Artificial intelligence is rapidly revolutionizing the field of history, offering historians both powerful new tools and unprecedented dilemmas, according to recent research and expert testimony reported by The New York Times (nytimes.com). As large language models (LLMs) like Google’s NotebookLM and OpenAI’s ChatGPT gain traction, scholars are reckoning with the profound changes AI might bring to the process of researching, organizing, and ultimately narrating our shared past—a phenomenon with significance for educators, students, and even policymakers in Thailand.
The driving force behind this transformation is AI’s ability to process vast troves of digitized material—ranging from hand-written historical records to books and academic papers—in minutes rather than months or years. For professional historians, such as those at Stanford University or Wilfrid Laurier University in Canada, the technology promises much greater research efficiency, and even creativity, by surfacing new connections or suggesting innovative narrative structures. Yet this very power also raises urgent questions about accuracy, context, and the historian’s evolving role.
One of the clearest examples described in the report involves a historian experimenting with Google’s NotebookLM. By feeding the AI software surveys and texts from the California Gold Rush era—including accounts from both settlers and Native Americans—the historian could rapidly compare perspectives, uncover overlooked figures like Maria Lebrado (a key Yosemite Valley Native American presence), and generate vivid biographical sketches in a fraction of the time traditional research would take. “Everything I’ve just showed you is, like, 30 minutes of work,” the historian marveled, underlining the leap in efficiency.
Stanford communications faculty have taken a similar approach, employing AI not only to organize their massive research outlines but also to press the AI for feedback on structure, clarity, and narrative approach—effectively using it as a “middlebrow reader.” AI-generated suggestions often help clarify which storylines resonate with non-specialist audiences without diminishing scholarly rigor.
Canada-based researchers, meanwhile, have deployed AI to analyze tens of thousands of hand-written fur trading records. The models identify complex webs of relationships and trading networks, demonstrating analytic feats “that would take a graduate student weeks, but the AI can do in 20 seconds,” according to a university expert.
But even as these advances unlock new levels of productivity and enable scholars to tackle previously daunting projects, experts stress that AI’s accuracy problems can undermine its promise. The technology’s dream of perfect synthesis runs aground on persistent “hallucination”—the tendency to fabricate plausible but false details. A 2025 review cited in The New York Times found OpenAI’s latest model still produced inaccurate outputs up to 33 percent of the time—double that of its predecessor. Scholars warn that AI currently lacks the “bullshit detector” of a good human editor, who might spot flaws in logic or subtle mis-readings of context (nytimes.com).
Many experts also highlight institutional and cultural obstacles. The historian community—as well as the broader education sector—remains deeply ambivalent, torn between the seductive possibility of AI-augmented research and the fear that such tools might erode scholarly skills or put human expertise at risk. Some academics expressed concerns about “hypocrisy,” noting that while they worry about student cheating via AI, they themselves are tempted to use the same tools for professional work.
Reflective commentary from historians points to a broader issue: the risk that digital search, and now AI, shift history-writing away from place-specific labor, nuances, and “serendipitous” discovery. The growing dominance of digitized and English-language sources accentuates systemic biases; non-digitized, local, or non-official histories may be overlooked, rendering some communities or perspectives invisible. This is of particular importance to countries like Thailand, where digitization is ongoing and national historical narratives rely heavily on official, institutionally preserved documents, often at the expense of marginalized or regional voices.
Historian Lara Putnam, for example, warns that “gazing at the past through the lens of the digitizable makes certain phenomena prominent and others less so.” The more quickly researchers move online, the riskier it becomes that the “expense” of fieldwork or archive visits pushes out traditional historical craftsmanship, even though such labor may be essential for well-rounded understanding.
Still, AI’s potential to “raise the ceiling” in creativity and analysis is hard to ignore. As a Google Labs executive put it, these tools lower barriers for new creators and offer established scholars new vistas to explore “so many more stories, so fast.” Moreover, new features such as automatic mind maps and podcast-style conversational summaries promise to democratize historical access and bring complex topics into the reach of much broader audiences.
For Thai society, these developments carry a clear and immediate significance. Thailand’s education system is increasingly incorporating digital platforms and AI-powered tools, both to support students and to streamline research by educators and academic historians (UNESCO Bangkok). The risk of bias towards English-language and official sources, as highlighted in global research, is mirrored here; local languages and oral traditions, essential to understanding the diversity of Thai society—from the northern Lanna culture to the traditions of the Deep South—may be under-represented or misrepresented unless specific attention is paid to digitizing and encoding them into future AI systems.
Moreover, as Thailand grapples with debates about history curriculum reform, AI offers both promise and peril. On the one hand, AI could make it easier for teachers to assemble, compare, and present multiple narratives—including local or regional histories previously underrepresented in national curricula. On the other, the opaque nature of AI’s content selection and synthesis could unintentionally reinforce official biases or oversimplify the diversity of historical experience. As a Thai educational policymaker might note, leveraging AI needs to be coupled with new teacher training, digital literacy, and safeguards that empower educators to interpret and contextualize automatically generated content.
Historically, Thai society values the notion that history is more than a record of events; it is also a tool for nation-building and moral education. The way AI tools select, synthesize, and “retell” stories, then, could shape collective memory in subtle but consequential ways. If institutional actors become overly dependent on AI-generated output, there is a risk that controversial or inconvenient topics—such as the experiences of ethnic minorities, borderland communities, or moments of political upheaval—could be filtered out or neglected, echoing concerns already seen in Western scholarship.
A key recommendation emerging from this research is the need for hybrid approaches: using the speed and breadth of AI for synthesis and first-pass exploration, but always combining it with expert human judgment and rigorous fact-checking. Thai universities and cultural institutions might consider building local-language archives and maintaining investment in physical and oral history, even as they develop digital and AI-driven initiatives. Thai historians may also need to consider new collaborations with technologists to develop culturally sensitive AI models, ensuring that regional diversity and non-digitized knowledge are made present and powerful within next-generation historical work.
Looking ahead, the influence of AI in history-writing is likely to increase, especially as tools become more sophisticated, collaborative, and embedded in educational practice. The future may see ebooks and online textbooks “bundled” with AI-driven guides, podcasts, or interactive maps tailored to students’ interests and backgrounds. For those shaping Thailand’s future educational and cultural strategies, the message is clear: AI is not a replacement for human expertise but, handled wisely, it could be a transformative partner in making history both more democratic and more inclusive.
For Thai readers, educators, and students, the call to action is to become critically engaged with these new tools. Ask: Whose voices are missing from digital archives? How transparent are the algorithms shaping historical synthesis? How can you combine AI-driven summaries with your own research or local knowledge? Consider helping local archives digitize unique collections or supporting oral history initiatives in your community. Embrace AI as a research assistant—but insist that it serves, not supplants, the depth and diversity of Thailand’s multifaceted historical tradition.
Sources: The New York Times, UNESCO Bangkok, Stanford University, Wilfrid Laurier University.