AI, History and Historians Video

An online discussion and questions, hosted by the RHS, for historians to discuss the challenges and opportunities of AI and GenAI.

[paularblaster]

Matthew Jones mentions "slides" a couple of times. Can those be viewed somewhere?

[fabiodeoliveiraribeiro]

The further we go back in time, the greater the possibility of science merging with fiction and history becoming literature. But this phenomenon paradoxically also occurs in relation to the very recent past. Not because there are no reliable written sources, but because there are so many sources and the documentation is so vast that the historian is forced to make selections, excluding an immense amount of information from his analysis.
Some people imagine that Artificial Intelligence will bring about a revolution in archaeological and historical studies. But I have doubts, because it is easier for an AI to spew anachronistic hallucinations than for a trained human to make serious errors when applying scientific methods to analyze sources from the past. And anyway, when a historian makes a serious mistake others can correct it, but mistakes made by one AI will be treated as information by other AIs in the future. Carefully purging outdated interpretations and theories when writing new history and archeology books is something human experts must do if they want to gain recognition in their field. Can this be programmed into an AI? I do not think so.
Another aspect that deserves attention here is the illusion that everything can be reduced to data and that analyzing the largest amount of data is equivalent to producing the best knowledge. I have my doubts about this, because it is a well-known fact that historians must be careful with data mentioned by ancient writers such as Herodotus, Livy, Thucydides, Suetonius, Flavius ​​Josephus, Polybius and others. An AI will never have the same sensitivity, care and distrust as a human historian. A machine cannot interpret some data with the necessary skepticism, paying greater attention to the context in which certain data were collected and the possible motivations that may have contaminated their collection and use.
Making history is filling in gaps, that is, trying to perceive data that were not recorded or were recorded in a suspicious manner to differentiate possible, plausible and improbable situations based on the interpretation of ancient texts and data. With education and training, a human can learn to do this well. I do not believe that an AI can imitate this type of human activity.