Asking Better Questions in the Age of Generative AI

Each year, I journey forth with an intrepid group of students to tackle the fearsome beast known as the research paper. Faced with the certain knowledge that they must complete this challenge in order to advance, there are always one or two who attempt to cobble together an argument from conjecture and Wikipedia in the first 40 minutes of class. For these students, wading into the deep waters of research without a ready-made flotation device is terrifying. Any subsequent examination of sources that they do involves sussing out the facts that agree with a prepared truth. It’s a form of reverse engineering. The result is often a weak, poorly-executed argument that can be easily dismissed by an obvious counter, provided by the sources they did not consider.   

The step that these students miss when they jump to an answer is, outside of choosing a topic, the most important of the process. It is figuring out what question will guide their research. The job of the research question is two-fold.  It narrows the research parameters, while still providing room for discovery.  

A research question takes specific information and elevates it to the argumentative level by demanding analysis (e.g., a how or why). In what war did New York’s 369th regiment earn the nickname the Harlem Hellfighters? Not a good research question. It’s a closed question. The answer–World War I–is a fact. Once you state that fact, the question has been answered. 

A research question about the Harlem Hellfighters, on the other hand, might look like this: How did the Hellfighters’ military successes during World War I impact African Americans in Harlem in the early 20th century? It can’t be addressed with a yes or no, or basic fact. If it’s to be answered well, the student will need to consider what life was like for African Americans there before, during, and after the war. Using primary sources, they will need to track domestic responses to news of overseas military victories. Their thesis will emerge from their research and serve as an answer to the posed question.  

Still, it’s important to remember that research questions, which are a type of open question, are not the only ones that advance knowledge. Who, what, when and where are targeted closed questions that never go out of style. In fact, in a research question, they are the key to identifying the problem, limiting options, and focusing effort.

In the example presented above, we’re not looking at every war.  We’re looking at World War I. Nor are we considering every regiment. We’re examining New York’s 369th alone. Those deep waters of research that so often trouble students can be as vast as the oceans. Yet with the aid of a few closed questions and their answers, they can be reduced to the more manageable lake or pond.

But wait, there’s more. The information garnered by closed questions is the key to successful analysis and argumentation. What if we mistakenly believe that the Harlem referred to above is Haarlem, the city in the Netherlands, not Harlem, the neighborhood in New York City?

The analysis that follows will be faulty, because the context–that web of connections through which we filter information–is faulty. Yes, a student can always “look it up,” but only if they know to do so. It’s essential to ensure that they have a clear understanding of the fundamental, indeed, foundational information obtained through the use of closed questions. That, in turn, yields more successful, higher-order, critical-thinking. Closed questions shouldn’t be used exclusively, of course, but as I hope the examples here illustrate, they can and should be used to scaffold more complex reasoning.  

We stand at the forefront of a technological event on par with the First Industrial Revolution. We must teach students how and when to use these different types of questions in order to make better use of generative AI as a tool. The coming age doesn’t end the need for complex thinking. On the contrary, it demands a better understanding of how to elicit it, and more opportunities to practice it, among educators and students alike.  

What do you think?

Tisha Hooks, Ph.D., teaches history at Hopkins School in Connecticut.  She is a former member of the Teacher Advisory Council for the National Humanities Center.