A 5e student can recite that Julius Caesar crossed the Rubicon in 49 BC. They can also name the year of the Battle of Alesia, describe the structure of the Roman legions, and copy a sentence about the crisis of the Republic. What they generally cannot do is explain why Caesar took this risk, what exactly he was risking, what would have happened if he had refrained. The difference between these two types of knowledge is precisely what AI can help bridge — not by replacing the teacher, but by giving a voice to history.
The Problem of Memorized History
The teaching of history in middle and high school suffers from a structural paradox. Official programs set ambitious goals: understanding causality, analyzing viewpoints, exercising critical thinking. But practical constraints — thirty students, fifty-five minutes, a packed curriculum — often push towards frontal transmission. The teacher speaks, the students write, the lesson is printed in the notebook. What remains in memory after two weeks rarely resembles deep understanding.
This is not a criticism of teachers — it is a structural observation that cognitive science research has confirmed for forty years. Passive memorization produces inert knowledge: available for the next day's test, untraceable three months later. Active learning — where the student questions, doubts, reformulates, confronts — produces usable knowledge because it has been processed in depth, not just recorded.
According to research on long-term memory encoding, students retain about 10% of what they read and 20% of what they hear — compared to 70% of what they debate or teach to someone else. Conversation with a historical character engages exactly this active mechanism.
Conversational AI opens a third way between the lecture and traditional documentary work. It allows each student to directly question a historical character, confront their hypotheses with time-situated answers, and discover for themselves the contradictions and blind spots of a witness of the time. This is not gamification — it is active pedagogy with an interlocutor capable of maintaining a coherent historical position.
What Conversation Changes
When a student asks Julius Caesar a question, something happens that does not occur during a textbook reading: they must formulate. Formulating requires knowing what one already knows, identifying what one does not know, choosing where to tackle the problem. A student who asks Caesar "Why did you need the legions to do politics?" has already understood, even implicitly, that Rome is not a modern democracy and that the relationship between military power and civil power operates differently there. The question itself reveals an understanding that reciting dates would never show.
What conversation adds is also surprise. Caesar responds from 49 BC. He does not know he will be assassinated in five years. This temporal constraint — which Parlova calls the golden rule — fundamentally changes the nature of the exchange. The student who asks "Are you afraid of being killed?" receives a response that reveals not the character's naivety, but the coherent logic of a man convinced of his legitimacy. It is precisely this gap between what Caesar believes and what the student knows that generates historical understanding.
A Concrete Example: Julius Caesar in 5e
Here is how a typical session is structured in a 5e class. In the introduction (ten minutes), the teacher recalls the context of 49 BC: Caesar returns from Gaul, he is in conflict with the Senate, crossing the Rubicon is imminent. They distribute a short character sheet — no more than one page — that situates Caesar in his own logic, without revealing the rest of the story. The students know who Caesar is from the program; the sheet gives them his point of view.
For twenty minutes, students work in pairs on Parlova and ask their questions to Caesar. Some groups choose the questions suggested in the pedagogical sheet (available on the site). Others formulate their own. What always surprises teachers testing this approach for the first time is the quality of spontaneous questions: "Would your soldiers obey you if you ordered them to turn against Rome?", "Do you really think you are defending the Republic or are you defending yourself?".
This type of exchange — which the teacher can project and comment on in the whole class — generates exactly what the program demands: the analysis of a historical actor's discourse, the detection of their justifications, the highlighting of their contradictions. The collective synthesis (ten minutes) naturally arises from what the groups have gathered: what does Caesar say? What does he omit? What does he not know?
Five Principles for Integrating AI in History Classes
- 1Anchor the character in a specific moment. Caesar in 49 BC, Napoleon in 1799, Freud in 1900. The temporal anchoring prevents the character from doing the philosophy of history — they speak of what they live, not of what will follow.
- 2Prepare students for dramatic irony. Students know things the character does not. It is precisely this gap that creates the pedagogical tension. Indicate it explicitly before the session.
- 3Let students formulate their own questions. Suggested questions are starting points. Encourage students to reformulate them, deepen them, invent new ones. An awkward question often reveals more than a beautiful answer.
- 4Organize a collective synthesis after the dialogue. The conversation alone is not enough. It is in the debriefing — confronting the answers with known historical facts — that understanding is consolidated.
- 5Keep a written record. Ask students to identify a response from the character they find convincing and one they find questionable. This record anchors the work in the school framework and prepares for essays or presentations.
What AI Does Not Replace
Conversational AI is not an infallible encyclopedia nor a guarantor of historical accuracy. Parlova's characters are trained to respond coherently with their era, but they can make mistakes, simplify, or reflect the biases of their own position. This is precisely why the presence of the teacher remains indispensable — not to control every exchange, but to contextualize, correct, and deepen.
AI also does not replace work on primary sources. A letter from Caesar, a text by Plutarch, a map of Roman conquests retain an irreplaceable pedagogical value. What conversation adds is engagement — the motivation to read Suetonius because one has just had a discussion with Caesar and wants to verify if what he said corresponds to what the sources report.
Always specify to students that the character speaks from their time and does not know the rest of the story. This rule avoids confusion between what Caesar says on Parlova and what historiography establishes. The boundary between pedagogical simulation and historical knowledge must remain explicit.
Where to Start
The best way to start is to choose a character you know well and a concept your students struggle to grasp. Julius Caesar to understand the crisis of the Roman Republic. Napoleon to untangle the French Revolution from the Empire. Zeus to anchor mythology in a societal logic. Freud to make the unconscious tangible rather than abstract. The character you master allows you to anticipate mistakes, correct simplifications, and guide the debrief with authority.
Begin with a short session (twenty minutes of dialogue maximum) with guided questions. Observe what the students ask. The first few times, some groups will test the limits — asking Caesar if he has a phone, or if Cleopatra was his girlfriend. It's not a problem: Caesar responds within his historical logic, which quickly defuses these attempts and naturally brings the focus back to historical questions. After two or three sessions, students have integrated the framework and formulate questions of a quality that often surprises even the most experienced teachers.
All you need is a free teacher account. Your students join the mission with a link and a password.
Create my account →