Over the years a type of AI called a Large Language Model (LLM) has risen to fame. Trained on billions of words from the internet, an LLM is able to provide human-like writing. But the device you use to write your English assignments may have sinister effects.
The first LLM piercing the public was GPT-3 created by OpenAI in 2020 going by the name ChatGPT. With the massive reach it attained, students found they can use it to help with school work.
This tool has helped many students understand various topics, as the LLM can explain in a different way to what the student may have found confusing; using the tool as a secondary teacher when their main teachers are unavailable.
Many, however, began to use LLMs to completely take over their thought process; using them as a means to complete their assignments rather than help them.
The National Institutes of Health revealed that overusing LLMs can hinder students’ problem solving abilities and reduce their creativity. These LLMs can also perpetuate societal biases, potentially providing racist, sexist, or other socially unacceptable responses due to the training data used.
LLMs also often provide wrong and misleading information; but due to overreliance and trust in these systems, some students take the output provided as fact.
Students don’t even have to go to a separate app or website to get responses to their questions. The widely used Google Search has AI responses built in, with flaws of the service prevalent during its shaky launch in May 2024; Citing non credible sources and giving dangerous information such as putting glue in pizza to make the cheese stick and eating one small rock a day.
In a survey of 121 Mt. Eden students, 57% of students said that they don’t use LLM tools to assist them with school work.
Out of the students that reported that they use such tools, most claimed that it helps them think better. And of the students who said they don’t use them, most claimed that AI tools hinder your ability to think.
Amongst users, there’s an even split of people who claimed to increase or decrease their use of LLMs and other AI tools for school work; with freshmen and sophomores reporting that their usage increased from the previous year, while juniors and seniors had reduced theirs.
While the majority of the students that took the survey claimed that they don’t use LLM tools to assist them with their schoolwork, 14% of students that use LLMs report that they use them a lot (reported a 4 or 5 on a 1-5 scale). As LLMs and AI continue to grow, it has become increasingly difficult to avoid. Though useful when used in moderation, the dangers of overusing LLMs are ever increasing.
