Exercise #3: Generative AI and the Future of History

What Kind of Tool Is AI?

After the AI demo in class, I don’t think generative AI is just one type of tool. It is not only a search engine, but it is also not fully a co-author. To me, it feels most like a mix between a research assistant and a pattern-finding tool.

AI is different from a search engine because it does not just find websites or articles. Instead, it creates new text. When we asked it to summarize historiography or fix OCR mistakes, it did not give us links,

it gave us answers. That makes it feel like a research assistant. It can organize ideas, explain complicated topics in simpler ways, and help us see patterns across many texts.

In the corpus analysis activity, AI showed how quickly it can find themes in a large group of documents. That is something that would take a human a very long time. In that way, it works like a distant reading tool. It can scan a lot of material and point out repeated words, ideas, or trends.

But I do not think AI should replace historians in every task. There are things we should not ask AI to do. For example, historians carefully interpret sources. They think about bias, missing voices, and historical context. AI does not truly understand context, it predicts what words are likely to come next. When it summarizes historiography, it sounds confident, but it might miss important debates or misunderstand subtle arguments. That can be risky.

OCR repair seems like a helpful use of AI because it fixes messy text from scanned documents. However, we still need humans to check that nothing important was changed by mistake. Even small changes in wording can affect meaning in historical documents.

When it comes to who benefits from AI, I see both positives and concerns. AI can make research more accessible. Students can use it to understand difficult readings or get help starting a project. That can make learning easier and less intimidating. At the same time, these tools are owned by large technology companies. That means power and control are concentrated in a few places. Universities and researchers may depend on systems they do not fully control.

Overall, I see AI as a powerful support tool, but definitely not a replacement for historians. It can help us work faster, but it cannot take responsibility for interpretation.

Is this the Last Generation of Historians?

After reading Joshua Gans and Dr. Humphries, I kept thinking about one main question: if AI can now “do research,” what does that mean for historians?

Joshua Gans argues that AI challenges the way research currently works. Right now, scholars spend a lot of time writing articles and books that may or may not be used later. From what I heard, genuine research is slow It takes months or even years to complete a paper. Then it goes through peer review, gets published, and becomes part of the larger body of knowledge. Gans suggests that this system is based on the idea that research is expensive and time-consuming, and because it takes so long, we try to prepare knowledge in advance just in case someone needs it in the future.

But AI changes that idea, if AI can generate summaries, models, or explanations quickly, then maybe we do not need to prepare everything ahead of time. Maybe we can simply “research on demand.” Instead of spending years writing something that might be useful later, we could wait until a question comes up and then ask AI to produce an answer. That idea makes research faster and more flexible.

At first, this sounds easy and honestly a lot more appealing. It could make gathering information so much more efficient. It could reduce wasted effort. It might even allow more people to participate in research because the barriers are lower. But it also raises serious concerns. 

If knowledge can be produced instantly, what happens to traditional academic work? What happens to the value of expertise?

Dr. Humphries takes this concern even further. He describes Deep Research, an AI system that can search archives, collect sources, and write detailed responses that sound like they were written by a strong graduate student. That is surprising and honestly a little unsettling. If AI can now gather primary sources, analyze them, and produce a long explanation, then it is doing many of the tasks that historians are trained to do.

This leads to a difficult question: if AI can complete these tasks quickly and cheaply, will human historians still be needed?

However, Humphries also shows that AI still makes serious mistakes. In one example, the AI claimed to run advanced computer analysis, but when tested, the results were wrong. Some quotations were even made up. That is a major problem because historical research depends on trust. If AI invents evidence or misrepresents sources, it can create false knowledge. 

Even if the writing sounds confident and professional, it may not be reliable.

This makes me think that AI is powerful but not independent, it definitely still needs human supervision. It can gather information and produce drafts, but humans must check the evidence carefully.

I also think there are parts of history that AI cannot replace. Historians do more than find information, they also ask meaningful questions that AI simply cannot do. For example, when historians study marginalized communities, they are not just analyzing data, they are trying to recover voices that were ignored. That requires care and ethical awareness, and AI does not have empathy. It does not understand why certain stories are important to living communities.

Another important point is that AI is trained on existing data, that means it reflects the biases already present in the historical record. If archives already focus more on powerful groups, AI may repeat that imbalance. Without human historians to question those patterns, AI could strengthen existing inequalities in knowledge.

At the same time, I do not think we should reject AI completely. It offers real opportunities. It can scan thousands of documents quickly. It can help historians see patterns across long periods of time. It can support language translation and make difficult texts easier to understand. For large projects, AI could save time and energy. So instead of asking whether this is the last generation of historians, maybe the better question is: what kind of historians will we become? Historians may need to develop new skills. They may need to learn how to work with AI, how to test its accuracy, and how to guide its analysis. The role of the historian might shift from being only a researcher to also being a supervisor and critical evaluator of machine-generated work.

In the future, research might become a collaboration between humans and AI. AI could handle large-scale scanning and drafting. Humans would focus on interpretation, ethics, and deeper meaning. That balance will be important.

I do not believe this is the end of historians. But I do believe it is a turning point. The profession will likely change, just like many other jobs have changed because of technology. What will remain important is human judgment. AI may change the speed and tools of research. But it does not remove the need for thoughtful human interpretation, and that is why I think historians will still matter in the future.