“Axios last month published a list of "The 10 jobs least and most threatened by AI." No. 2 on the threatened list, between translators and passenger attendants, was "historians." The source is a study by researchers at Microsoft on the implications of generative artificial intelligence on the labor force. The study uses data from Copilot, Microsoft's competitor to ChatGPT, to match user requests with relevant career paths. In doing so, it purports to demonstrate the level to which work activities in certain careers are "covered" by generative AI.
To their credit, the researchers are modest about their findings. They note that in the past, efficiency tools have often tended to increase the value of work, not decrease it. Whether generative AI will have a similar effect is too early to tell.
The many news outlets that have publicized the study have been less scrupulous. Almost every popular news report on the Microsoft study claims that it shows which jobs are most "vulnerable" or "threatened" by AI -- even though those words never appear in the original paper.
While the study seeks to determine the level of overlap between career tasks and generative AI, it doesn't purport to measure how large the gulf is between that "coverage" and professional-level mastery of a subject. In some of the careers with the highest AI "coverage," the gulf is wide indeed.
That explains the appearance of history, along with political science, journalism and writing -- fields that lean heavily on humanity's creative spark -- in the study's top 20. All these disciplines are practiced at a variety of levels of depth: by professionals, students and inquisitive amateurs. While for professionals generative AI is merely a slight improvement over a Google search, it can more easily satisfy a curious mind or create the appearance of familiarity for a passable term paper.
This is seen with the profession of history, which the study declared -- out of hundreds of professions in America -- to be the second most "covered" by AI, at 91%. The study found that Copilot successfully completed "historians' " requests 85% of the time.
According to the Bureau of Labor Statistics data cited in the study, more than 3,000 historians work in the U.S. As a recent history graduate, I'd bet that few of the AI requests captured in the study came from any of them. More likely, they came from young people seeking a shortcut on a research paper, research assistants preparing a precis for a policymaker, or tourists hoping to learn more about Gen. George Meade before a trip to Gettysburg -- those for whom a repackaging of received wisdom on a given subject is good enough.
The essence of the professional historian is the opposite. He exists to challenge received wisdom, whether by asking a question no one has thought to ask, finding evidence to topple ancient orthodoxies, or casting old evidence in a new light. That spark of creativity and individual genius resides not in the summary of existing information but within the mind of the practitioner.
That is why past leaps forward in research efficiency -- such as online databases -- haven't stunted the historical profession but caused it to grow. It's the reason, for professional historians, that generative AI will be another useful tool for finding and summarizing literature, not a threat. Whatever the gulf between "coverage" and mastery, it won't be bridged any time soon.
One could make a similar argument for the other humanities or social-science professions. Amateurs or lazy professionals might use generative AI for convenience, or to write slapdash articles such as those claiming that historians' or journalists' jobs are "vulnerable." Perhaps AI will replace the writers of such articles. Perhaps it should. But serious practitioners in those fields have little to fear from AI's advance.
The Microsoft study should caution us about drawing bright lines between what sorts of tasks AI platforms can do and what human work they can replace. In some fields, the human spark burns far brighter, and is far harder to quantify or replicate, than in others.
---
Mr. Masko is a freelance writer based in Boston.” [1]
1. AI Won't Replace Historians. Masko, John. Wall Street Journal, Eastern edition; New York, N.Y.. 18 Aug 2025: A17.
Komentarų nėra:
Rašyti komentarą