AI is everywhere now. You can ask it to write a speech, find evidence, summarize a 40-page study, and spit out a rebuttal in seconds. It’s fast and convenient, yet not perfect. This leads to what a lot of people are missing: being able to generate an argument is not the same as being able to make one.
When you first start using AI for anything, it feels like the best shortcut ever created. Anything you think of, type it into a generative AI server and it is done within seconds. However, as you pay attention, you realize something’s off. The speech that AI just pushed out sounds hollow. You can’t answer questions about your own evidence during crossfire. And when an opponent pushes back, you aren’t sure what to say. Welcome to the details of what we call the “best” shortcut.
The Speech That Sounds Great and Says Nothing
During my conversation with Potomac Debate Academy’s Director of Debate, Matt Stannard, we discussed how AI is changing debate. One question I was really curious about was whether judges can tell when someone is reading an AI-written speech. According to Coach Matt, judges almost universally say they can tell. After hearing enough rounds, they start recognizing the patterns. AI tends to write with the same polished tone every time. The wording is clean, everything flows nicely, but it doesn’t sound like a student. More importantly, it doesn’t sound like someone who actually knows the argument. The moment a problem comes up that isn’t expected, the students’ confidence starts to disappear. There’s a certain flatness to an AI-generated speech nobody actually wrote. The biggest thing is there is no person behind the words. Judges who’ve heard thousands of rounds can immediately tell as it is less about the argument being wrong, but there’s no conviction that the speaker actually understands what they’re saying.
When a student outsources their thinking, they lose the entire point of the activity, which is to build an argument yourself and defend it under any circumstance with critical thinking.
What Students Give Up
We also discussed the versatility of the skills students miss out on. Things like learning how to break apart an argument, thinking through evidence instead of just collecting it, or responding when someone challenges your case. Those are skills you only develop by practicing them yourself. And one thing AI isn’t able to do, is to allow the user to think on their own feet. Some of the most important aspects we talked about that students need to know is the following:
The ability to construct an argument. Which is knowing how to put together a claim, warrant, and impact (CWI) for any argument.
Genuine understanding of evidence. If AI found it and summarized it, then it is difficult to explain it, contextualize it, or defend it when someone challenges the methodology.
Confidence under fire. Students’ confidence comes from knowing your case like the back of their hand. Borrowed words crack the second an opponent asks a question you didn’t anticipate.
Adaptability. Last but not least, rounds are always unpredictable. Debate demands improvisation, whether it be during crossfire, summary etc. Having a solid understanding of your arguments, evidence and reasoning is necessary in order to succeed.
Ideas Many Weren’t Meant To Find
One of the most interesting points Coach Matt brought up had nothing to do with writing speeches, but research itself. He talked about something called serendipitous research, the idea that some of the best learning happens when you stumble across information you weren’t even looking for.
Think about researching a debate topic on your own. Maybe you start looking into election reform, but halfway through reading an article you find a completely different study that changes how you think about voter turnout or campaign finance. Suddenly, your argument shifts because you discovered something unexpected.
Through research, we realized that is one of the biggest things students lose when they rely too heavily on AI. Large language models are designed as prediction machines to give you the most likely answers, based on what others have put into the model as well. This includes the most common arguments, and the safest responses. They’re great at pointing you toward the middle of a topic, but they can’t recreate the experience of exploring a topic yourself and finding an idea that completely changes your perspective.
Potomac Debate Academy’s Philosophy Towards AI
Here’s where it gets interesting, though, as we also discussed how AI isn’t necessarily always bad. As a college student myself, I know exactly how useful AI can be. It cuts down on the many time-intensive projects and also helps cast wider nets of research.
However, there’s a world of difference between AI doing the work for you and AI helping you do the work better. Potomac Debate Academy is not pretending AI doesn’t exist, but instead teaches students how to use it responsibly from day one.
Coach Matt explained that every class begins by setting expectations around AI use. All Potomac coaches have access to a presentation that walks students through what responsible AI use actually looks like. The presentation even features Potomac’s AI mascot, Lemmy, a clunky robot whose name comes from LLMs. Throughout the presentation, Lemmy reminds students that he’s not a good debater. He tells students it’s dishonest to use him as a script or pretend its words are their own, but he also shows where he can genuinely help, whether that’s organizing research, brainstorming ideas, or making preparation more efficient.
That philosophy really stood out to me because it doesn’t treat AI as something students should fear or avoid. Instead, it teaches students to think critically about when AI is helping them learn. At PDA, our goal is to help students become better debaters with responsible usage of this new tool.
