A first-grader asks an AI chatbot why the sky is blue and takes the answer at face value. A high-schooler scrolls through a social media feed of takes on a current event without much framework for sorting the factual from the manufactured. Both scenes share the same pattern: a generation growing up on fluent, confident-sounding information, with no working practice for deciding what to trust. By the time the first-grader is old enough to vote, the pattern will have been reinforced for more than a decade.
Every information shift demands a new literacy. Twentieth-century kids learned to tell the news from the commercials, and to recognize a reporter’s byline as a different kind of authority than a pundit’s opinion. The internet brought a harder task, because anyone could publish anything, and a slick-looking website wasn’t automatically a trustworthy one. Social media layered an invisible filter on top, since the content reaching a kid had been sorted and shaped by systems optimizing for engagement long before anyone chose what to look at. AI is the latest turn of this arc, and it collapses the task further: a chatbot answer arrives confident and without citations, potentially with errors nested inside well-formed sentences and no visible trace of who produced it or how. So for a kid without evaluation skills, a news article and a viral post can land with roughly the same weight, and the cost of failing to tell them apart can have significant consequences as they get older.
Kids who grow up practicing evaluation skills enter adulthood able to meet whatever new claim shows up in front of them and hold it against a working understanding of how information gets made and whose interests it serves. They can disagree with the people around them because they’ve traced arguments back to their origins, and they’re better equipped to update their thinking when the evidence calls for it. What they then carry into adulthood is the ability to form their own opinions rather than inherit someone else’s. That ability shapes everything downstream of it, from the choices they make about their health and money to the votes they cast and the people they trust.
What we mean by information literacy
Library science has been developing evaluation skills for well over a century, and the field has a formal framework for teaching them. In 2016, the Association of College and Research Libraries published the Framework for Information Literacy for Higher Education. The framework replaced an earlier set of teaching standards from 2000 that had been organized around a checklist approach, where students learned to locate sources and evaluate them against fixed criteria.
The new framework moved toward a set of “threshold concepts.” A threshold concept is the kind of idea that reorganizes how someone thinks about a whole field once they grasp it. Learning about compound interest, for instance, permanently shifts how a person reads any decision involving money over time, in a way that wouldn’t have been available before the concept landed. The framework identified six such concepts in information literacy and called them “frames,” each describing a dimension of how information functions in the world. A kid who’s been taught the frames doesn’t just know the procedural steps of evaluating a source; she sees information differently, which is what carries her across whatever new tools and contexts she’ll meet over a lifetime.
The framework was written for college students and their librarians. But the frames it describes apply just as well to the information environments kids are navigating at home and in K-12 classrooms. One librarian practice sits upstream of the frames, shaping every information interaction that follows it: the reference interview.
The reference interview
At library reference desks, the question someone asks out loud is almost never the question they really need answered. A student might walk up and ask for “something about the French Revolution” when the material they need is something more specific, like a study of women’s political organizing in late-eighteenth-century Paris. Answering the literal question would send them in the wrong direction, so librarians developed a short protocol for surfacing the real question first. That protocol is called the reference interview.
The interview uses a handful of specific techniques:
Open-ended follow-ups that invite the patron to say more about the underlying project
Questions about what the patron already knows, and what they plan to do with the answer once they have it
Restating the question back to check that it’s been heard correctly
A competent reference interview can turn a vague question into a sharp one, and that sharper question is what makes the rest of the research succeed. Without it, hours can be lost following a poorly-framed question down paths that were never going to answer what the patron actually needed to know.
The same practice transfers to how kids interact with any information source, including school research and AI chatbots. Teaching kids to interview their own questions before consulting any source is one of the most useful habits information literacy can build. A kid who learns this practice young carries it into every research task they’ll do, and the small early work of clarifying a question saves substantial wasted effort on vague or misguided answers later.
In the home: This move matters most right before your kid types into a chatbot or search engine, since these tools produce confident-sounding answers to vague questions just as readily as to sharp ones. Next time you’re about to look something up together, pause before either of you starts the search. If your kid asks “what’s the best video game console,” try responding with “best for what kind of games?” The question that comes back is usually one with a more useful answer. With a younger child wondering why sharks are dangerous, asking “what are you trying to figure out about them?” can surface what they were really after. Doing this with your kid gives them a model they can draw on later, when they’re consulting a chatbot or search bar without anyone next to them.
In the classroom: Before any research assignment, pair students up and have them interview each other’s topics, with one student playing librarian and the other playing researcher. A short exchange of this kind reliably turns a vague prompt into better research questions. The habit transfers to how students interview their own questions when they’re working alone, which is where the exercise has its biggest effect across the course of a research project.
The six frames
The reference interview clarifies what’s actually being asked before any source gets consulted. The six frames that follow do the next layer of work, which is evaluating what those sources contain once the question itself is clear. Each frame names a different dimension of how information functions in the world, and each one comes with its own way of reading sources that holds up across formats, platforms, and tools.
1. Not every expert is an expert on everything.
Framework name: Authority Is Constructed and Contextual
Authority isn’t a permanent quality a person has. Specific communities grant authority for specific claims, and that authority rarely travels smoothly across domains. A cardiologist has real authority on heart disease, but that authority doesn't extend to questions about mental health treatment, even though she holds a medical degree. A celebrity endorsing a wellness product brings name recognition to the product, but name recognition isn’t the same as knowing whether the product actually works.
This frame replaces the flat question of whether a source is trustworthy with the sharper question of what a source is trustworthy on. AI output complicates the picture further, because a chatbot answer carries the tonal texture of expertise without coming from any particular expert. Kids who learn this frame young develop the habit of asking what is this person trained in, and does this claim fall inside that area?
In the home: When your kid says “my teacher said...” or “the doctor said...” take the opening to ask a light contrast question. You can say “Does your dentist know a lot about teeth? What about building rockets?” The contrast makes the point without lecturing: expertise is specific to the area someone trained in. Over time, your kid can begin to ask on their own whether a source is speaking from their area of training or reaching beyond it.
In the classroom: Give students a short piece of writing where a credentialed person makes several different kinds of claims. An op-ed by a famous doctor on a political issue works well, as does a business executive writing about a scientific topic. Have students mark which claims fall inside the writer’s training and which reach outside it. The exercise demonstrates that credentials don’t transfer automatically across domains, and the question of where someone’s authority ends becomes one students can ask of any source going forward.
2. How information gets made shapes what it can tell us.
Framework name: Information Creation as a Process
A peer-reviewed paper and an AI answer can both claim to be sources of knowledge, but they were produced by radically different processes; that process shapes what kind of knowledge each one can hold. A peer-reviewed paper takes months or years to produce, and it passes through multiple rounds of expert review before it gets published. A chatbot answer takes seconds to generate, with no human editor involved at the moment of writing. Both arrive at a reader looking like information, but only one was shaped by a process designed to catch errors.
This frame teaches kids to ask how a piece of information was produced before deciding how much to trust it. A peer-reviewed article has been through checks that catch errors, so a kid can lean on it more heavily. A chatbot answer or a social media post has been through no checks at all, so anything that matters in it should be verified against a source that did go through verification checks. The habit we’re learning is how much to trust a source in proportion to how carefully it was made.
In the home: When your kid quotes something to you, take the moment to ask where they got it. “Did you read that in a book, hear it in class, see it on YouTube, or get it from AI?” The answer decides how much work the claim still needs. A fact from a textbook has been through some kind of review, which catches basic errors even though it doesn’t catch things like motivated omissions or framing choices. A chatbot answer hasn’t been through any review at all, so anything important needs a second verifiable source before the family treats it as true.
In the classroom: Pick a topic students are already researching, and have the class find two kinds of sources on it: an AI answer and an encyclopedia entry or textbook chapter on the same question. Ask students to infer what they can about how each source was made, based on the kind of source it is. From there, they can calibrate their trust: lean on the more carefully made source, and verify anything important from the less carefully made one.
3. Behind every piece of information, somebody wants something.
Framework name: Information Has Value
Anything created for an audience came into existence for a reason, and that reason shapes the content in ways that aren’t always obvious from the surface. Advertisements are built around the goal of converting viewers into customers, which determines the kinds of claims they make and the kinds of evidence they offer. Political messaging follows the same logic for a different end, shaping content around the goal of moving an audience toward a particular position. AI has added a new dimension of this issue, since many of the major models were trained on copyrighted material without compensation or consent, and what those models now produce serves the commercial interests of the companies running them.
The frame teaches kids to ask what the maker was after when they made the piece. The answer changes what shows up in the final content and what gets left out, which means noticing the maker’s motivation is where most of the evaluation work happens. Kids who grow up reading information this way develop a default question they can ask of anything: who benefits when I take this at face value? Motivation is one of several dimensions of this ACRL frame, alongside questions of attribution, access, and whose voices get heard; it’s the dimension most directly applicable to the content kids encounter in everyday life.
In the home: When your kid wants to watch a free video or play a free game, ask them “who do you think is paying for this to exist?” Whatever they answer, the question opens a conversation you can come back to: free content always has a cost attached, and the cost just gets paid in something other than money. For a free video, your kid is paying with attention that the platform sells to advertisers. For a free game, they’re paying with time and data, and often with temptation to buy upgrades inside the game. The same approach surfaces motivations beyond money in any other kind of content. Naming this out loud, repeatedly, builds the question your kid will start asking on their own about anything that reaches them: who’s getting something out of me reading or watching this?
In the classroom: Pick a piece of content the class has seen, and spend twenty minutes tracing what the maker wanted from the audience. The incentive might be commercial, where an advertiser or sponsor is working to drive sales, or it might be persuasive, where a campaign or organization is working to shift opinion on an issue. The exercise puts the question of motivation in front of students directly, so they can practice reading content for the purpose driving it rather than only for what it says on the surface.
4. Good questions get sharper as you ask them.
Framework name: Research as Inquiry
Research doesn’t happen in a single act of looking something up. A question goes in, partial answers come back, the question gets refined based on what those partial answers reveal, and the cycle repeats. Meaningful questions rarely have clean single answers, and researching well includes the capacity to stay with a question while it sharpens rather than collapsing it too early into a premature conclusion.
AI chat trains the opposite reflex: a chatbot offers a single finished answer to a single question, and the interaction feels complete as soon as it ends. Kids who only practice that pattern miss the underlying skill of refining questions as they learn more. The skill is what separates researching a topic from simply collecting answers.
In the home: When your kid asks you a question, try bouncing it back first: “What do you already think about that?” or “How could we find out together?” These simple questions invite your kid into the thinking process rather than handing them a finished answer. Over time, children can develop their own opinions and their own research instincts, instead of waiting for the adult in the room to tell them what’s true.
In the classroom: For any research assignment, build in two checkpoints where students rewrite their research question based on what they’ve learned so far. By the end of the project, students can compare their final question to the one they started with and see how it changed. The point of the exercise is to make the iterative nature of research visible, so students experience research as something that reshapes the question itself (and not just a hunt for answers to a fixed one).
5. No serious idea stands alone.
Framework name: Scholarship as Conversation
Substantive ideas rarely get developed by people working in solitude. They emerge from groups of people thinking and writing in response to each other over time, with each contribution shaped by what came before it and shaping what comes after. Making sense of something means recognizing the larger argument it's a part of, including the earlier work it's reacting to and the work it will provoke in turn.
The frame shows up vividly in academic writing, where literature reviews and citations make the conversation visible on the page. The same pattern operates in journalism that builds on earlier reporting and in political writing that engages older arguments. AI-generated summaries do something more concerning, because they compress many conversations into a single confident-sounding voice that erases the evidence a conversation ever existed.
In the home: When you’re reading a book or watching a show with your kid, point out the moments where the book or show is reacting to something else. You can say things like “This story is making fun of those old fairy tales we read last year” or “This creator made a video to argue with what that other person posted last week.” Comments like these position the book or show as part of a larger dialogue rather than as a standalone pronouncement, introducing the question a kid can carry into anything they read or watch: what is this responding to or building on?
In the classroom: Take any text students are already reading and teach them to look at its bibliography or works cited section. The simplest question to ask is who the author is building on or arguing with. Walk the class through a single citation chain: pick one reference in the text, and look at that source together to see what it says and how the original text engaged with it. Students learn that every serious piece of writing is part of a longer, larger conversation, which can change how they read everything afterward.
6. Finding good information takes more than one search.
Framework name: Searching as Strategic Exploration
Skilled searching looks nothing like a single query producing a single result. It’s a process of trying something, seeing what comes back, adjusting based on what the results reveal about the topic and the tools, and trying again with sharper terms. The craft lives in knowing when to try again and what to vary when the first search doesn’t deliver.
AI chat has flattened this craft for many users, because a chatbot offers the surface of an answer on the first try. The pattern trains our reflexes away from iteration. Kids who grow up inside that reflex miss the underlying skill, which is the ability to move strategically through an information landscape when a first attempt doesn’t give them what they need.
In the home: When your kid watches you look something up, show them that you try more than one search. Say out loud when a search doesn’t give you what you need, and try a different approach. You can narrate it directly, with something like “Hmm, that didn’t work, let me try different words,” or “this site doesn’t seem reliable, let me look somewhere else.” The modeling teaches your kid that search is an iterative process rather than a single click.
In the classroom: Put several search tools in front of students and have them run the same question through each one. A library database and an AI chatbot will return noticeably different results for the same question, and comparing the differences shows students how information is structured in ways that aren’t visible from any single tool. The exercise also makes clear that no search tool gives a complete picture, because each one is designed to prioritize certain results over others. The ability to turn to multiple sources, or refine their search process over time, is the foundation to students’ verification strategies as they get older.
The arc over time
The information environment kids are growing up in now will keep shifting, and the shifts will inevitably keep coming. What makes the ACRL framework durable across these shifts is that the frames describe dimensions of information that stay stable when the tools change. Authority is always specific to a domain. Information is always produced through some process, and the process shapes what the output can carry. Every piece of information moves through systems of incentive, and every serious idea takes part in a larger conversation, whether or not the tool it arrives through makes the conversation visible.
Kids raised on these habits grow into adults who can encounter new information and truly see what’s in front of them. They can recognize when an author is reaching beyond their expertise, when incentives are shaping what gets said, when an article is in conversation with other articles, and when the first search result isn’t the last word. They can grow into voters and citizens who think the way they want to think rather than the way someone else wanted them to. This capacity gets built in the small conversations of home and classroom, long before the first vote gets cast or the first big decision needs making.
What these skills add up to has a name: information resilience, the capacity to meet any claim, from any source, without being knocked off course by what arrives. A resilient reader can stay with a question while it sharpens, and can hold her ground when the people around her are settling on conclusions faster than the evidence warrants. She has an internal sense for what’s worth a closer look, built up over years of practice. That kind of resilience is what parents and teachers can give the kids growing up now, and it will keep working long after any specific tool or platform has been replaced.
There's a website that I used to help kids evaluate information. Its called dhmo.org. The information is accurate, but is presented in an over-the-top, inflammatory way. Dhmo stand for dihydrogen monoxide. Its states, among other things, that it's involved in all drownings, is found in tumors and that the government refuses to do anything about it.
It contains links to valid websites. If you remember your chemistry you will know that it is H20 or water. I tell the kids after they are done reacting to the information they find.
Thank you for this post, Hana. I'm going to send it around the staff at my school as you suggest some very useful strategies for evaluating sources of information. I really like the way you listed activities under each framework. We have some similar frameworks in Australian states (where I'm from). One is the New South Wales Department of Education 'Information Fluency Framework'. While it is designed for our primary students (I think that equates to Elementary students in the US), it can be adapted for older students.
https://education.nsw.gov.au/content/dam/main-education/teaching-and-learning/curriculum/media/documents/Information_fluency_framework.pdf