J. People Plants Environ Search

CLOSE


J. People Plants Environ > Volume 26(6); 2023 > Article
Kwack and Im: A Study of Pre-Service Teachers’ Perceptions of the Use of ChatGPT in Practical Gardening Training

ABSTRACT

Background and objective: The purpose of this study is to determine the empirical tendency of pre-service teachers to use ChatGPT for practical gardening training and the impact of using ChatGPT characteristics on problem solving ability, and to draw implications. For the study, we investigated the general trend of horticultural practice (practical gardening classes), the trend of horticultural practice using ChatGPT, and the problem-solving ability after practical gardening training.
Methods: In this study, problem solving skills were categorized into knowledge, skills and attitudes and correlated with ChatGPT, and the effects of ChatGPT characteristics on immediacy, continuity of questions, adjustability of the level of responses, suitability of media to the current generation, advantages of the lack of necessity to ask directly, advantages of conversational knowledge, expertise, applicability of educational utilization, and diversity of range of questions were analyzed.
Results: The results showed that pre-service teachers found ChatGPT to be very useful in teaching horticultural practice, with a high preference for customized lessons and discovery of different teaching materials. In particular, we examined the problem solving skills of the pre-service teachers in horticultural practice and found that they had difficulties when problems arose. We therefore organized them into knowledge, skills and attitudes to examine the correlation with ChatGPT characteristics. As a result, knowledge and skills in practical gardening training were positively correlated with ChatGPT characteristics, while attitudes and ChatGPT characteristics were negatively correlated in immediacy, continuity of questions, adjustability of the level of responses, and suitability of medium, and the others showed a relatively low correlation. The effects of knowledge, skills and attitudes on the use of ChatGPT by pre-service teachers’ problem solving skills in practical gardening training were analyzed using multiple regression analysis. The knowledge aspect of immediacy and continuity of questions had a significant effect. It can be seen that ChatGPT characteristics such as ‘immediacy’ and ‘continuity of questions’ had the greatest impact on the knowledge aspect of problem solving ability. Attitude aspect had a smaller effect. ‘suitability of media to the current generation’ had a significant effect on attitude, whereas ‘expertise’ had no significant effect. Applicability of educational utilization and diversity of range of questions had a significant effect on both knowledge and skills.
Conclusion: Based on the findings in this study, we explored the possibility of using ChatGPT to educate pre-service teachers about horticultural practice and suggested implications for future related research.

Introduction

Satisfaction with practical education is increased when class performance standards and objectives are achieved, which motivates learners more and is an important complementary factor in the acquisition of knowledge in their major (Kim and Lee, 2023). Practical gardening training is carried out according to school conditions in addition to horticultural theory education, and the essence of horticultural education is to maintain a complementary relationship between theory and practice (Im, 2018; Oh, 2019). However, the large number of students compared to instructors, the difficulty in answering learners’ questions, the lack of feedback on practical results, and the lack of plant knowledge and material preparation in the case of horticultural practice often result in a low density of experience in practical education (Chi et al., 2018). For practical education to be meaningful as a learning experience, the process of voluntary interest is essential, especially for students who are not horticulture majors. However, this process is difficult in traditional placements, and placements tend to be one-off and unstable (Kim and Lee, 2023). In addition, Kim (2014) found that practicum classes require problem-solving skills and diffuse thinking about activities. However, it has been reported that it may be difficult to develop the problem-solving skills required for practicums because they are one-time events. (Ahn, et al., 2022).
ChatGPT (a combination of Chat and Generative Pre-trained Transformer), an AI-based tool, can be a very effective alternative to this problematic situation for practical gardening education (Sun et al., 2023). ChatGPT is an artificial intelligence dialog model developed by OpenAI, first introduced in 2021, which can answer user questions or perform various language-based tasks through natural language processing. In particular, ChatGPT can help students enhance their learning experience through immediate and interactive conversations (Crompton, 2023; OpenAI, 2023), personalized learning, and effective feedback based on the learner’s level (Lee et al., 2023).Using a chat-based platform that is familiar to the digital native generation, ChatGPT can be useful for students who are reluctant to ask questions directly to teachers (OpenAI, 2023).As a state-of-the-art natural language processing (NLP) model (Lund and Wang, 2023), it can understand context using large training data and provide knowledge in a wide range of contexts.(Jeon et al., 2023).
Practical gardening training needs to be properly implemented to effectively teach horticultural activities, which are currently reported to be successful in many tertiary industries beyond the primary and secondary sectors to education, service, and healing. Therefore, this study aims to explore how ChatGPT can be effectively utilized to overcome various limitations of horticultural practice and foster problem-solving skills required in practice (Lim, 2015; Oh, 2019; Lee et al., 2023). In this study, we examine the empirical trends of horticultural practice classes using ChatGPT, and then empirically investigate how ChatGPT affects the educational characteristics of horticultural practice according to B. Bloom’s taxonomy of educational goals: knowledge, skills, and attitudes (Borich, 2000). We also aim to provide implications for the direction of practical gardening training using the results of the analysis. To this end, the following research questions were posed.
First, to identify general trends in horticulture education.
Second, to understand the views of horticulture practicum classes using ChatGPT and the extent to which A.I. chatbots such as ChatGPT have an impact on future teachers.
Third, to determine the correlation between ChatGPT and the identification of three variables of knowledge, skills, and attitudes of problem solving skills in horticulture practicum classes.
Fourth, to determine the impact of ChatGPT on knowledge, skills and attitudes towards practical gardening classes and its pedagogical characteristics.

Theological Background

Nature of practical training and horticultural practical training

Practical education is the process of implementing educational contents that transfer knowledge from theoretical classes and has the purpose of providing sufficient interaction between instructors and learners. In addition, most programs that require internships provide students with the opportunity to fully acquire integrated adaptability of key knowledge and skills as core competencies. However, internships also require satisfactory learning methods that allow students to learn a variety of knowledge to improve their understanding. In addition, it can not only improve the level of achievement and motivation in class, but also provide practical help to acquire major knowledge. Therefore, the satisfaction of practical education is highly correlated with the reduction of academic burden; therefore, quality practical education is also a learning method that requires a case-centered problem-solving approach and efforts to ensure practical knowledge (Kim, 2014). Horticultural education and practical activities can be used in various environments through horticultural plants. Learning can be developed through functional and design education and therapeutic aspects. In addition, direct horticultural activities can be used as a variety of comprehensive learning, such as stimulation of cognitive, social, and emotional aspects, natural experience with plants, five senses, and understanding the relationship with the environment (Seo et al., 2022). Therefore, this horticultural knowledge must be accompanied by many experiences and experiences depending on the situation, which can usually be achieved through program activities or practice. It has also been shown that horticultural activities can be used as a form of STEAM in education and complex problem solving (Kwack et al., 2020).

Characteristics of utilizing ChatGPT

Although AI chatbot programs existed before ChatGPT, ChatGPT is a state-of-the-art natural language processing (NLP) model (Lund and Wang, 2023) that allows users to interact with it in a very natural and human-like way. The applications of ChatGPT are endless, including writing (Jang et al., 2023), journalism (Lee et al., 2023), and coding (Nigar et al., 2023). Since the first ChatGPT coauthored paper was published in 2022, many researchers have been actively using it in their research. It has been claimed that ChatGPT has a positive effect on creative thinking and problem solving by allowing users to provide short notes and bullet points of sporadic and fragmented knowledge, which can then be completed or coherently transcribed into text using ChatGPT. In addition, the emergence of ChatGPT suggests that the ability to ask questions is more important than getting the right answer, and that the era of “how to ask” has arrived, whereas in the past it was “how to know”. Therefore, learners can improve their academic performance and quality of experience through the complementary process of asking questions and receiving continuous feedback from generative AI (Kim et al., 2023). This ChatGPT-based lesson design is characterized by the flexibility of customized lesson design based on the time and effort required by students for the lesson content or related knowledge and experience, the diversity of material utilization, and the continuity of questions, and the strength in terms of the convenience of providing feedback (Kim, 2023).

ChatGPT utilization and troubleshooting skills

Problem solving is a creative thinking process that involves discovering problems that arise in practical gardening classes, and solving them by generating various ideas and alternatives. The Ministry of Education (2016) in South Korea emphasizes problem solving to create something new by using basic knowledge and various experiences creatively and convergently. Practical gardening training can be understood with the visual process shown by the teacher, and the effectiveness of practical gardening training is enhanced. However, students have different preferences and knowledge and skill levels, so it is necessary to introduce different learning methods. To solve this problem, ChatGPT is suitable for use (Lee, 2023). The problem-solving pedagogical characteristics of ChatGPT include the immediacy of solving questions, the continuity of questions that can be deepened, the possibility of adjusting the level of answers, the suitability of the medium for the Millennials and Generation Z, the advantage of not having to ask questions directly to the instructor, the ease of asking questions and getting the desired answers, and the advantage of conversational knowledge that removes emotional barriers to knowledge acquisition (Chung and Bang, 2015; Lee et al., 2023). In addition, the expertise of chat-GPT information can be said to have great educational value in many ways (Liu et al., 2023). Current education emphasizes multidisciplinary, i.e., interdisciplinary integration, but the educational field does not reflect this, and the diversity of ChatGPT’s academic scope is expected to be highly utilized (OpenAI, 2023). In addition, practical education requires skills and attitudes that ensure an individual-centered problem-solving approach and practical knowledge (Kim, 2014), and the use of ChatGPT enables students to understand the problem situations they encounter during practice, establish questions about them, and enable continuous learning based on them. In this way, ChatGPT focuses on the pedagogical possibilities of supporting the learning of problem-solving skills (Oh and Kim, 2019).

Research Methods

Study model

This study aims to empirically investigate how the problem-solving abilities (knowledge, skills, and attitudes) of pre-service teachers participating in horticultural practice affect the pedagogical characteristics of ChatGPT. For this purpose, a study model was established as follows. (Fig. 1).

Study subject

The subjects of this study were 62 pre-service teachers who were taking “Horticulture” as an elective course in their second year of study at a university of education located in the city of S. Purposive sampling was used. At that time, the students were asked to set aside a certain amount of time before the end of the practical training period, in addition to the theoretical training classes, to search for questions related to the practical content using the free version of ChatGPT 3.5. The survey period was from September 24, 2023 to September 30, 2023, and the survey was conducted online for a total of one week. The general characteristics of the study participants are shown in Table 1.
In addition, the level of questioning and the scope of questions were set as shown in Table 2 for the training on the use of prompts for ChatGPT use, and a total of three training sessions were conducted during the theory course on the level of ChatGPT use by instructors before the practical course (OpenAI, 2023).

Study tools

Organization of survey questions

The measurement tool to determine the objectives of this study is a questionnaire. The questionnaire consisted of 3 questions on general characteristics, 3 questions on experience of gardening practice class, 4 questions on gardening practice using ChatGPT, 16 questions on problem solving ability test, and 27 questions on ChatGPT characteristics to fit the variables of this study. The composition of the survey questions is shown in Table 3.

Problem solving test sheet

A problem solving test was conducted to analyze the effect of ChatGPT on practical gardening training. The questionnaire used in this study was modified and supplemented to fit the questionnaire used in Kwak, Hye-ran (2022). Each item is measured on a 5-point Likert-type scale ranging from 1 (not at all) to 5 (very much). Higher scores indicate greater problem solving ability. The overall reliability coefficient of this study is Cronbach’s α of .800, and the item content is shown in Table 4.

Characteristics of ChatGPT

To analyze the impact of using ChatGPT in horticultural practice classes, we created factors based on the nine concepts analyzed (Crompton, 2023; OpenAI, 2023; Lung and Wang, 2023) and composed a total of 27 questions, three for each factor, as shown in Table 5. Each item is measured on a 5-point Likert-type scale ranging from 1 (not at all true) to 5 (very true). The overall reliability of the instrument is Cronbach’s α of .933.

Data analysis

In this study, the data collected through the survey was analyzed using the SPSS Statistics 26.0 statistical package. The method of analysis is as follows
First, a frequency analysis was conducted to identify the general characteristics of the survey participants.
Second, Cronbach’s a coefficient was calculated for exploratory factor analysis and reliability of the measurement tools.
Third, Pearson correlation analysis was conducted to analyze the relationship between the extracted factors and the main concepts of ChatGPT characteristics.
Fourth, we conducted multiple regression analysis to verify the influence of ChatGPT characteristics.

Results and Discussion

General trends in practical gardening training

The mean and standard deviation were calculated based on a 5-point scale to find out the trend of pre-service teachers’ opinions on “importance of practice”, “impact of practice on horticulture learning”, and “most difficult parts” of practical gardening training. As a result of the analysis, the majority of pre-service teachers responded to the importance of practice as 79.1% of the respondents selected “important” and “very important”, and the mean of the importance of practice was 4.03 (SD = 1.21). Regarding the impact of practicum on horticultural learning, the mean was 4.40 (SD = .89), with 88.7% of the respondents ranging from “somewhat influential”, “fairly influential”, and “almost critically influential”, and the mean was 4.40 (SD = .89). In addition, when practicing horticultural activities, the most difficult part was “plant related information” at 46. The mean was 2.41 (SD = .77). Therefore, most of the pre-service teachers considered the importance of horticultural practice to be very high and the impact on horticultural learning to be very high. It was also found that there is a need for information about plants and information for direct application to learning.

Empirical trends in practical gardening training using ChatGPT

Means and standard deviations were calculated based on a 5-point scale to determine the pre-service teachers’ experience with ChatGPT in horticulture and their views on ‘views on horticulture education using ChatGPT’, ‘benefits of designing horticulture education using ChatGPT’, ‘thoughts on using ChatGPT in horticulture in the future’, and ‘the role of future teachers if AI chatbots such as ChatGPT are used in actual horticulture classes in the future’ (Table 6). The analysis showed that 90.4% of the respondents selected ChatGPT as “useful”, with a mean of 2.06 (SD = .98). Regarding the benefits of using ChatGPT to design practical horticulture education, 37.1% of the respondents answered that they could design customized lessons for students, 29% of the respondents answered that they could discover and use various materials that teachers had not thought of, and 24.2% of the respondents answered that they could save time and effort, with a mean of 2.43 (SD = 1.15). These experiences were found to have a significant impact on their gardening practice using ChatGPT, and 96.9% of respondents were very likely to use ChatGPT for gardening in the future. The mean score was 2.06 (SD = .66). Finally, when asked about the role of teachers in the future if A.I. chatbots such as ChatGPT were used in practical horticulture classes, 50% of respondents said that the role of teachers would become more important, and 46.8% said that it would be similar to today, with a mean of 2.53 (SD = .56). Thus, pre-service teachers expressed a high degree of usefulness of ChatGPT in practical horticultural education and indicated that most students would use ChatGPT in practice classes in the future. Based on the results of Chung and Bang’s (2015) study, which showed that agricultural literacy was low among college of education students, it was assumed that ChatGPT would be effective for pre-service teachers.

Problem-solving skills based on practical gardening training

The results of pre-service teachers’ problem solving ability after horticulture practicum are shown in Table 7. Among the 16 questions, “I believe I can solve most problems if I put in enough time and effort” has the highest mean of 4.148, followed by “When deciding how to solve a problem, I consider which option will be more effective” and “When making a decision, I evaluate and compare the results of each alternative” with a mean of 4.101. However, “Sometimes I get overwhelmed by emotions and can’t think of different ways to solve a problem” had a mean of 2.445, “When I encounter a problem, I don’t know if I can solve it” had a mean of 2.382, and “Sometimes I don’t have enough time to deal with a problem and feel overwhelmed” had a mean of 2.359, which is below 3 points, indicating that they have difficulty solving problems when they encounter them. This is because practical horticulture education requires complex problem solving skills, such as various forms of creativity and convergence by integrating information (Hong et al., 2013). ChatGPT can facilitate the search for various resources for practical learning and provide interactions with ChatGPT such as immediacy, continuity of questions, diversity of scope, and educational utility (Crompton, 2023; OpenAI, 2023).
Among the 16 questions related to problem solving skills in practical gardening training, the principal component analysis was conducted by the Varimax method, a right-angle rotation method, without specifying the number of factors, and after five repeated rotations, three factors with an eigenvalue of 1 or more were established. The first factor consisted of questions 5, 1, 2, 4, 11, 9, 7, 14, and 3; the second factor consisted of questions 8, 13, 10, and 12; and the third factor consisted of questions 16, 15, and 6. The loadings of the first factor ranged from .906 to .559, the loadings of the second factor ranged from .862 to .584, and the loadings of the third factor ranged from .896 to .837, and the reliability Cronbach’s α coefficients between each item were adequate (Table 8). Therefore, the concept of practical gardening training was established with these three factors, “knowledge”, “skills”, and “attitude” in practical gardening training (Table 8). Knowledge, skills and attitude are concepts derived from Bloom’s pedagogical taxonomy. Each concept complements the other as a taxonomy that systematizes the problem of “how to determine and present specific learning content at a given level of behavior” (Borich, 2000) (Table 9).

Correlation between ChatGPT characteristics, knowledge, skills and attitudes

Just as teachers’ knowledge, skills, and attitudes relate to teachers’ competence, pre-service teachers’ knowledge, skills, and attitudes need to be analyzed together and interpreted integrally. Therefore, it is necessary to consider the interrelatedness of knowledge, skills, and attitudes in problem solving ability in order to effectively utilize ChatGPT characteristics for practical gardening training (Koehler and Mishra, 2009). As mentioned earlier, the reliability of the three factors was analyzed, and the Cronbach’s a values of 0.921 for knowledge, 0.811 for skills, and 0.864 for attitudes were found to be acceptable for use as a scale. Therefore, we categorized various characteristics of ChatGPT presented in OpenAI (2023), Crompton (2023), Lund and Wang (2023), Lee (2023), and Kim and Lee (2023) into nine characteristics: Immediacy, Continuity of Questions, Adjustability of the Level of Responses, Suitability of Media to the Current Generation, Advantage of the Lack of Necessity to Ask Directly, Advantage of Conversational Knowledge, Expertise, Applicability of Educational Utilization, and Diversity of Range of Questions, and examined the relationship between knowledge, skills, and attitudes and ChatGPT characteristics using Pearson’s correlation analysis. The results are presented in Table 10.
Practical gardening knowledge and skills are all significant and positively correlated except for expertise, while attitude is negatively correlated with ChatGPT characteristics such as immediacy, continuity of questions, and adaptability of responses. First, the correlation between knowledge of gardening activities and ChatGPT characteristics was significant for all items except expertise, with immediacy having the highest correlation of r = .42 (p < .01). Next, the correlations between gardening skills and ChatGPT characteristics were statistically significant for all items except expertise. The highest correlation with technology was r = .44 (p < .01) for medium aptitude and r = .41 (p < .01) for conversational knowledge advantage. Finally, in the correlation between attitudes toward gardening activities and ChatGPT characteristics, only immediacy, continuity of questions, and adaptability of responses showed significant differences, with continuity of questions showing the highest correlation at r=.40 (p<.01). In summary, knowledge and skills in practical gardening training were moderately correlated with most ChatGPT characteristics and were positively correlated. In terms of attitude and ChatGPT characteristics, immediacy, continuity of questions, controllability of answers, and suitability of medium were negatively correlated, and the rest were relatively low (Table 10). This was recognized as a similar result to Kim’s (2023) study, which found that the use of ChatGPT is very advantageous for providing various teaching materials, discovering the latest horticultural materials, and creative lesson design, but there is a somewhat critical view on the role of the teacher and attitude change.

How knowledge, skills, and attitudes affect ChatGPT utilization

Multiple regression analysis was used to examine the effects of knowledge, skills, and attitudes on the use of ChatGPT, which are problem-solving skills based on practical gardening training of pre-service teachers, and the results are shown in Tables 11, 12, 13.
The fitted value of the regression line for the effect of knowledge, skills, and attitudes on immediacy in using ChatGPT is F = 5.872, which is statistically significant at the p < .001 level. Among the independent variables of immediacy, knowledge (β = .343, p < .05) had a significant effect, while skills (β = −.237, p < .057) and attitude (β = .019, p < .91) had no significant effect (Table 11). Therefore, the immediacy of ChatGPT appeared to be the most influential factor on knowledge, and it seems that questions related to various knowledge characteristics of horticultural education practice can be addressed immediately (Table 11). The continuity of questions was found to have a statistically significant effect at the p < .001 level with F = 8.131. The coefficient of determination for the regression analysis shows an R2 value of .296, indicating a model explanatory power of 29.6%. The most powerful tool of ChatGPT is the ability to continue the conversation, which has a significant impact on knowledge (β = .386, p < .017) and skills (β = −.291, p < .016). Attitude (β = −.004, p < .982) was not significant (Table 11). Regarding the adjustabiltiy of the level of responses, the regression model is interpreted as good with F = 3.754 at p < .016, and there is no statistically significant effect of knowledge (β = .118, p < .492), skill (β = −.244, p < .061) and attitude (β = .152, p < .384). The overall explanatory power is approximately 16.3% (R2 = .163) (Table 11). These results indicate that ChatGPT characteristics such as “immediacy” and “continuity of questions” have the greatest impact on the knowledge aspect of problem solving ability, especially the knowledge aspect. However, the impact of the attitude aspect is not significant, which is related to the ease of knowledge provision (Lung and Wang, 2023) and the possibility of personalized and safe learning in ChatGPT.
The fit of the medium to the generation is F = 5.009, which is statistically significant at the p < .004 level. The explanatory power of the regression model is 20%. Among the independent variables of media fit, knowledge (β = .077, p < .644), skills (β = −.069, p < .583) were not significant, while attitude (β = .368, p < .033) was significant (Table 12). Most people prefer to use a medium that suits them, that they are trained in, and that they have accepted, and ChatGPT seems to be a very suitable medium for the Millennials and Generation Z. The advantages of the lack of necessity to ask directly is statistically significant at the p < .040 level with F = 2.962. The regression model has an explanatory power of 13.3%. Knowledge (β = .237, p < .178), skills (β = .277, p < .086), and attitude (β = .161, p < .366) do not have a statistically significant effect (Table 12). The benefit of conversational skills was F = 6.412, p < .001, and the regression model was significant. It is not significant for knowledge (β = .288, p < .080), skills (β = .225, p < .067), and attitudes (β = .281, p < .092), but it is significant at p < .10. The total explanatory power is about 24.9% (R2 = .249) (Table 12). Among these results, especially the medium suitability for the current generation, which showed a significant effect in terms of attitude, indicates that ChatGPT is a suitable platform for digital native generation learners who are used to mobile and chat formats, and is effective in motivating learners with low horticultural literacy (Chung and Bang, 2015) and learning attitudes (Lee et al., 20–23).
Expertise (F = 1.490, p < .227) was not statistically significant. The regression model has an explanatory power of 0.72% and a modified explanatory power of 0.24%. Expertise had no significant effect on any of the independent variables: knowledge (β = .252, p < .167), skills (β = .168, p < .216), and attitude (β = .015, p < .933) (Table 13). From the ChatGPT characteristics, it can be concluded that expertise can be said to be accurate information about clear knowledge, but there is an overall difficulty in sifting through the information or finding literary information. The regression explained 18.1% (R2 = .181) of the total variance in educational use, with an F-value of 4.263, indicating that the regression is statistically significant (p < .009). Specifically, we found a statistically significant impact on knowledge (β =. 342, p < .048) and skills (β = .254, p < .049). There was no significant effect on attitude (β = .114, p < .509) (Table 13). These results suggest that ChatGPT is highly educational in terms of knowledge and skills. Regarding the diversity of range of questions, the analysis of variance to test the significance of the regression model was F = 4.701, p < .005, and the regression model was appropriate. The coefficient of determination of the regression analysis was .196, which means that the R2 value was .196, which explains 19.6% of the variables used in the statistics. The independent variables knowledge (β = .366, p < .033) and skills (β=.260, p < .042) have statistically significant effects. Attitude (β = .106, p < .534) was not statistically significant (Table 13). This suggests that a wider range of questions has a positive effect on knowledge and skills.
Of these outcomes, ‘expertise’, where knowledge, skills, and attitudes did not have a significant effect on problem solving, was attributed to the fact that the training data was not maximized due to the non-complete nature of the platform (Lund and Wang, 2023). Both ‘pedagogical usability’ and ‘diversity of question set’ showed significant effects on problem solving knowledge and skills (Table 13). This suggests that ChatGPT can provide learners with personalized feedback and a wide range of domain knowledge. These features were also considered as ChatGPT features that can be effectively used in horticulture education, which may have limitations due to experience, interest, and program materials (Lee et al., 2023).

Conclusion

In this study, we explored the possibility of teaching horticultural practice using ChatGPT and analyzed its impact. For this purpose, a survey of pre-service teachers was conducted to find out their perceptions on the use of ChatGPT in horticulture education. Based on the results, the discussion and conclusions are as follows.
First of all, as an empirical trend of practical gardening training, the importance of practical education was very high, and it was possible to customize the design for students, and the usefulness, lesson design ability, and utilization for future practical gardening training were also very high. It was found that there are difficulties related to problem-solving skills in horticultural practical training, and ChatGPT-based prompt engineering can be used to design lessons that reflect the latest educational trends, discover and apply various teaching materials, and save teachers’ time and effort (Kim, 2023). In addition, based on the problem-solving ability according to practical gardening training, it was classified into three factors such as knowledge, skills, and attitudes that determine the quality of education through factor analysis. Based on this, we analyzed the correlation between knowledge, skills and attitudes and ChatGPT characteristics, and found that knowledge and skills are all significant except for expertise, and have a positive correlation. In the case of attitude, immediacy, continuity of questions, and the ability to adjust the level of answers were significantly different, and the rest showed relatively low correlations. Since attitude emphasizes an individual’s internal motivation and values, it seems to have a low correlation in horticultural education.
We then conducted a regression analysis to determine the impact of knowledge, skills, and attitudes on the use of ChatGPT. The significant findings of the study were as follows.
Immediacy was shown to have an impact on knowledge. ChatGPT is a conversational artificial intelligence that demonstrates the potential of ChatGPT as a learning method for solving questions related to various knowledge attributes in gardening practice through immediate conversations (OpenAI, 2023). This experience helps students deepen their knowledge and create new knowledge. Growing plants is a task that requires constant information and guidance. ChatGPT can serve as such a guide, allowing students to have a continuous and ongoing learning experience, maximizing the effectiveness of their education. In addition, it has been shown that in horticultural therapy activities, it is difficult to transfer continuous plant knowledge, such as how to grow plants after practice, and the efficient use of ChatGPT can help a lot (Kim, 2023). The continuity of questions was found to have an impact on knowledge and skills. The knowledge gained in a ChatGPT conversation is continuous and can contribute to the depth of knowledge and skills (Crompton, 2023). In addition, a conversation can be revisited and continued at a later time, rather than at that moment, which differs from traditional questioning and can increase interest in the experience. Horticulture practicums are typically organized on a weekly, monthly, or quarterly basis, so the timing can vary greatly (Oh, 2019). Therefore, the use of ChatGPT has the beneficial property of connecting the dots of knowledge. In addition, gardening activities stimulate participants’ interest, curiosity, and questions about various plants, and the fact that they can ask questions to ChatGPT continuously or at different levels until they understand on their own, instead of relying only on the instructor, is considered to be very helpful in overcoming the limitations of gardening practice (Liu et al., 2023). Although it was found that there was no statistically significant effect on the ability to adjust the level of responses, it may be possible to facilitate the acquisition of expertise by requiring the level of responses in actual activities (Kim, 2023). In particular, by allowing pre-service teachers to ask questions at the level of elementary students using ChatGPT, it increased their interest in the lesson and helped them develop appropriate study skills.
Attitudes were found to have a significant impact on the appropriateness of the medium for the generation. ChatGPT can be used in education by utilizing a mobile and chat-based platform that is familiar to today’s generation of digital natives. While there was no statistically significant effect on the benefits of not having to ask direct questions and the benefits of conversational knowledge, students who are reluctant to ask their teachers direct questions may benefit from the ease with which they can ask questions through ChatGPT (OpenAI, 2023). Conversational knowledge, or the narrative approach, is a representative knowledge retrieval method of the new generation of AI, and ChatGPT differs from the existing one-way communication knowledge retrieval. In particular, ChatGPT, which allows users to ask questions in the language of their generation, will be effective in overcoming this psychological barrier because the existing materials in written language show a high schema level due to the characteristics of the Millennials and Generation Z with very low agricultural or horticultural literacy (Chung and Bang, 2015; Lee et al., 2023).
Expertise was not significant for knowledge, skills, or attitudes. Applicability of educational utilization was found to have a statistically significant effect on knowledge and skills. In a recent review of ChatGPT research, the most researched areas were medical and nursing education (Jang and So, 2023). In medical education, it appears that once the reliability of ChatGPT has been established, ChatGPT can be used for simulation-based learning (Gilson et al., 2023). In the case of horticultural practice, it is believed that learners can more easily apply problem-solving skills because it has technical features that can be interacted with based on conversations (Kim, 2023). In this regard, Kim (2023) developed a ChatGPT-based edible plant project learning program for elementary school students based on prompt engineering and used it very effectively to design horticulture lessons. The variety of questions in ChatGPT showed a statistically significant effect on knowledge and skills. It predicts the possibility of gradually diversifying the types of questions, such as experiments, practices, and use of new technologies, as well as curriculum-related questions covering all areas. Most importantly, it is equipped with a feature that remembers and learns from previous conversations with users, so it can accurately identify the context of the conversation, which can lead to a variety of questions (Kim, 2023). In fact, pre-service teachers used ChatGPT to ask and understand questions about horticulture activities, especially at the elementary school level, such as precautions for the activities, necessary practical materials and points to note in teaching, and the chronological division required for the classes.
In conclusion, we can see that ChatGPT can play an important role in knowledge, skills and attitudes when used appropriately in practical gardening training and is a variable that should be considered in the future. However, it is not enough to use ChatGPT alone. It is necessary to deepen and apply the knowledge so that learners can reorganize it in their own context and create new outputs based on it, so that the information gained through ChatGPT is not just “volatile” information which is likely to be decontextualized. Ultimately, it is important to actively introduce, review, and utilize new educational tools and strategies such as ChatGPT to ensure that horticultural practice is not just for students to receive knowledge, but to internalize and utilize it. Thus, true learning is achieved.
Finally, this study examines how pre-service teachers perceive the use of ChatGPT in practical gardening training in relation to the ChatGPT characteristics that emerged during the process. In addition, each variable was individually categorized to reveal its characteristics and provide basic data on the use of ChatGPT in practical gardening training. We expect this study to provide implications for the practice of horticultural education using AI, especially ChatGPT. However, it may be somewhat difficult to generalize the results of this study due to the lack of a survey of pre-service teachers at different universities by grade level. In the future, it is necessary to conduct additional studies with students in many universities and compare the performance of students who participated in horticulture practicum with those who did not.

Fig. 1
Study model.
ksppe-2023-26-6-693f1.jpg
Table 1
General characteristics of participants (n = 62)
Items n %
Gender Male 15 24.2
Female 47 75.8

Age Under 20s 10 16.1
20 to 23 years old 33 53.3
More than 23 19 30.6

Practice participation time Less than 5 hours 21 33.9
Less than 5–10 hours 29 46.8
More than 10 hours 12 19.4
Table 2
ChatGPT search training
Class Level of questioning Scope of a question


Level Content Level Content
1 General to Specific Accessing from higher knowledge to lower knowledge when entering prompts in ChatGPT. Users start with general questions and gradually enter prompts with specific questions Morphological and biological knowledge of plant Use specific plant names and variety names

2 Few shot learning A method of providing a few examples when learning a new task to a language model such as ChatGPT. In contrast, there are 0-shot learning (not providing an example) and 1-shot learning (only 1 example). Nurturing and technical knowledge of plants Clarify the cultivation conditions

3 General to Specific Few shot learning Combination of the concepts of 1 and 2 Other plant-related knowledge Ask questions in various fields including social, cultural, and educational knowledge
Table 3
Composition of questionnaire
Variable Sub-factor n
General characteristics Gender, Age, Practice participation time 3
Experience in practical gardening class Significance, effects of practical training, parts which are most difficult 3
Practical gardening training using ChatGPT Usefulness, advantages, intentions for utilization, roles of future teachers 4
Examination for problem solving ability Problem solving ability in practical gardening class 16
Characteristics of ChatGPT Immediacy, continuity of questions, adjustability of the level of responses, suitability of media, points of questions, advantages of conversational knowledge, expertise, applicability of educational utilization, diversity of range of questions 27
Total 53
Table 4
Problem-solving skills test for horticulture practical training
Questions Composition of questions Cronbach’s α
1 I can come up with creative methods to solve problems. .800
2 I formulate as many ways as possible to solve problems.
3 I believe that problems can be solved even if no solution is initially in sight.
4 The problems I encounter can be solved without help from anyone else.
5 If there is a problem, I can discover an answer that can solve it.
6 I sometimes lack the capacity to address problems and feel flustered.
7 When deciding solutions to problems, I consider which method will be more effective.
8 When making a decision, I evaluate and compare the results of each alternative.
9 When taking an action, I try to predict its consequences.
10 I believe that most problems can be solved with enough time and effort.
11 I discover new, innovative ways to solve problems.
12 I know systematic methods of comparing various alternatives and making decisions.
13 When I encounter a problem, I examine what things in my surrounding will help me solve it.
14 When there is a problem, I identify the situation first, and then review relevant information.
15 I sometimes become so engrossed in emotions that I am unable to devise various methods to resolve problems.
16 When I encounter a problem, I feel helpless about whether I can solve it.
Table 5
Characteristics of ChatGPT
Classification Composition of questions Cronbach’s α
Immediacy Questions that arise during practical training classes can be immediately addressed.
The use of ChatGPT can expand the continuity and scalability of experiences.
Interest in practical training activities has increased because questions for which answers are sought to can be addressed immediately.
.816
Continuity of questions The depth of knowledge increases since questions are asked in the form of a continuous conversation.
The acquired knowledge is also evoked and used for other experimental factors.
The depth of the experimental element is likely to increase because the information in ChatGPT conversations is recorded and managed.
.777
Adjustability of the level of responses The understanding of professional knowledge is broadened when practical training is conducted using ChatGPT.
Answers to questions using ChatGPT are educationally highly applicable because they are easy to explain to others.
Confidence in professional knowledge is created by directly adjusting and leading the level of questions and answers.
.843
Suitability of media to the current generation ChatGPT may be a suitable medium for Millennials and Gen Z because it comes in familiar forms of typing, touching, and chatting.
The ChatGPT method is made so that it is easy to ask and encourages people to wish to ask further.
I prefer a medium that is suitable for myself and easy for me to accommodate.
.816
Advantages of the lack of necessity to ask directly ChatGPT enables questions to be asked conveniently, and answers desired by the inquirer to be easily received.
The ChatGPT communication method enables the reception of answers by indirect language, chatting, and touching actions instead of actual language.
Interest is induced as the user may gain desired knowledge without asking questions in class.
.919
Advantages of conversational knowledge ChatGPT gains knowledge by bilateral communication.
ChatGPT’s friendly tone of communication induces more questions and helps the user acquire practical skills...
ChatGPT induces more questions with friendly and attractive answer, and effectiveness increases with the connection between practical training experience and knowledge.
.876
Expertise ChatGPT bears expertise and accuracy as a learning tool.
Professional information on practical gardening training can be easily found and utilized using ChatGPT.
I am confident that professional information on practical gardening training which I have not experienced can be found using ChatGPT.
.931
Table 6
Empirical trends in the use of ChatGPT in practical gardening classes
Classification n % M SD
Opinions on practical gardening classes using ChatGPT Very useful 22 35.5 2.064 .989
Useful 20 32.3
Moderately useful 14 22.6
Not useful -. -.
Not useful at all 6 9.7
Total 62 100.0
Advantages of planning practical gardening classes using ChatGPT Saves time and effort 15 24.2 2.435 1.154
Enables class plans to be customized for students 23 37.1
Enables class plans to reflect the latest trends in the field of agricultural education 6 9.7
Facilitates discovering and applying various class materials necessary for learning in practical training which teachers may not have called to mind 18 29.0
Others -. -.
Total 62 100.0
Intention to utilize ChatGPT in practical gardening training I absolutely do not intend to utilize it. 1 1.6 3.774 .663
I do not intend to utilize it. 1 1.6
I moderately intend to utilize it. 13 21.0
I intend to often utilize it. 43 69.4
I intend to always utilize it. 4 6.5
Total 62 100.0
If artificial intelligence chatbots such as ChatGPT are used in actual practical gardening class in the future, the roles of future teachers will be The importance of the roles of teachers will increase significantly -. -. 2.532 .564
The importance of the roles of teachers will increase 31 50.0
The importance of the roles of teachers will be similar to the present 29 46.8
The importance of the roles of teachers will decrease 2 3.2
The importance of the roles of teachers will decrease significantly -. -.
Total 62 100.0
Table 7
Descriptive statistics of problem-solving skills in practical gardening classes
Questions Composition of questions M SD
1 I can come up with creative methods to solve problems. 3.882 .769
2 I formulate as many ways as possible to solve problems. 3.867 .777
3 I believe that problems can be solved even if no solution is initially in sight. 3.851 .774
4 The problems I encounter can be solved without help from anyone else. 3.851 .710
5 If there is a problem, I can discover an answer that can solve it. 3.875 .878
6 I sometimes lack the capacity to address problems and feel flustered. 2.359 1.106
7 When deciding solutions to problems, I consider which method will be more effective. 4.101 .771
8 When making a decision, I evaluate and compare the results of each alternative. 4.101 .696
9 When taking an action, I try to predict its consequences. 3.984 .720
10 I believe that most problems can be solved with enough time and effort. 4.148 .804
11 I discover new, innovative ways to solve problems. 3.945 .890
12 I know systematic methods of comparing various alternatives and making decisions. 4.070 .733
13 When I encounter a problem, I examine what things in my surrounding will help me solve it. 4.046 .719
14 When there is a problem, I identify the situation first, and then review relevant information. 4.093 .747
15 I sometimes become so engrossed in emotions that I am unable to devise various methods to resolve problems. 2.445 1.175
16 When I encounter a problem, I feel helpless about whether I can solve it. 2.382 1.204
Table 8
Factor analysis of problem-solving skills in practical gardening classes
Item Question number Factor Analysis common variance

1 2 3
Knowledge 5 .906 .824
1 .813 .739
2 .789 .700
4 .737 .619
11 .717 .651
9 .709 .562
7 .704 .652
14 .655 .509
3 .559 .505

Skills 8 .862 .643
13 .731 .657
10 .680 .786
12 .584 .548

Attitude 16 .896 .808
15 .894 .810
6 .837 .746

Cronbach’s α .921 .811 .864
Eigenvalue 7.189 2.311 1.258
Explained variance ratio 44.93 14.44 7.86
Cumulative explained variance ratio 44.93 59.37 67.24
Table 9
B. Bloom’s taxonomy of education objectives
Category Contents Relationship with learners
Knowledge Refers to theoretical or practical understanding. The ability of learners to remember and understand specific information. Places focus on memorizing and understanding clear information, such as facts, concepts, and procedures. Contents which the learner knows about a particular subject
“What does one know?”
Skills Refers to the ability necessary to apply learned knowledge to actual situations or to solve problems. May be manifested in various forms such as critical thinking, analysis, communication, technical ability, etc. Practical aspects of learners
“How does one do it?”
Attitude Refers to internal attitudes toward individual values, beliefs, emotions, motivations, and emotions, etc. Concerns the way learners respond to and value new information or experience. Can include the attitude toward learning, cooperation, responsibility, passion, and respect. Affects learners’ internal motivations and values. Internal motivations and values of learners
“How does one feel or react?”
Table 10
Correlations between knowledge, skills, and attitudes and ChatGPT characteristics
Item Knowledge Skills Attitude
Immediacy .425** .339** −.341**
Continuity of questions .468** .365** −.401**
Adjustability of the level of responses .296* .315* −.327**
Suitability of media to the current generation .358** .445** −.210
Advantages of the lack of necessity to ask directly .286* .265* .106
Advantages of conversational knowledge .423** .412** .051
Expertise .215 .140 .090
Applicability of educational utilization .350** .274* .118
Diversity of range of questions .367** .281** .120

* p < .05,

** p < .01

Table 11
Impact of problem-solving knowledge, skills, and attitudes on ChatGPT immediacy, continuity of questions, and adjustabiltiy of the level of responses
Item Immediacy Continuity of questions Adjustability of the level of responses



B β t (p) B β t (p) B β t (p)
(Constant) 3.003 4.317 (.000) 2.888 4.119 (.000) 3.263 3.981 (.000)

Knowledge .382 .343* 2.098 (.040) .452 .386* 2.464 (.017) .149 .118 .692 (.492)

Skills −.155 −.237 −1.939 (.057) −.200 −.291* −2.489 (.016) −.180 −.244 −1.911 (.061)

Attitude .022 .019 .113 (.910) −.004 −.004 −.022 (.982) .197 .152 .877 (.384)

R2 .233 .296 .163
Adjusted R2 .193 .260 .119
F-value(p) 5.872(0.001) 8.131(0.001) 3.754(0.016)
Table 12
Impact of problem-solving knowledge, skills, and attitudes on ChatGPT’s suitability of media to the current generation, advantages of the lack of necessity to ask directly and advantages of conversational knowledge
Item Suitability of media to the current generation Advantages of the lack of necessity to ask directly Advantages of conversational knowledge



B β t (p) B β t (p) B β t (p)
(Constant) 2.329 3.374 (.001) .372 .319 (.751) −.327 −.335 (.738)

Knowledge .084 .077 .465 (.644) .415 .237 1.364 (.178) .454 .288 1.781 (.080)

Skills −.044 −.069 −.553 (.583) .234 .227 1.747 (.086) .209 .225 1.865 (.067)

Attitude .412 .368* 2.181 (.033) .290 .161 .911 (.366) .456 .281 1.712 (.092)

R2 .206 .133 .249
Adjusted R2 .165 .088 .210
F-value(p) 5.009 (0.004) 2.962(0.040) 6.412(0.001)
Table 13
Impact of problem-solving knowledge, skills, and attitudes on ChatGPT’s expertise, applicability of educational utilization, and diversity of range of questions
分类 Expertise Applicability of educational utilization Diversity of range of questions



B β t (p) B β t (p) B β t (p)
(Constant) 1.439 1.194 (.237) .200 .193 (.848) .024 .023 (.982)

Knowledge .442 .252 1.401 (.167) .551 .342* 2.022 (.048) .600 .366* 2.187 (.033)

Skills .173 .168 1.251 (.216) .240 .254* 2.010 (.049) .250 .260* 2.080 (.042)

Attitude .028 .015 .085 (.933) .189 .114 .664 (.509) .179 .106 .626 (.534)

R2 .072 .181 .196
AdjR2 .024 .138 .154
F-value(p) 1.490(0.227) 4.263(0.009) 4.701(0.005)

References

Ahn, T.U., I.A. Lee, Y.J. Kwun. 2022. The effect of entrepreneurship theory and practical education on career maturity: focusing on the mediating effect of problem solving ability. Asia-Pacific Journal of Business Venturing and Entrepreneurship. 17(3):269-280. https://doi.org/10.16972/apjbve.17.3.202206.269
crossref
Borich, G.D. 2000. Effective teaching methods New Jersey: Prentice-Hall.

Chi, S.A., C.Y. Kim, J.H. Jung. 2018. An analysis of the status and teachers’ awareness of horticulture education for young children. Early Childhood Education Research and Review. 22(5):191-217. https://doi.org/10.32349/ECERR.2018.10.22.5.191
crossref
Chun, J.E., E.S. Chung, N.I. Lee, A.E. Kim. 2023. Designing ChatGPT and Midjourney applications for various subjects. Journal of Korea Society of Computer Education. 27(1):117-118.

Chung, N.Y., G.H. Bang. 2015. he direction for revising 2015 national curriculum of the agricultural life science contents in rractical arts education. TJournal of Korean Practical Arts Education Research. 21(2):6-22. https://doi.org/10.17055/jpaer.2015.21.2.67
crossref
Crompton, H. 2023. Role playing challenging conversations Old Dominion University.

Gilson, A., C.W. Safranek, T. Huang, V. Socrates, L. Chi, R.A. Taylor, D. Chartash. 2023. How does ChatGPT perform on the united states medical licensing examination? The implications of large language models for medical education and knowledge assessment. JMIR Medical Education. 9(1):e45312. https://doi.org/10.2196/45312
crossref pmid pmc
Hong, S.M., H.J. Kim, G.W. Jo, G.B. Park, S.W. Kim. 2013. Talent development strategy in response to future science and technology talent Seoul: Science and Technology Policy Institute.

Jang, H.J., H.J. So. 2023. The analysis of research trends and topics about the educational use of ChatGPT. Journal of Research in Curriculum and Instruction. 27(4):387-401. https://doi.org/10.24231/rici.2023.27.4.387
crossref
Kim, S.G., S.Y. Sim, S.Y. Yoon, K.S. Han. 2023. Can ChatGPT be a consensual peer? - Use of ChatGPT in classifying question types. Journal of Gifted/Talented Education. 33(3):273-290.

Kim, D.S. 2023. ChatGPT prompt engineering Seoul: Communication Books.

Kim, H. 2023. Developing an edible plants project learning program of elementary school student using ChatGPT-4. Master’s thesis. Seoul National University of Education, Seoul, Korea.

Kim, H.Y., E.H. Lee. 2023. Correlation between practical education satisfaction, adaptability to college life, job-seeking anxiety in health-related college students in the untact era of COVID-19. Journal of the Korea Contents Association. 23(1):418-429. https://doi.org/10.5392/JKCA.2023.23.01.418
crossref
Kim, Y.H. 2014. Learning motivations, academic self-efficacy, and problem solving processes after practice education evaluation. Journal of the Korea Academia-Industrial cooperation Society. 15(10):176-186. https://doi.org/10.5762/KAIS.2014.15.10.6176
crossref
Koehler, M.J., P. Mishra. 2009. What is technological pedagogical content knowledge? Contemporary Issues in Technology and Teacher Education. 9(1):60-70.

Kwack, H.R. 2022. Development and application of curriculum convergence problem solving program for plant fine dust removal learning. Journal of Korean Practical Arts Education. 35(1):89-114.
crossref
Kwack, H.R., S.H. Jang, I.W. Park, W.C. Chun, M.G. Park. 2020. Development of interdisciplinary elementary integration programs to foster complex problem-solving skill. Korean Journal of Elementary Education. 31(4):343-360. http://doi.org/10.20972/kjee.31.4.202012.343
crossref
Lee, Y.J. 2023. An analysis of pre-service teachers’ mathematics lesson design using ChatGPT. Communications of Mathematical Education. 37(3):497-516.

Lee, Y.H., C.S. Kim, H.C. Ahn. 2023. A study on the chatGPT: Focused on the news big data service and ChatGPT use cases. Journal of the Korea Society of Digital Industry and Information Management. 19(1):1-13.

Lim, M.J. 2015. Horticultural activity program for improving school happiness for students in the department of agriculture in a specialized high school. Master’s thesis. Konkuk University, Seoul, Korea.

Lund, B.D., T. Wang. 2023. Chatting about ChatGPT: How may AI and GPT impact academia and libraries? Library Hi Tech News.
crossref
Ministry of Education. 2016 2015 Revised curriculum general commentary. Elementary School Retrieved from http://ncic.re.kr/mobile.index2.do.

Oh, E.S., Y.H. Kim. 2019. The core competencies of young children and the direction of early childhood education in the fourth industrial revolution. Journal of Digital Contents Society. 20(5):1011-1021. http://doi.org/10.9728/dcs.2019.20.5.1011
crossref
Oh, Y.A. 2019. Development of horticultural activity program for improving psycho-emotional stability and attention in elementary school students. Master’s thesis. Konkuk University, Seoul, Korea.

OpenAI. 2023 Teaching with AI Available at: Retrieved from https://openai.com/teaching-with-ai.

Seo, J.J., K.K. Jeong, Y.S. Kang. 2022. Effects of Gardening Activity-Oriented Environmental Education on Children’s Environmental Susceptibility and Nature-Friendly Attitudes. Journal of the Korea Academia-Industrial cooperation Society. 23(7):466-475. http://doi.org/10.5762/KAIS.2022.23.7.466
crossref
Yu, J.J. 2023. Application of artificial intelligence for geography education - Focusing on question answering on ChatGPT. Landscape and geography. 33(1):162-173. PMID: 10.35149/jakpg.2023.33.1.011

TOOLS
Share :
Facebook Twitter Linked In Google+ Line it
METRICS Graph View
  • 0 Crossref
  •    
  • 771 View
  • 17 Download
Related articles in J. People Plants Environ.


ABOUT
BROWSE ARTICLES
EDITORIAL POLICY
AUTHOR INFORMATION
Editorial Office
100, Nongsaengmyeong-ro, Iseo-myeon, Wanju_Gun, Jeollabuk-do 55365, Republic of Korea
Tel: +82-63-238-6951    E-mail: jppe@ppe.or.kr                

Copyright © 2024 by The Society of People, Plants, and Environment.

Developed in M2PI

Close layer
prev next