general knowledge. Ask which parts came from the page.

You feel like you learned it, but cannot explain it. You skimmed with extra steps. Go back to step 4 and write two sentences without looking.

The article turns out to be weak. That is a result, not a failure. Find better coverage, Caulfield's third move, and start from a stronger source.

You have ten links, not one. Then you have a reading list problem. See how to turn a reading list into a learning plan, and use better questions for your AI tutor along the way.