EUROCLIO published an article “Bridging the Past and Future: The Impact of AI on History Education – A Journey into AI-Enhanced History Teaching” and I have some comments.
What we call simulating conversations with people from the past is actually maximum fiction. We do not talk to people from the past, but only the chatbot reproduces the materials on the basis of which it was programmed. And then there is the issue of data, i.e. on the basis of which primary, secondary and even tertiary sources it was programmed. How broad or limited are these sources, whether they were only from one available archive or several, whether they only included one language (most often English) or others, etc.
Simulating conversations using AI is just playing with history. Talking to AI Napoleon is just scratching the surface of historical texts and not talking to an actual historical person. There is also the example of Napoleon and it is well supported because it is about the ruler and military leader who was written about. But if you try, say, to “talk” to a slave on a southern plantation in the early 19th century, what will you get? Or with a medieval serf in France, etc. We can look at it as an attempt to stage historical figures so that the new “TikTok generations” have minimal additional motivation.
