There is a lot of chatter about where AI fits in education. As EducAIte's Head of Growth, I set out to get the answers. A few weeks ago I presented to students about EducAIte and the importance of having guardrails for students when they are using AI.

I ran an experiment to see what the students would do and say. I split the group into 4 different groups. The topic: Should students wear school uniforms?

  • Group 1: Research the topic and agree and DON'T use AI.
  • Group 2: Research the topic and agree and USE AI.
  • Group 3: Research the topic and disagree and DON'T use AI.
  • Group 4: Research the topic and disagree and USE AI.

My original idea was to have the students debate to see if the students who used AI could argue their points. I ran out of time and asked each student to talk about their experience doing the research and the results shocked everyone.

Here is what the students said…

1. Students were honest about AI's limitations and said that it was helpful when given context and used for a specific purpose.

The representative from Group 4 talked about using AI but giving it a specific prompt to make sure he got the best results.

2. Students want to collaborate and learn about other student's experiences. This helps them create their own.

Group 1, did not use AI. Their representative talked about the conversations they had and what they learned about other student's points of view. They discussed how some students had positive experiences wearing uniforms and others did not. The key was that there was conversation and collaboration.

3. Students who use traditional AI models are not learning; they are just reading structured output.

Group 2 representatives used AI. He talked about the arguments presented but later said that if he was going to be asked to argue on behalf of the team he would not be able to remember.

4. All students agreed that critical thinking and learning is more important than ever. Studies are showing that traditional large language models are removing deep learning.

As highlighted by Melumad (2025), research indicates that individuals who rely on syntheses from LLMs rather than standard web searches feel decreased personal investment when crafting advice; consequently, their recommendations tend to be less original, more sparse, and less persuasive to others.

EducAIte was built for students like these who want to use a tool that empowers them to do their own research, formulate their own opinions, and perspectives, and present well-thought-out arguments.

Sources
Melumad S, Yun JH. Experimental evidence of the effects of large language models versus web search on depth of learning. PNAS Nexus. 2025 Oct 28;4(10):pgaf316. doi: 10.1093/pnasnexus/pgaf316. PMID: 41163786; PMCID: PMC12560091.

To learn more about how to bring safe AI usage to your schools, contact campbell@educaitelearning.com, or contact us here.