2.1. School Library Education

In 1999, the International Federation of Library Associations and Institutions/United Nations Educational, Scientific and Cultural Organization’s School Library Manifesto defined the role of a school library as “to develop students’ enjoyment of reading and learning, habits and attitudes to use the library throughout their lives, and to guide them to evaluate and utilize information containing various materials” (International Federation of Library Associations and Institutions & United Nations Educational, Scientific and Cultural Organization, 1999).

The Role of the School Library (American Association of School Librarians, 2019) released in 2019 stated that the school library is an essential component of the educational environment, because it is a dynamic learning environment that bridges the gap between access and opportunities for all students. In addition, the education and services provided through school libraries are developed through the following six essential sharing bases, listed as “Inquire,” “Include,” “Collaborate,” “Curate,” “Exploration,” and “Engage.”

Song (2018) stated that school library education is an educational information service performed by school libraries to achieve school education goals. As the role of school libraries in school education increases, library use and reading education, which were previously operated as informal curricula, are expanding to information literacy instruction. In addition, school library education is divided into education on school libraries and education using school libraries. Education on school libraries aims to increase ability toward self-directed production of new information, and knowledge based on the ability to utilize various learning materials owned by school libraries. In addition, school library-based education is an activity that enhances learning ability by allowing students to self-directedly solve the subject’s learning topics by using various learning tools owned by school libraries.

However, there are many school libraries in Korea where librarians do not work. Even if one is working, there may not be any assistant, and so many tasks other than training must be handled alone. Therefore, if there is a way to reduce workload and focus on education, it should be seriously investigated.

Song (2019) suggests that librarians should be in charge of school libraries, which are physical and digital learning spaces that perform teaching/learning activities based on reading, exploration, research, thinking, imagination, and creativity. To help or aid librarians with their ever-increasing workload, ChatGPT should be used. ChatGPT is expected to be an essential element as a future learning tool.

Hence, what activities can librarians at school libraries do using ChatGPT? According to Yorio (2023), who received and organized the answers to this question from a chatbot: “Providing reference assistance,” “Recommending books and other materials,” “Helping with writing assignments,” “Providing digital resources and online tutorials,” “Creating interactive learning experiences,” and “Providing language assistance for non-native speakers” were possible.

2.2. Chatbot Education Use

ChatGPT is attracting attention as an educational tool that realizes ‘learning initiative,’ which is emphasized in the current curriculum and future educational settings direction (Shin et al., 2023).

Hetler (2024) stated that because ChatGPT is versatile, it can be used to solve math problems or find keywords optimized for search engines when people code computer programs or compose music. However, because it does not fully understand the complexity of language used by humans, the answers may be inaccurate. Various statistics can be cited and provided, but there are limitations, such as not being able to make practical comments on what it means or relevant topics. In addition, ChatGPT can be helpful for some tasks, but it is also necessary to use it with caution because it has ethical problems, such as those associated with plagiarism, privacy, and cyber security. Rouse (2024) said that some schools have completely banned the use of ChatGPT, but instead of this, teachers need to teach students how to use this type of AI ethically so that they can develop critical thinking skills.

Dilmegani (2024) cited the advantages of using ChatGPT in education as three categories: speed, availability, and personalized learning, and presented seven different cases from the perspective of teachers and students, respectively, as summarized in Table 1.

In Korea, measures to utilize ChatGPT for education are also being actively sought, and in July 2023, the Ulsan Metropolitan Office of Education developed a class resource book using ChatGPT for English teachers and distributed it to schools. The data collection includes instructional design using ChatGPT, project class cases, and evaluation methods, and includes how to organize and utilize prompts, reconstruct or produce dialogues presented in textbooks at various levels, and create role-play or scripted theater class materials.

Additionally, Jeong and Shin (2023) developed an AI chatbot that can be used in elementary school science classes. It was applied to sixth grade students and results confirmed that using it improved scientific academic achievement and science positivity.

As previously stated, research on the use of chatbots and specific measures for school library education in Korea has not yet been published. Nevertheless, cases are being introduced for training of librarian teachers with magazines such as school library journals and Internet newspapers. Among them, according to the case of Seoul Sinlim High School, published in the Internet Newspaper Education Plus (Woo, 2023), ChatGPT was used as an auxiliary or complementary tool in cooperation with classes with chemistry subjects. At this time, the chemistry teacher selected the keywords “medicine,” “new material,” “biochemistry,” and “cosmetics/fashion,” and the librarian teacher recommended a book suitable for each topic. Afterwards, students selected a topic to read and discuss from these books, and they had the opportunity to ask and correct whether it was appropriate as a topic and whether the arguments for or against were valid.

The above literature review results show that there are several issues not only in the use of AI chatbots themselves, but also in their use in school education. If they can be used positively while diminishing any possible negative aspect associated with this technology, it raises expectations that the effectiveness of education can be increased.

4.1. Chatbot Selections

Chatbot service can play the following roles in improving the learning experience by connecting educational institutions and students. First, the chatbot can improve the curriculum and adjust the learning path according to the progress of the student. Secondly, it is possible to provide an appropriate learning experience to each student. Thirdly, students can receive answers to solve tasks through the chatbot. Fourthly, student progress data generated by the chatbot can be used to track performance and provide improvement measures or study pathways. Fifthly, since the chatbot provides a conversation-based experience, it can increase interest in learning. Sixthly, learning support can be received whenever desired.

However, since the learning effect may vary depending on which chatbot is selected, it is necessary to first explore the possible viability of each as an educational tool. The following Table 2 summarizes the characteristics of the three chatbots selected to explore possibilities for use in school libraries.

4.2. Survey

The main premise of this study is to investigate the possibility of using chatbots as educational tools for school libraries. In order to achieve the research objectives, a survey questionnaire was developed from the librarian teacher recognition study (Kim et al., 2023) on the introduction of ChatGPT in school libraries and the librarian recognition study using metaverse in university libraries (Oh & Lee, 2022). Subsequently, an online survey was conducted for five days from August 17 to August 21, 2023, composed of questions in the form provided by the Naver Office. The survey used a snow sampling method that asks some librarian teachers working in the Metropolis of Seoul and Gyeonggi province to participate by sending links, and then gradually expanding them to recruit samples. Instead of predicting the number of samples, only the results of the number of responses during the survey period were set as analysis targets. The reason the survey target area was limited to Seoul and Gyeonggi province is because Gyeonggi province has the largest number of school libraries in the country and Seoul is the capital of Korea and also the center of education. Table 3 summarizes the survey questions.

4.3. Structured Questionnaire

Next, in order to practically evaluate whether each chatbot could be used as a school library educational tool, a structured questionnaire was created to ask questions directly and check the answers. The structured questionnaire was composed of the following Table 4 materials by applying “social dialogue ability,” “information provision ability in various fields,” “language command ability,” “creative ability,” and “moral discernment,” which were developed by Shin et al. (2023) to explore the possibility of using ChatGPT.

The performance missions assessment for each chatbot evaluation summarized in Table 3 includes five different areas with a total of 15 detailed questions. Among them, the “social conversation ability” area was included for priority evaluation because chatbots generally focus on topics of daily conversation. The performance mission of “ability to provide information in various fields” and “creative ability” were evaluated by chatbots and was a choice that took into account its function. Next, “moral discernment” took into account the controversy swirling over the unethicality of chatbots. Finally, “ability to understand libraries” was set as an implementation mission in terms of exploring the possibility of using it in school library education.

In addition, Table 5 is a performance evaluation paper for chatbots by area, and “educational utilization possibility” is included in the five areas presented in Table 3. The level for each area was composed of a five-point scale ranging from insufficient to excellent.

5.1. Analysis of Survey Results

5.2. Analysis of Structured Questionnaire

As shown in Table 4, the structured questionnaire contains missions by area of “social conversation ability,” “information provision ability in various fields,” “language command ability,” “creative ability,” and “moral discernment.” This questionnaire was delivered to 15 librarians working in elementary, middle, and high school libraries to carry out missions through Google’s Bard, Microsoft’s Bing, and OpenAI’s Nova, and to reflect the performance evaluation scores of Table 4. Table 8 summarizes the mix of work experience levels for librarian teachers who participated in the structured questionnaire.

According to the results summarized in Table 8, a total of 15 librarians participated in missions by area using chatbots, including five elementary schools, five middle schools, and five high schools. Among those who accepted participation in the study were students of the researcher or those who met at conferences and made friends, and the criteria for selection included librarian teachers with at least five years of experience working in the school library. The selection criteria were librarians with at least five years of experience working in school libraries. As a result, the average work experience of the librarians who participated in answering the structured questionnaire in this study was 12.2 years in elementary school, 11 years in middle school, and 8.6 years in high school.

Tables 9-Table 1011 are summarized by analyzing the evaluation results for Bard, Bing, and Nova, respectively.

In the evaluation results of the chatbot Bard summarized in Table 9, the area with the highest average score was “social conversation ability,” and the area with the lowest score was “educational usability.” Since “library understanding ability” ranks 5th out of 6 areas, the possibility of using it as a school library educational tool can be said to be low.

In the evaluation results of the chatbot Bing summarized in Table 10, the area with the highest average score was “ability to provide information in various fields,” and the area with the lowest score was “educational usability.” In addition, since “library understanding ability” ranks 5th out of 6 areas, the possibility of using it as a school library educational tool can also be said to be low.

In the evaluation results of the chatbot Nova summarized in Table 11, the area with the highest average score was “ability to provide information in various fields,” and the areas with the lowest were “ability to understand libraries” and “moral discernment.” However, since “educational usability” ranks 3rd out of 6 areas, it can be said that it has the highest possibility of being used as an educational tool.

Finally, Table 12 summarizes the results from Tables 9-Table 1011.

According to the results summarized in Table 12, among the three chatbots that were subject to performance evaluation, Nova received the highest score in the entire performance for each of the six areas. However, the overall average score was usually less than 2.9 points, and in particular, the “educational use possibility” was the lowest at 2.6 points. From the above results, librarians evaluated that while chatbots are still insufficient in other areas, they are higher than average in “social conversation ability” or “information provision ability in various fields,” so it is possible to use them as an educational tool in school libraries. Although it is clear that these results were evaluated by librarian teachers working in elementary schools, middle schools, and high schools, they do not take into account all differences in the context of information or educational levels required or expected by each school.

Fig. 1 is a graph of the results of Table 12.

Fig. 1

Results of chatbot performance evaluation by area.