Back to Blog

At the 2025 AIET IN STEM Forum, something unusual happened: researchers stopped presenting and started listening. Students from Jeju Science High School and Jeju National University were asked two questions: How are you using AI in your life? And what should this research group know about AI in education?

The Setup

Students discussed in small groups, then shared their perspectives with international researchers. The responses revealed a generation that is neither naive about AI nor afraid of it - they're pragmatically critical.

How Students Actually Use AI

Research and Report Writing

Students use AI as a research partner - not to write their reports, but to find references, understand difficult papers, and identify methodologies for their experiments.

View Details
  • Finding similar prior research and references for experiments
  • Summarizing difficult academic papers that would take too long to read fully
  • Getting topic suggestions and brainstorming research directions
  • Understanding theoretical backgrounds for performance evaluations
Original Quote

Student (Jeju Science High School): "When we write reports or do experiments in our research, we often have to decide what methodology we're going to use. Since we're all students here, it's actually really crucial to take references from other prior researches. This is where AI comes in. We often ask AI to find similar references, so we can actually take and get feedback from it."

Coding and Debugging

For coding, AI is a debugging partner. Students use it to interpret error messages, understand grammar, and troubleshoot - especially since coding has no fixed answers.

View Details
  • Interpreting cryptic error messages
  • Understanding coding grammar and syntax
  • Finding errors that would take hours to locate manually
  • Learning efficient approaches to problems
Original Quote

Student (Jeju Science High School): "Especially in the case of coding, there is no fixed answer, and we have to find the errors one by one, so when there are errors, we often ask AI how to interpret them, or how to write them grammatically."

Understanding Difficult Concepts

Science high school textbooks are difficult - sometimes without proofs or explanations. Students use AI to understand concepts that aren't explained in their materials, essentially as a faster alternative to Google.

View Details
  • Understanding concepts not explained in textbooks
  • Getting explanations faster than traditional search
  • Learning concepts that would require hours of self-study
  • Using AI as a "Google substitute" for initial exploration
Original Quote

Student (Jeju Science High School): "In our school, unlike general textbooks, we use textbooks that are difficult, even if it's not for our major. There are a lot of cases where we don't even have a proof, so we use it to ask questions."

Data Analysis

For efficient data analysis, students use tools like Publicity and other AI platforms. They also use AI to assess whether their approaches to math problems are efficient.

View Details
  • Efficient data analysis for research projects
  • Checking if math problem-solving approaches are efficient
  • Paper translations for international research
  • Assessing research possibilities

What Students Are Worried About

The Accuracy Problem

Every student mentioned errors. AI hallucinates, provides false content, and can't provide clear sources. Students have learned to cross-check everything.

View Details
  • AI frequently provides false or inaccurate information
  • Sources are often unclear or non-existent
  • Students must cross-check everything AI produces
  • "We use AI as a Google substitute" - but remain skeptical
Original Quote

Student (Jeju Science High School): "We often spot errors in the replies, so we shouldn't necessarily rely on everything on AI, and we shouldn't be skeptical... We don't believe in AI, and when we can't even figure out how to search for the first thing we see, we think about how to use it. After brainstorming, we think about how to use it when we can't make a clear source."

Over-Reliance and Dependency

As students use more AI, they find themselves relying on it more. When knowledge goes beyond what they know, it becomes hard to judge whether AI's responses are correct.

View Details
  • Increasing dependency makes it harder to verify AI outputs
  • When you don't know the domain, you can't catch AI errors
  • Need to think about using AI well without relying too much
  • Standing on your own is important - not just in AI, but in everything
Original Quote

Student (Jeju Science High School): "As I use a lot of AI, I rely on AI more and more. As I get out of the knowledge that I know, I have a hard time judging all the programs in AI. So I think we need to think about how we can use AI well without relying too much on AI."

The Copy-Paste Problem

Some students just copy AI outputs directly into reports. But the issue goes deeper: Can GPT be a standard for evaluation when even human-written work gets flagged as AI-generated?

View Details
  • Some students copy-paste AI answers directly into assignments
  • You need to "know what you are doing" before using AI output
  • AI detection tools are unreliable - human work gets flagged as AI
  • Question: Should GPT be the standard for evaluation?
Original Quote

Student (Jeju Science High School): "A few students even just get the answers from the generative AI and paste them into the report and submit them... Sometimes, even if you write everything yourself, the standard for GPT is lower than the content. Or sometimes, even if you use GPT, the standard for GPT is lower than the content."

What Students Think Researchers Should Know

The Core Message

AI should work as a guide, not a replacement. Students are neutral to slightly negative on AI's future - not because they fear job loss, but because they see the risks of over-reliance and want human-AI coexistence, not dependence.

A Balanced View

Students don't think using AI is bad. But they also don't think it's perfect. The key is knowing when to use it, how to verify it, and maintaining the ability to work independently.

View Details
  • Using AI isn't inherently bad - it's how you use it
  • AI is not yet perfect - errors are common
  • We should create a society where humans coexist with AI
  • Expecting too much from AI leads to negative outcomes
  • AI can be good if it works as a guide, not a replacement
Original Quote

Student (Jeju Science High School): "It is a negative future to expect too much from AI. I think AI can have a good future if it works like a guide."

How Researchers Responded

Fair Exchange

After students shared, researchers were surprised with a request to respond. "We asked you questions, you just shared. I think it's only fair that we respond to what we just heard." - Forum Facilitator

Prof. Janice Gobert's Response

Students' critical perspective on AI is exactly what's needed. The term "enshitification" captures the problem - so much unreliable content on the web that you can't separate good from bad.

View Details
  • "Enshitification" - term for web content quality degradation
  • Students using AI as a partner to troubleshoot is the right approach
  • Different countries have different perspectives on AI (Germany's surveillance concerns, US policies)
  • Critical thinking about AI is valuable and should be encouraged
Original Quote

Prof. Janice Gobert: "I really liked your responses because I think you're critical about AI and knowing that it's scraping the web and can be giving you very inaccurate information... There's a term that somebody encoded this - it's called 'enshitification.' What is it? Enshitification is that there's so much crap essentially on the web that you can't even separate out the good stuff from the bad stuff."

Prof. Jiliang Tang's Response

An AI professor shares a story: taking a Tesla Uber where the driver was doing homework while autopilot drove. "Who is driving now?" We use AI every day without realizing it.

View Details
  • The Tesla story: Student driver doing homework while AI drives
  • "We never use AI" - but who is driving the car?
  • Face ID, fingerprints, autocomplete, email responses - all AI
  • "AI is much better in newspaper than the real AI"
  • AI will not replace humans. Humans who don't use AI will be replaced.
Original Quote

Prof. Jiliang Tang: "The driver said, 'We never use AI.' So this driver. And then I asked, 'Who is driving now?' Right? So actually, the story tells us, we use AI almost every day. In many scenarios, we don't even realize... AI will not replace humans. But humans who don't use AI will be replaced."

Key Takeaways from the Student Panel

1
On Using AI

Consensus View

AI is a tool for research, coding, and understanding - not a replacement for thinking. Use it as a starting point for brainstorming, then verify everything.
2
On AI Errors

Universal Concern

Every student mentioned encountering errors and false information. The ability to detect AI mistakes is becoming a crucial skill.
3
On the Future

Balanced Perspective

AI should work as a guide. Human-AI coexistence, not dependence. Standing on your own is important - with or without AI.