As an aspiring public historian, it is difficult to make solid sense about the role of artificial intelligence beyond what has already been happening over the course of the past few years – namely, an unprecedented potential it brings. On the surface, my preconceived notions about AI have not particularly changed because my chosen fields, History and English, have been at the forefront of AI abuse which has caused me to regard it more negatively. However, I have never needed to utilize it for anything like what the readings mention, so while I do not discount the merits of certain tasks that can be allotted to an automated search engine, it is still difficult to focus on those aspects when the rest of the Pandora’s Box has been so fraught with controversy. Cameron Blevins’ example of using AI to undertake transcription processes illustrates both sides of my reservations: on one hand, the model’s ability to create a mostly-accurate transcription of the text is valuable, but on the other hand, the fact that every instance of automated transcription needs to be overseen and verified by an actual expert or historian seems to defeat much of the purpose of automating it in the first place. I would certainly trust the end result much less if it was done quickly through these tools. Moreover, I take issue with Blevins’ characterization of AI as a “research assistant,” not just because of the poor choice of humanizing an algorithm (when there is still so much misconceptions about how LLMs work), but because it impedes too much on what should be, in my opinion, the main human element in public history: bringing historical research and engagement into greater public access. The more historians rely on shaky automation, the less effective they will be in communicating that information to the public. Adina Langer’s point about “self-reinforcing” inaccuracy should AI models train on previous uses of AI remains extremely salient given how the technology is still a largely unregulated frontier, and in regards to the role of public historians, I agree with her position that it is an important responsibility to “extend and enhance the ways we educate our publics about information literacy” – ceding the historiographical process to unreliable automation will not help mend decreasing levels of reading comprehension and increasing levels of misinformation. I do not mean to be completely pessimistic about the use of new technology in public history, as the field certainly has new possibilities to explore with new tools and resources, but such means need to be carefully employed and evaluated on their merits and their consequences. Uses such as the ‘historical’ chatbots that Rob Nelson criticizes in his AI Log blog appear a horrendous misuse of LLM technology because of how much they dilute truth and human identity through its fundamental inability to understand context and (authentic) speech. Uses such as more advanced archival search systems and features, such as those Alex Houston mentions, seem more appropriate and measured for this technology, so long as it is reasonably implemented.

 

Asking ChatGPT to take those same articles and provide the same commentary (provided in full below), it came up with a substantial response that focused on the primary concerns shared across the sources as well as my own commentary, namely reliability and transparency. After asking it to write it without using bullet points, it came up with a five paragraph essay which ironically matched ‘s point that current AI models are a natural outgrowth of the “bureaucratic systems of schooling” that revolve around measured formats like the five-paragraph essay and multiple-choice exams. The rest of the AI’s commentary largely matches this description, providing a sanitary summation of the source’s overall consensus that AI cannot be used to replace human oversight, despite there being a few opportunities for AI to be used responsibly in public history, like facilitating a more inclusive historical record. It does bring up more examples of it being used irresponsibly, such as exacerbating existing public and academic biases and, obviously, spreading layers of misinformation upon itself. Regardless, ChatGPT’s answer does not deviate too much from my takeaways, except perhaps in its own positivity in itself.

 

Sources:

Adina Langer, “Generative AI and historical authority,” History@Work, October 19, 2023

Cameron Blevins, “A Large Language Model Walks Into an Archive,” October 25, 2024

Alex Houston, “Bringing hidden histories to light: An archivist reflects on AI, archives, and the future of digital stewardship,” JSTOR

Rob Nelson, “Banal impersonations and historical reenactment on the cheap,” AI Log, August 20, 2024

 

Full ChatGPT Response:

Taken together, these articles suggest that AI will not replace public history, but it will fundamentally reshape how it is practiced, especially in terms of access, authority, and interpretation. One of the clearest shifts is AI’s ability to expand access to historical materials at an unprecedented scale. Tools like large language models can transcribe, summarize, and analyze vast collections of documents, allowing historians and institutions to surface materials that were previously difficult to use or entirely hidden. This has important implications for public history, which has long been concerned with making the past accessible to broader audiences. In this sense, AI strengthens an existing goal of the field by enabling deeper and wider engagement with historical sources.

At the same time, the readings make clear that this expansion of access comes with a destabilization of historical authority. AI systems can produce highly convincing narratives that are partially or entirely incorrect, and they often blur the distinction between primary sources, interpretation, and fabrication. For public historians, who frequently present history in accessible and authoritative ways to general audiences, this creates a serious challenge. Authority can no longer rest simply on presenting information; it must be built through transparency about sources, methods, and the limits of AI-generated content. In an environment where misinformation can be generated at scale, trust becomes a central concern.

Rather than functioning as independent historians, AI tools are better understood as collaborators that require careful oversight. The articles consistently emphasize that these systems are most useful when treated as assistants that can help process information but not replace human judgment. They remain limited in their ability to interpret context, nuance, and historical significance. As a result, public history is likely to become a more hybrid practice in which AI handles large-scale data processing while human historians provide interpretation, ethical framing, and narrative coherence.

Finally, the readings point to the emergence of new opportunities alongside new ethical responsibilities. AI can help uncover and amplify marginalized or overlooked histories, contributing to a more inclusive public historical record. However, it can also reproduce existing biases in archival collections and training data, or flood the public sphere with plausible but misleading historical content. This tension means that public historians will increasingly act not just as interpreters of the past, but as curators of authenticity and mediators of technology. They will need to guide audiences in understanding how historical knowledge is produced, especially when AI is involved.

Overall, AI is pushing public history away from a model centered on simply presenting narratives and toward one focused on navigating, evaluating, and explaining an overwhelming abundance of historical information. The role of the public historian will shift accordingly, emphasizing critical engagement, transparency, and trust in an increasingly AI-mediated landscape.