Historian warns AI may overwrite history by missing human suffering in testimonies

Human historians are increasingly vital in the AI ​​era, in particular with the crucial need to grasp the emotional and moral complexity behind global events.

It is according to a leading academic, Dr Jan Burzlaff, an expert from Nazi Germany from Cornell University, who, when Tasking Chatgpt to summarize the experiences of the Holocaust survivors, found that the AI ​​tool had failed to capture intimate and vital details.

“With the testimony of Luisa D., a survivor of the 7 -year -old holocaust, the AI ​​has neglected heartbreaking details on her mother cutting her own finger to give him drunk blood -” the lowest trace of humidity ” – to stay alive.

“This omission alone shows why human historians remain essential in the era of artificial intelligence.

“Historical writers have skills that are currently lacking, in particular the ability to capture human suffering,” said Dr. Burzlaff, a postdoctoral partner on the Jewish Studies program at the College of Arts and Sciences. “If the historical writing can be carried out by a machine, then it has never been historical enough.”

Its results are published today in the journal Rethink the storyIn a play that analyzes Chat GPT attempts to summarize the recorded testimonies of Holocaust survivors made to La Paz, Krakow and Connecticut in 1995.

The results expose the limits of AI, which creates new content according to what it learns from existing data. He stresses that if AI can identify the angles that historians may not have taken into account, the drawback is that algorithms can distort history or, as in this case, try to clarify the holocaust which, according to him, cannot be resolved “.

“He essentially ignored the measure where these people suffered at the emotional level,” said Dr. Burzlaff.

“A recent Microsoft study has classified historians as on the list of jobs that AI could replace. But AI does not have the capacity to capture human suffering.

“If he vacillates testimonials from the holocaust – the most extreme case of human suffering in modern history – it will also take more subtle stories.

He adds: “Like tools like Chatgpt Satuching more and more about education, research and public discourse, historians must count with what these systems can and cannot do.

“They summarize but do not listen, do not reproduce but do not interpret and do not exceed coherence but weaken to contradiction.

“The problem we have now encountered historians is not whether AI can recognize the meaning, but if we will continue to do so.”

The article divides five directives developed for teachers, academics and any other person who writes on history in the modern era, in particular for those who teach trauma, genocide and historical injustice.

The author says that his advice will help historians keep “ethical, intellectual and stylistic issues of historical writing”.

“The AI ​​feeds on patterns, frequency and proximity. Historians should avoid this approach – they should draw written testimonies, and not become a collection of texts,” describes Dr. Burzlaff.

“Essentially, as Historians, we must not try to” surpass the machine “but do not look like anything.

“The stake is not only the memory of the holocaust, as in this case, but on the way in which societies everywhere will remember and interpret their past in the era of prediction.

“The accounts of people from the past differ according to their individual experiences and some are different to categorize. Historians must embrace this lack of uniformity and moments of human experience that algorithms cannot anticipate.”