Author: Le Thanh Binh - UPP Global Technology

Abstract

In recent times, the subject of ChatGPT has gained considerable attention on social media platforms. This article provides an overview of ChatGPT and its model architecture, which built upon Transformers architecture. After that, ChatGPT was situated within the framework of ongoing Artificial Intelligence in Education (AIEd) research, talk about applications geared toward student-facing, teacher-facing, and system-facing, and assess potential and risks. In conclusion, we provide some suggestions for students, professors, and higher education institutions as a way to wrap up the essay.

Introduction - a brief of GPT (Generate Pre-trained Transformer) and ChatGPT

GPT (Generative Pre-trained Transformer) is a family of natural language processing (NLP) models. These models are designed to generate human-like text by predicting the next word in a sentence or generating an entire sentence given some initial input. GPT models are pre-trained using unsupervised learning techniques, which means that they are trained on large amounts of text data without any human supervision or labeling. For specific tasks, GPT can be fine-tuned for apply in a wide range of applications, including summarization (Grail, Q., Perez, J., & Gaussier, E. (2021), sentiment analysis (Zhang, L., Fan, H., Peng, C., Rao, G., & Cong, Q. (2020)), language translation (Sawai, R., Paik, I., & Kuwana, A. (2021)), and even writing entire news articles (Dale, R. (2021)).

At UPP Technology, we can offer the involvement of human experts in fine-tuning GPT models for diverse tasks that would benefit your organization.

On November 30, 2022, ChatGPT was introduced as a prototype and rapidly gained popularity due to its ability to provide detailed and articulate responses across various domains of knowledge. Developed using a InstructGPT model that was trained using supervised techniques, ChatGPT offers real-time conversations and offers advice and solutions to complex problems. Additionally, since ChatGPT is designed to emulate human conversations, a number of experts have observed that it can be difficult to distinguish between interacting with a machine program and communicating with an actual human, particularly if we are not aware of it in advance. (Cotton, D., Cotton, P., & Shipway, J. R. 2023)

We can outline some notable characteristics of ChatGPT in comparison to other chatbot products that are currently on the market (generated by ChatGPT):

• Language Understanding: ChatGPT has a highly advanced ability to understand and interpret natural language, including idiomatic expressions and complex sentence structures. This allows it to respond to queries and conversations more accurately than many other chatbots.
• Flexibility: Unlike many other chatbots, ChatGPT can be trained to perform a wide range of tasks, from answering general knowledge questions to offering personalized recommendations based on a user's interests and preferences.
• Contextualization: ChatGPT can take context into account when generating responses, allowing it to provide more relevant and nuanced answers to queries. It can also maintain a conversation over a longer period of time and adjust its responses based on previous interactions.
• Multilingual Support: ChatGPT can support multiple languages, allowing it to communicate with users from different countries and regions. This makes it a powerful tool for businesses and organizations that operate globally.

The implication of ChatGPT for education system

In this portion, we will delve into the specific applications and technologies of ChatGPT that can be utilized to enhance the education system.

In context of applying existing artificial intelligence in education, we can divide as 3 main parts: Students-facing applications, Teacher-facing applications, System-facing applications.

1. Student-facing applications

Regarding student learning, ChatGPT can be beneficial by assisting with:

• Personalized learning: ChatGPT can provide personalized learning experiences to students by tailoring the content to their individual needs and learning styles.
• Instant feedback: ChatGPT can provide instant feedback on students' responses, helping them identify areas of strength and weakness, and allowing them to improve their learning.
• Access to vast knowledge: ChatGPT can provide students with access to a vast array of knowledge, from different sources and in different formats, giving them the opportunity to explore new ideas and concepts.
• Engaging learning: ChatGPT can provide an engaging and interactive learning experience, using conversational language and multimedia resources to keep students interested and motivated.

2. Teacher-facing applications

ChatGPT can be used for reducing workloads from teachers such as: Automating assessment, creating education content (quizz, homework), provide real-time feedback when teacher is not in-charge., etc. Some of most popular applications include:

• Automated Essay Scoring (AES): using statistic, NLP, LSA (latent semantic analysis), and the algorithms can be used to identify patterns in text responses and prompt students to revise the responses. ChatGPT has the capability to assist teachers by checking the grammar of written work and automatically comparing students' assignment answers with the answer keys provided by teachers.
• AI-enabled automatic feedback: The system establishes the appropriate answers based on Bloom’s cognitive levels and recommends additional learning resources and challenges.
• Writing Assistant: ChatGPT was developed to facilitate English writing practices and to enhance writing skills, particularly in higher education. ChatGPT is capable of assisting with various tasks, such as verifying grammatical accuracy and rewording paragraphs to aid in writing, among others.

3. System-facing application

Some AI-based applications that focus on the system side of education can offer academic administrators and managers high-level data, such as the patterns of attrition in different schools or institutions. Nonetheless, this aspect has been given the least amount of attention.

However, with the appearence of ChatGPT and GPT-3, we can expect a future of more opportunities for reseachers and scientists in this application.

Risks and challenges when using ChatGPT for education

If AI can produce the answers, there is less incentive for students to acquire knowledge, skills and independent learning. Relying too heavily on the chatbot can result in negative consequences for the student, such as a weak grasp of the subject matter, reduced motivation to study, and an inability to independently research or explore alternative resources. Additionally, without seeking other sources of information for comparison, the student may be at risk of receiving inaccurate or misleading information.

Plagiarism is also a concernable problem. It is difficult to distinguish between a student’s own writing and the responses generated by a chatbot application, mostly in case students can easily access to ChatGPT for asking. (Cotton, D., Cotton, P., & Shipway, J. R. 2023)

OpenAI itselft notes that the bot may sometimes provide plausible-sounding but incorrect or nonsensical answers and may exhibit biased behavior. This could lead to misunderstandings or errors on the part of students who rely solely on the system for information. As a result, it is important to use ChatGPT as a supplement to traditional learning methods rather than a replacement for them, to ensure that students develop critical thinking skills, problem-solving abilities, and creativity.

Recommendations for students and educators in the age of Artificial Intelligence

To address the issue of plagiarism and improve accessment process, there are several strategies that teachers can implement, including:

• Provide clear and detailed instructions to students regarding how to structure their assignments. Ensure that the assignments are written in a more structured and coherent manner.
• Use a rubric to evaluate the quality of student work. The rubric should focus more on the process instead of only on the results of the work.
• Use a combination of automated and manual assessment techniques to evaluate the student’s understanding of the material.

I suggest that students should consider the following recommendations:

• Be aware of academic integrity policies and understand the consequences of academic misconduct. (RudolphA, Jürgen 2022).
• Read widely and voraciously to improve critical and creative thinking. (RudolphA, Jürgen 2022).
• Practise the use of AI language tools (like ChatGPT) to solve real-wor65ld problems (Zhai, 2022).
• Focus more on other skills that between humans and can not be solved by machine like working in groups, communication.

Conclusion

Tools such as GPT-3 and ChatGPT are bringing AI to the threshold of being a routine aspect of daily life. While the effects of this advancement on society are only beginning to emerge, the potential for substantial impact is enormous. In my opinion, AI will play a key role in shaping the future of education by facilitating communication between individuals, promoting creative pursuits, decreasing tedious homework in favor of more collaborative and imaginative group assignments.

Reference

Cotton, D., Cotton, P., & Shipway, J. R. (2023, January 10). Chatting and Cheating. Ensuring academic integrity in the era of ChatGPT.

RudolphA, Jürgen. "Journal of Applied Learning & Teaching." Journal of Applied Learning & Teaching 5, no. 1 (2022).

Azaria, A. (2022). ChatGPT usage and limitations. Preprint. DOI: 10.13140/RG.2.2.26616.11526

Deng, J., & Lin, Y. (2022). The benefits and challenges of ChatGPT: An overview. Frontiers in Computing and Intelligent Systems, 2(2), 81-83.

Zhai, X. (2022). ChatGPT user experience: Implications for education.

Grail, Q., Perez, J., & Gaussier, E. (2021, April). Globalizing BERT-based transformer architectures for long document summarization. In Proceedings of the 16th conference of the European chapter of the Association for Computational Linguistics: Main volume (pp. 1792-1810).

Zhang, L., Fan, H., Peng, C., Rao, G., & Cong, Q. (2020, August). Sentiment analysis methods for hpv vaccines related tweets based on transfer learning. In Healthcare (Vol. 8, No. 3, p. 307). MDPI.

Sawai, R., Paik, I., & Kuwana, A. (2021). Sentence Augmentation for Language Translation Using GPT-2. Electronics, 10 (24), 3082.

Dale, R. (2021). GPT-3: What’s it good for?. Natural Language Engineering , 27 (1), 113-118.