AI Activities for Teaching Primary Sources

AI activities for teaching primary sources work best when they scaffold the four historical thinking skills researchers actually measure: sourcing, contextualization, close reading, and corroboration. AI can generate leveled questions for a document, draft a companion source for comparison, or simulate a "sourcing detective" dialogue — but it should never write history for students or stand in as the document itself.

Quick answer: Use AI to generate scaffolded questions, leveled transcriptions, and comparison documents around a real primary source — never to invent the source content itself. The strongest activities pair one authentic document (from the Library of Congress, National Archives, or a textbook) with AI-generated sourcing questions, corroborating context, and differentiated reading levels.

Here's the uncomfortable finding behind that advice. In a widely cited 2016 study, the Stanford History Education Group (Wineburg & McGrew, 2016) gave middle school, high school, and college students a series of online sources to evaluate — and found the large majority could not reliably distinguish a sponsored ad from a news story, or a lobbying group's report from independent research. That was before generative AI existed. The skill gap it exposed — sourcing before believing — is exactly what primary source instruction is supposed to build.

The stakes have not improved on their own. The National Assessment of Educational Progress (NAEP, 2022) reported that eighth-grade U.S. history scores fell to their lowest level since the assessment began in 1994, with only 14% of students performing at or above the "proficient" level. Reading a document and reading it critically are different skills, and most primary source instruction still asks students to do the harder one without much scaffolding.

This guide walks through what actually works: the historical thinking framework AI activities should target, five specific classroom activities, a tool comparison, a step-by-step workflow, and the pitfalls that turn a promising AI-assisted lesson into a shortcut around thinking rather than a support for it.

Why Primary Source Analysis Is Hard to Teach Well

Primary source analysis is difficult to teach because it asks students to hold two things in mind at once — what a document says, and how much to trust what it says — and most classroom time is spent on the first while the second gets a worksheet at best.

Historians call this historical thinking, and the Stanford History Education Group's "Reading Like a Historian" framework (Wineburg, Martin & Monte-Sano, 2013) breaks it into four moves:

• Sourcing — who wrote this, when, and why, before trusting a word of it
• Contextualization — what was happening in the world at the moment the source was created
• Close reading — what the words, images, or artifacts actually say and imply
• Corroboration — checking one source's claims against others

Most students default to what researchers call "naive reading" — treating a document as a neutral window onto the past rather than a perspective shaped by its author's position and purpose. A diary entry, a propaganda poster, and a government report all feel equally authoritative to an untrained reader, even though they are not.

The College, Career, and Civic Life (C3) Framework published by the National Council for the Social Studies (NCSS, 2013) makes sourcing and evaluating evidence a formal inquiry standard from elementary grades onward — not just a high school AP History skill. That is precisely where AI activities can help most, because AI is good at generating the scaffolding questions that turn naive reading into sourced reading, at whatever grade level you teach.

Elementary vs. Middle School Approaches

The same four historical thinking skills apply from Grade 3 through Grade 9, but the scaffolding depth changes a great deal. Younger students need the sourcing questions modeled aloud, in whole-class discussion, before they attempt anything independently.

For Grade 3-4, keep documents short — a single photograph, a short advertisement, or one paragraph of a letter — and ask only one or two sourcing questions at a time. For Grade 7-9, students can hold a full document plus a companion source and work through all four thinking moves in a single class period.

Say you teach Grade 4 and you want to introduce sourcing for the first time: you could start with just one question — "who took this photograph, and why might they have wanted people to see it?" — asked about a single Depression-era image, before ever introducing corroboration or contextualization as separate ideas.

The Gap Between Access and Analysis

Digitized archives have made more primary sources available to K-9 classrooms than at any point in history — the Library of Congress's Teaching with Primary Sources program and the National Archives' DocsTeach tool both offer thousands of free, classroom-ready documents. Access was the old bottleneck; interpretation is the current one.

Handing a class a scanned 1863 letter and asking "what does this tell us?" without scaffolding tends to produce surface-level answers. Students paraphrase instead of analyze. AI-generated sourcing prompts close that gap by asking the questions a historian would ask before students form a conclusion.

Five AI Activities for Teaching Primary Sources

The most effective AI activities for primary sources fall into five types: sourcing detective prompts, leveled transcription support, perspective comparison sets, corroboration companions, and Socratic bias-spotting dialogues. Each targets a specific historical thinking skill rather than treating "analyze this document" as one undifferentiated task.

1. The Sourcing Detective Warm-Up

Before students read a document's content, have them interrogate its origin. You could feed a real document's citation information — author, date, format, and archive — into an AI tool and ask it to generate five sourcing questions specific to that document, rather than generic ones.

For a Grade 5 unit on westward expansion, say you're using an 1870s homestead advertisement from the National Archives. You could prompt an AI tool to generate questions like:

1. Who paid to create and distribute this advertisement, and what did they gain if people believed it?
2. What does the advertisement leave out that a settler might have needed to know?
3. Would a Native American living on this land in the 1870s have described it the same way?

Students answer these before reading the document's content closely, which research on historical thinking instruction consistently shows produces more skeptical, source-aware reading than content-first approaches.

2. Leveled Transcription and Vocabulary Support

Handwritten and archaic-language documents are often the biggest barrier to primary source work in K-9 classrooms — not the historical thinking itself, but the raw decoding. A student who cannot parse 1860s cursive or Elizabethan-era phrasing never gets to the analysis stage at all.

You could use an AI tool to generate a plain-language, grade-appropriate paraphrase of a difficult passage alongside the original text — never as a replacement for it, but as a bridge. Pair the original document with:

• A vocabulary list of 5-8 unfamiliar terms with kid-friendly definitions
• A modern-English paraphrase of the hardest paragraph, displayed side-by-side
• Guiding questions that send students back to the original wording to check the paraphrase

This mirrors the "text-leveling" approach already common in ELA instruction, applied to the specific problem of archival language rather than general reading level.

3. Perspective Comparison Sets

Corroboration — checking one account against another — is hard to teach when a classroom only has one document on a topic. AI can help you assemble a comparison set: a real primary source alongside AI-generated discussion prompts that surface how a second, differently-positioned observer might have described the same event.

Say you teach a Grade 7 unit on the Industrial Revolution and you have one real factory owner's report praising working conditions. You could ask an AI tool to generate discussion questions — never a fabricated second document — that prompt students to imagine and then research how a child laborer, a labor organizer, or a government inspector might have described the same factory differently, then send them to find a real second source to test their prediction.

4. The Corroboration Companion

Once students have analyzed one document, AI can generate a structured comparison worksheet that pairs it with a second, real source you provide — a newspaper account next to a personal letter, or a government record next to an oral history excerpt.

A useful three-column format for this:

You could ask an AI tool to generate the specific comparison questions for your two chosen documents, then have students fill in the columns themselves. The corroboration work — actually comparing the evidence — stays entirely with the student.

5. Socratic Bias-Spotting Dialogue

For upper-elementary and middle school classes ready for more independence, AI chat tools can role-play a Socratic questioner. A student explains their initial read of a document, and the AI asks follow-up questions ("What makes you say the author is being neutral here? What would change your mind?") rather than supplying an answer.

This works best as a teacher-mediated, whole-class activity for younger grades and a monitored individual activity for Grade 7-9 students, always with a debrief where the class compares what the AI pushed them to notice.

Tools for Primary Source AI Activities

Not every tool built for primary sources is an AI tool, and not every AI tool is built for primary sources — the useful combination pairs an archive with an AI-assisted question generator.

EduGenius can generate a differentiated primary source worksheet — sourcing questions, a vocabulary scaffold, and a short comprehension check — from a document you paste in, and its Class Profiles feature lets you set the grade level and ability range so the questions adjust automatically for a mixed-readiness classroom. Its Bloom's Taxonomy alignment is useful here specifically because sourcing and corroboration sit at the "analyze" and "evaluate" levels, not the "remember" level most generic worksheets default to.

A Step-by-Step Classroom Workflow

Here is a repeatable five-step sequence for building an AI-assisted primary source lesson, from document selection through corroboration.

1. Choose one authentic document. Pull it from the Library of Congress, National Archives DocsTeach, or a textbook — never an AI-generated "source." Confirm the citation: author, date, and original context.
2. Generate sourcing questions before content questions. Use an AI tool to draft 4-5 sourcing questions specific to that document's author, date, and format. Have students answer these first.
3. Level the language, not the thinking. If the document's vocabulary is a barrier, generate a paraphrase or glossary alongside the original — but keep the analytical questions at full rigor for every student.
4. Add a second, real source for corroboration. Use AI to generate the comparison questions, but the second document itself should be authentic — pulled from an archive, not invented.
5. Debrief with a Socratic follow-up. Close with a class discussion (optionally AI-assisted for individual practice) that asks students to justify their conclusions with specific textual evidence.

A concrete illustration: say you teach Grade 6 and you're covering the Boston Tea Party. You could pull a real Loyalist newspaper account and a real Patriot pamphlet from a digitized archive, generate sourcing questions for each with an AI tool, and have students complete a corroboration table before writing a paragraph on which claims both sources agree on.

That sequence takes roughly one 45-minute class period once you've built the routine, and it scales down easily — for a younger class, drop step 4 entirely and stop after the sourcing questions in step 2, saving corroboration for a later unit once sourcing is second nature.

Fitting This Into an Existing Unit

You don't need a dedicated "primary sources week" to use these activities. A single well-sourced document dropped into an existing social studies, science history, or civics unit does more for historical thinking than a standalone unit that never recurs.

Consider building sourcing questions into the routine you already use for warm-ups or exit tickets, so the skill gets repeated practice across the year rather than a single isolated lesson students forget by spring.