Teaching History in the Age of AI: Using Artificial Intelligence Without Surrendering Historical Thinking
Teaching History in the Age of AI: Using Artificial Intelligence Without Surrendering Historical Thinking
Artificial intelligence is already in the history classroom.
The important question is no longer whether students will use it. The question is whether educators will teach them to use it responsibly.
For history educators, this question carries particular importance. Our discipline has never been simply about finding information. Historians evaluate evidence, question sources, reconstruct context, compare interpretations, recognize uncertainty, and build arguments from incomplete records. These habits of mind are increasingly valuable in an information environment where a polished paragraph can be generated in seconds, but accuracy, provenance, and interpretation remain distinctly human responsibilities.
The American Historical Association recognized this distinction in its 2025 Guiding Principles for Artificial Intelligence in History Education. The AHA argues that generative AI can provide opportunities for teaching and learning but cannot substitute for teachers, disciplinary expertise, or historical thinking. Most importantly, it warns against allowing AI to become a shortcut that replaces meaningful intellectual work.¹
That principle should become the foundation of AI use in college history courses, high school classrooms, and homeschool education:
AI should support historical thinking, not perform historical thinking for the student.
History Is More Than Information
One of the greatest dangers posed by AI is also one of the oldest problems in history education: treating history as a collection of information to memorize.
Dates matter. Names matter. Chronology matters. Students cannot analyze a past they know nothing about. But factual knowledge is the beginning of historical education, not its final purpose.
In my own work on teaching historical thinking and civic literacy, I have argued that educators should help students analyze primary sources, evaluate evidence, distinguish fact from interpretation, recognize competing interpretations, and understand that complexity does not represent confusion. Complexity is often where meaningful historical understanding begins.
John Lewis Gaddis similarly describes historical inquiry as a process of making connections across time rather than simply accumulating disconnected facts.² My America250 work applies that principle directly to education: events such as the Revolution, Civil War, Reconstruction, women's suffrage, the New Deal, and the Civil Rights Movement should not be taught as isolated chapters. They form parts of continuing historical arguments about liberty, citizenship, equality, power, and opportunity.
That is precisely where AI can either weaken history education or strengthen it.
Ask an AI system, “What caused the Civil War?” and it can produce a competent-looking summary almost instantly.
A student who copies that answer has learned very little.
Instead, give students several historical documents and ask them to investigate:
What did different Americans believe they were fighting over in 1861?
Now the intellectual task changes.
Students must identify authors, audiences, dates, contexts, arguments, omissions, and contradictions. AI might assist somewhere in that process, but it cannot be allowed to replace it.
Put the Primary Source Back at the Center
Primary sources provide one of the best safeguards against passive AI-assisted learning.
The National Archives explicitly describes document analysis as a means of transforming students from passive recipients of information into participants in the work historians perform. Students examine evidence, establish context, make judgments, and ultimately use documents as historical evidence.³
The Library of Congress uses a similarly accessible process built around observing, reflecting, questioning, and investigating further.⁴
These methods work at virtually every educational level.
A high school student studying Reconstruction might receive a Freedmen's Bureau document, a Black political convention resolution, a newspaper account, and testimony concerning racial violence. Instead of asking AI to explain Reconstruction, the student first analyzes the documents.
Only afterward might the teacher permit AI.
The assignment could become:
Ask an AI system to explain why Reconstruction ended. Compare its explanation with the evidence you examined. Identify two claims supported by your primary sources, one claim requiring additional evidence, and anything important the AI omitted.
Now AI has become an object of historical criticism rather than an answer machine.
That is a radically different educational experience.
Teach Students to Challenge AI
We should not teach students that AI is an authority.
We should teach them to interrogate it.
The Digital Inquiry Group's Reading Like a Historian curriculum emphasizes sourcing, contextualization, corroboration, and close reading. These practices grew from research into how historians approach historical documents.⁵
Those same habits transfer remarkably well to AI literacy.
When an AI system makes a historical claim, students can ask:
- What is the claim?
- What evidence supports it?
- Where did that evidence originate?
- Can I verify it independently?
- What historical context is missing?
- Are there credible competing interpretations?
- What voices or perspectives are absent?
- Is the AI presenting interpretation as settled fact?
This turns AI literacy into an extension of historical literacy.
The College Board's AP U.S. History framework already asks students to develop related abilities: sourcing, evaluating claims and evidence, contextualization, comparison, causation, continuity and change, and argumentation.⁶
We therefore do not need to invent an entirely new intellectual framework for teaching students how to evaluate AI.
Historians already have one.
AI Should Never Invent the Archive
There is another principle I consider essential:
If you cannot verify the source, do not present it as a source.
Generative AI can produce convincing titles, quotations, page numbers, archival collections, and citations that do not exist. The prose may sound authoritative. That does not make the evidence real.
This is especially dangerous in history because historical arguments depend upon provenance.
Students should therefore be taught a firm distinction between discovery and verification.
AI can help generate search terms, identify possible lines of inquiry, formulate questions, explain unfamiliar terminology, or suggest places to begin looking.
It should not be treated as proof that a source exists.
If an AI system says that Frederick Douglass wrote something, the student's work has only begun. Find the speech, newspaper, letter, or publication. Determine its date. Read the surrounding passage. Establish context. Cite the actual source.
The National Archives' document-analysis framework reinforces precisely this kind of inquiry by asking students to determine who created a document, who received it, when and where it originated, why it was created, what historical circumstances surrounded it, and what additional evidence is necessary to understand it.⁷
That is historical research.
AI output is not a substitute for it.
AI Can Make a Good Tutor
Responsible use does not mean rejecting the technology.
There are valuable applications.
A college student struggling to understand the difference between primary and secondary sources could ask AI for additional examples before completing an instructor-designed exercise.
A high school student could use AI to practice identifying thesis statements.
A homeschool student studying the Constitutional Convention could ask an AI tutor to quiz them on the Virginia Plan, New Jersey Plan, Great Compromise, and Three-Fifths Clause, then verify uncertain answers against assigned materials.
Students might also ask AI to:
- generate practice questions from their own notes;
- explain difficult vocabulary;
- create timelines that students subsequently verify;
- compare two student-supplied documents;
- simulate an oral review session;
- provide feedback on whether a thesis is specific and arguable;
- suggest research keywords;
- help identify questions requiring further investigation.
The critical distinction is that AI assists the learning process rather than replacing the student's intellectual labor.
UNESCO's guidance takes a similar human-centered approach. Its framework emphasizes ethical, safe, equitable, meaningful, and age-appropriate use while preserving human agency. UNESCO's student competency framework likewise emphasizes critical judgment, ethical responsibility, foundational AI knowledge, and students' development as responsible users rather than passive consumers of automated output.⁸
College Classrooms: Raise the Intellectual Bar
At the college level, banning AI from every intellectual activity may eventually become as impractical as pretending students do not have access to internet search engines.
That does not mean anything goes.
It means assignments must become more intentional.
A traditional prompt such as “Write 1,000 words about the causes of the Great Depression” is increasingly vulnerable to outsourcing because the assignment primarily asks for information synthesis.
A stronger assignment might provide students with speeches, economic statistics, newspaper articles, political cartoons, and competing scholarly interpretations. Students could be required to develop an argument using those materials, document their reasoning, and explain why particular evidence supports their conclusions.
AI might be permitted for brainstorming or revision but prohibited from generating the final argument.
An instructor could even require an AI disclosure statement:
I used ChatGPT to generate possible research questions and identify keywords for database searches. I independently located, read, and cited all sources. AI-generated prose was not incorporated into the submitted essay.
That teaches transparency instead of encouraging concealment.
The AHA specifically recommends concrete and transparent AI policies that tell students when AI is permitted, when it is prohibited, and how its use should be acknowledged.⁹
That approach protects academic integrity while recognizing technological reality.
High School: Teach the Skill Before the Shortcut
High school presents a somewhat different challenge.
Students are still developing the foundational skills AI can easily imitate: summarizing, outlining, constructing thesis statements, comparing evidence, and writing analytical paragraphs.
If AI performs those tasks too early, students may produce better-looking work while developing weaker abilities.
Teachers should therefore consider sequencing.
First, the student performs the skill.
Then, the student examines AI performing the same skill.
Finally, the student evaluates the difference.
For example, students might independently analyze Abraham Lincoln's Gettysburg Address. Only after submitting their initial observations would they see an AI analysis.
The assignment then becomes:
What did the AI notice that you missed? What did you notice that it missed? Which interpretation is better supported by the text?
Suddenly the machine is not replacing thinking.
It is provoking more thinking.
Homeschooling: AI as a Resource, Not the Teacher of Record
AI offers particularly interesting possibilities for homeschool families because a parent may be teaching several subjects and may not have specialized training in every historical period.
AI can help parents generate age-appropriate questions, vocabulary exercises, timelines, discussion prompts, quizzes, reading guides, and ideas for primary-source activities.
But homeschool families should follow the same verification principles as professional educators.
Use trusted collections as anchors.
The Library of Congress provides digitized primary sources and analysis tools suitable for remote learning, including an Observe-Reflect-Question process that can be used online.¹⁰ The National Archives likewise offers DocsTeach, document-analysis materials, milestone documents, and other freely available resources.
A homeschool lesson on westward expansion, for example, could begin with a map, emigrant diary, government document, Native account, photograph, or newspaper.
The student examines the evidence first.
AI comes afterward.
That sequence matters.
Evidence before algorithm.
Teach Competing Interpretations Without Teaching Relativism
One of history education's greatest responsibilities is helping students understand that interpretation is unavoidable without teaching them that every interpretation is equally valid.
My America250 work emphasizes this point. Historical traditions can ask different questions of the same past. One interpretation might emphasize constitutional institutions, political leadership, and national ideals. Another might foreground workers, enslaved people, Native Americans, women, immigrants, social movements, and people excluded from political power. The educator's responsibility is not to require students to choose the approved interpretation. Students should learn to compare interpretations against evidence, context, and historical reasoning.
AI can help demonstrate this process if carefully controlled.
Ask students to compare two real historians first.
Then ask AI to summarize their disagreement.
Then return to the historians.
Did the machine represent them fairly?
That final step is essential.
The historian remains the source.
The AI remains the tool.
History Education Should Produce Informed Skeptics
The most important contribution history education may make in the AI era has little to do with technology.
It is disciplined skepticism.
Not cynicism.
Not assuming everything is false.
Rather, students should develop the habit of asking:
How do we know?
Where did this claim originate?
What evidence supports it?
What evidence challenges it?
Who created the source?
For what audience?
Under what circumstances?
What changed?
What remained the same?
What caused this?
What alternatives existed?
Why do historians disagree?
Those questions belong in college classrooms, high school classrooms, and homeschool dining rooms alike.
They are also exactly the questions citizens need when navigating an increasingly complicated digital information environment.
The Teacher Still Matters
The emergence of AI does not diminish the importance of teachers.
It clarifies their purpose.
A teacher's greatest value was never possessing information that students could not obtain elsewhere.
Teachers create intellectual environments. They select meaningful evidence. They recognize misconceptions. They ask the follow-up question a student did not anticipate. They understand when confusion represents intellectual growth and when it represents genuine misunderstanding. They connect individual students to a larger community of inquiry.
UNESCO's recent work on AI and teachers similarly emphasizes that educators remain fundamental even as AI alters teaching practices.¹¹
History teachers have another responsibility as well.
We teach students to encounter people who lived in worlds profoundly different from their own.
That requires imagination disciplined by evidence.
No algorithm should replace that encounter.
A Simple Rule for the History Classroom
If I had to reduce responsible AI use in history education to one principle, it would be this:
Use AI wherever it increases inquiry. Restrict it wherever it replaces inquiry.
Let AI generate questions.
Do not let it replace the student's answer before the student has learned to construct one.
Let AI help identify possibilities.
Do not confuse possibilities with evidence.
Let AI assist revision.
Do not let it erase authorship.
Let students interrogate AI.
Do not teach them to obey it.
And above all, keep bringing them back to the sources.
The goal of history education has never been to provide students with a finished story. It is to give them the knowledge and habits of mind necessary to evaluate evidence, understand context, recognize complexity, compare interpretations, construct arguments, and participate thoughtfully in the world around them. That principle already sits at the center of my work on historical thinking and civic education.
Artificial intelligence does not make those skills obsolete.
It makes them more important.
The history classroom of the future should therefore be neither anti-technology nor technology-dependent.
It should remain what the best history classrooms have always been:
a place where evidence, questions, and context matter, and students learn to think rather than being told what to think.
Endnotes
- American Historical Association, “Guiding Principles for Artificial Intelligence in History Education,” approved July 29, 2025, published August 5, 2025.
- John Lewis Gaddis, The Landscape of History: How Historians Map the Past (New York: Oxford University Press, 2002). This work also forms part of the historical-thinking framework developed in Buie's America250 education research.
- National Archives and Records Administration, “Teaching With Documents,” last reviewed January 4, 2023.
- Library of Congress, “Teacher's Guides and Analysis Tool,” Getting Started with Primary Sources, accessed August 7, 2026.
- Digital Inquiry Group, “Getting Started with Reading Like a Historian Lessons,” accessed August 7, 2026.
- College Board, “AP United States History Course Audit,” accessed August 7, 2026.
- National Archives and Records Administration, “Analyze a Written Document,” last reviewed April 6, 2023.
- Fengchun Miao and Wayne Holmes, Guidance for Generative AI in Education and Research (Paris: UNESCO, 2023); Fengchun Miao, Kelly Shiohira, and Natalie Lao, AI Competency Framework for Students (Paris: UNESCO, 2024).
- American Historical Association, “Guiding Principles for Artificial Intelligence in History Education.”
- Library of Congress, “Distance Learning and Primary Sources,” Getting Started with Primary Sources, accessed August 7, 2026.
- UNESCO, “AI in the Classroom: Empowering Teachers for the Future of Learning,” October 17, 2025, updated March 31, 2026.
Bibliography
American Historical Association. “Guiding Principles for Artificial Intelligence in History Education.” Approved July 29, 2025. Published August 5, 2025.
Buie, David. “America250 and the Social Studies Classroom.” Unpublished History_Academix manuscript, 2026.
College Board. “AP United States History Course Audit.” Accessed August 7, 2026.
Digital Inquiry Group. “Getting Started with Reading Like a Historian Lessons.” Accessed August 7, 2026.
Gaddis, John Lewis. The Landscape of History: How Historians Map the Past. New York: Oxford University Press, 2002.
Library of Congress. “Distance Learning and Primary Sources.” Getting Started with Primary Sources. Accessed August 7, 2026.
Library of Congress. “Teacher's Guides and Analysis Tool.” Getting Started with Primary Sources. Accessed August 7, 2026.
Miao, Fengchun, and Wayne Holmes. Guidance for Generative AI in Education and Research. Paris: UNESCO, 2023.
Miao, Fengchun, Kelly Shiohira, and Natalie Lao. AI Competency Framework for Students. Paris: UNESCO, 2024.
National Archives and Records Administration. “Analyze a Written Document.” Last reviewed April 6, 2023.
National Archives and Records Administration. “Teaching With Documents.” Last reviewed January 4, 2023.
UNESCO. “AI in the Classroom: Empowering Teachers for the Future of Learning.” October 17, 2025. Updated March 31, 2026.
Comments
Post a Comment