During an interview with Wired, Professor Benjamin Breen of the University of California, Santa Cruz, described how he tasked OpenAI’s latest model, o1, with analyzing an 18th-century Mexican medical manuscript. To his surprise, the AI not only transcribed the text flawlessly in colonial Spanish but also identified graphic details that had previously escaped his trained eye.
The remarkable capability raises pressing questions about the role of historians in the age of artificial intelligence. With AI demonstrating proficiency in transcribing and analyzing historical documents, scholars are debating whether traditional skills—such as working in archives, deciphering manuscripts, and mastering ancient languages—will remain essential.
“I absolutely believe historians still need to engage in field research and archival work,” Breen said. “But these AI models have shown me that we are far ahead of where we typically assume, and in ways we hadn’t anticipated.”
Breen highlighted the sophistication of modern AI with a striking example: OpenAI’s GPT-4 and Anthropic’s Claude Sonnet can analyze multilingual texts containing Renaissance Latin interwoven with Hebrew—something even a skilled Latinist might struggle to interpret. AI is now capable of parsing a wide range of historical languages, including those used by Jesuit missionaries in the New World, written in highly erudite Latin. “These systems can analyze texts across multiple languages simultaneously, detect hidden patterns, and suggest relevant bibliographies with a level of precision that seemed impossible until recently,” Breen explained.
Yet AI has its limits. According to Breen, these systems tend to converge on conventional interpretations, producing analyses that, while technically sound, lack originality. “It’s like having a brilliant student who can recite lectures flawlessly but struggles to generate truly innovative ideas,” he said. This, he argues, underscores AI’s fundamental limitation: a lack of creative thought.
The intersection of AI and creativity has also been a focal point for Bill Gates. In a recent interview on the Italian television program Che tempo che fa, Gates speculated that most job roles could eventually be automated, with a few key exceptions: programmers, biologists, and energy experts. These fields, he explained, require human oversight to guide and advance AI systems.
Gates also addressed AI’s inability to fully replace creative professionals. While AI can generate logos or design promotional materials based on text prompts, it struggles with deeper creative processes, such as forming a brand identity or crafting narratives that have never been told before. “AI can’t build relationships with brands or find new ways to connect with audiences,” Gates said. “These are critical elements of many creative roles. While AI can assist in execution, it lacks the strategic insight and originality that define human creativity.”
Ultimately, the responsibility for shaping AI’s role in society falls to humans. “My concern is ensuring we handle this correctly,” Gates concluded. The challenge ahead lies not only in harnessing AI’s potential but also in preserving the uniquely human qualities that machines cannot replicate.
