Page 84 - AGRANI SAMKALP SILVER JUBILEE SOUVENIR

AGRANI SAMKALP SILVER JUBILEE SOUVENIR

IMPACT OF LLMS ON EDUCATION: THE RESEARCHERS' PERSPECTIVE
Prof. (Dr.) Sachin Gupta
Dean (Research and Innovation)
Maharaja Agrasen Institute of Technology, Delhi

“Generative AI is the key to solving some of the world's Figure 1.
biggest problems, such as climate change, poverty, and LLMs offer promising
disease. It has the potential to make the world a better place opportunities to personalize learning,
for everyone.” ~Mark Zuckerberg enhance student engagement, and
The emergence of Large Language Models (LLMs) streamline tasks like content creation.
including ChatGPT, Google Gemini, Microsoft Bing chat Many educators and students view
and other sophisticated AI language models, has generated them as a tool for democratizing
significant interest in its potential applications and education, making learning more
implications across various sectors, particularly in education. accessible and equitable. However, there are concerns about
But are these systems recommended in education? Should academic integrity, with the potential for plagiarism and
the students use them? Should they even be allowed? Would superficial learning. Over-reliance on the tools could hinder
the use of LLMs make students dumb? Let us find out the the development of critical thinking, creativity, and genuine
researchers’ perspectives. knowledge acquisition. To understand these concerns,
Introduction researchers are employing diverse methodologies to
LLMs are a subset of Natural language generation understand how LLMs are being used in educational settings
(NLG), which in turn is a class of Generative AI. LLMs are and their impact on learning, teaching, and skill development.
trained on a large corpus of documents to learn the Experimental Studies and Perspectives
distribution of words in text and predict a possible word The novelty of the research area and its potential impact
sequence to generate new text. The resulting AI Applications on human cognition, society, job markets and careers have
have been showing outstanding results in creative and triggered a lot of studies in this domain. The studies are being
professional writing including poetry, novels, stories, blogs, designed to cover a variety of research questions, a summary
posts and code. These systems are interactive, and answer for some of which are presented below.
the queries of users which may include subject specific and Contributions to Learning Outcomes: These studies
general queries. Students often use these LLMs for assistance use controlled experimental designs to evaluate the causal
with their school or college subjects, and the usage is growing effects of ChatGPT on specific learning outcomes. For
with sophistication and features being offered by different instance, one study used a quasi-experimental approach to
LLMs. examine the quality and originality of essays generated by

postgraduate students using ChatGPT compared to those
written without AI assistance. By comparing the performance
of students who use ChatGPT with those who do not,
researchers could isolate the specific contributions of AI tools
to learning outcomes.
Integration into Learning Practices: Researchers are
Figure 1: Relative popularity of contemporary LLMs as also conducting surveys to gain insights into how students
per Google Trends are integrating ChatGPT into their learning practices. Latent
This article focuses on LLMs in general, however most profile analysis (LPA) is often used to identify distinct groups
of the research considered for summarization discusses of students based on their patterns of ChatGPT reliance. For
ChatGPT in particular, which is the most popular amongst example, one study identified five profiles of students-
all LLMs as apparent from Google Trends illustrated in ”versatile low reliers,” “all-rounders,” “knowledge seekers,”

82