Keywords

1 Introduction

AI chatbots have emerged as versatile tools with applications in various fields, showcasing their potential to revolutionize communication and generate possibilities through the use of generative AI. The emergence of chatbots has brought about a new way of viewing communication, especially when considering the possibilities that can be unleashed using generative AI tools. The introduction of ChatGPT in November 2022, as well as the access to other chatbots such as Perplexity, Bing, or Bard, has sparked numerous perspectives and proposals for its application in various fields, including education, professional settings, creative endeavours, or the arts, among many others.

Chatbots can prove especially valuable as consultation channels for managing critical situations in a professional environment. These situations can arise in various contexts such as handling complaints, offering apologies, resolving conflicts, facilitating brainstorming sessions, discussing key aspects of a project, negotiating, designing, or conducting a marketing campaign. The amalgamation of personal creativity with the information provided by chatbots undoubtedly enhances the outcome, leading to a more comprehensive and effective work.

2 Context and State of the Art. A Review of the Literature on Generative AI and Chatbots for Language Learning

The utilization of chatbots in lecturing presents both advantages and limitations that should be considered. We need to convey to our students that their interaction with AI tools goes beyond simply receiving the output of their request through a prompt. Instead, we take on the role of reviewers and validators of the generated text. Therefore, it is essential for tourism students to understand the specific requirements of the text they need. These generative AI apps continuously improve through feedback and training provided by users and developers, allowing the generated texts to surpass previous limitations and shortcomings. However, it is important to note that the outputs or results produced by these apps are not to be automatically accepted. Hence, we stress the importance of verifying every outcome, drawing on our own knowledge of the required material.

In essence, while generative AI apps provide aid, we hold the ultimate authority in confirming and approving the result. Schmidt and Strasser (2022) highlight that some students may lack access to digital technology, or even have limited literacy; they also point out a risk that technology could replace or reduce real human interaction and opportunities to speak and listen. Lecturers will adapt technology to their own methods to ensure that the experience is being effective and stimulating.

2.1 Enhancing Language Learning. Tailoring Chatbot Usage for Language Tourism Education

To fully capitalize on the potential benefits for students, it is essential to foster self-learning automated processes. By instilling such habits in graduating students, they will be better equipped to continue their professional development beyond the academic setting. Li et al. (2022) highlight the need for conversational skills and support the lack of access to qualified native instructors with large-conversational data set, grammar correction, and reinforcing the system to make it more adaptive to individual user profiles. An experiment using online language learning tools showed an 84% feeling more confidence after the interactions. This may support the recognition of chatbots for increasing self-confidence in individual habits of improving fluency, as there are apps that contributes to these individual habits of learning and improving second language performance. Thus, Nykon (2023) mentions Duolingo, Rosetta Stone, Babbel, Busuu, Mondly, Memrise, or ChatGPT. AI models can replicate real speech and conversations with increasing accuracy, and language learning chatbots can respond to messages with personalized, relevant information.

The importance of considering lecturers’ feelings on their use of AI apps, in addition to learners’ perceptions, may be overlooked in specialised English learning. The need for greater attention to users experience in educational settings should be highlighted, mainly when the experience is still in progress with the newest chatbots. Belda-Medina and Calvo-Ferrer (2022) state that knowledge, level of satisfaction and perceptions are to be examined among educators, as there are practical implications for future educators, given for granted the need for good information about different chatbots and their benefits and limitations in language learning. It is pointed out a gap between the preparation of language lecturer candidates and the recent advancements in AI application to language learning, highlighting the need for improved training in the curriculum of English as a Foreign Language, hereafter EFL. Chatbots can be effectively utilized in the subjects of English for specific purposes, hereafter ESP, in tourism to enhance learning outcomes and facilitate engagement among students. Chatbots have shown effectiveness in providing personalized and interactive language learning experiences, offering immediate feedback and support in various language context, as chatbots can support self-directed learning, allowing individuals to improve their reading competence and develop skills such as questioning, answering, conversing, quizzing, and generating ideas or narratives (Dokukina & Gumanova, 2020). However, it is important to acknowledge the limitation of communication and potential challenges that may arise during interaction.

Training tailored chatbots to specific purposes and proficiency levels of students may enable structured interactions; these chatbots could be referred to as “intelligent tutoring systems” (Dokukiva & Gumanova, 2020). Given that the subject of English taught in the program of Bachelor’s Degree in Tourism is one of the language for specific purposes subjects, there is a need for specifically designed chatbots, appropriately trained to support effective progress in self-learning English for the tourism industry, as well as their adequate utilization in the lesson setting.

Consideration should be given to the intercultural contexts in which students at Malaga University typically engage in their professional lives. Malaga is widely recognized as an internationally sought-after tourist destination, and this characteristic extends to other places in Spain and many other international destinations. Given the prevalence of intercultural interactions, it is crucial to incorporate this aspect into the data used for chatbot consultations. Zhao (2023) emphasizes the significance of incorporating courses on culture and global competence, highlighting the importance of a multi-dimensional education that encompasses knowledge, skills, attitudes, and values, which proves successful when tackling global issues or navigating intercultural situations. By including a diverse range of cultures in the database, the motivation of tourism students can be enhanced, as it allows them to develop the necessary skills for effective performance in cross-cultural contexts.

2.2 Overcoming the Lack of Experience in Using This Technology in Teaching

This knowledge gap in the EFL curriculum regarding the integration of AI in language lecturing, particularly in preparing language lecturers, highlights the need for better training in the EFL curriculum to bridge the disconnection between the preparation of language lecturers and the recent advances in AI application to language learning. AI chatbots may improve EFL learners’ mastery, fluency, and skills. Therefore, lecturers must face AI-powered technologies in foreign language to ensure their students receive the best education experience.

Support and training are also required to enhance willingness and self-reliability for effective use in lessons. For Dittmeyer (2023) AI offers a significant advantage as it enables the system to determine the appropriateness of tasks efficiently and systematically for individual learners based on their respective skill levels and learning trajectories. AI can also assist lecturers and textbook writers in selecting and customizing educational materials; this capability allows lecturers to identify suitable texts, modify them as needed, or even generate entirely new texts to offer students a range of linguistically suitable resources for skill development. The potential for personalized learning afforded by AI technology is a transformative advancement in the field.Footnote 1

Thus, to effectively leverage the potential of chatbots for English for specific purposes learners, prioritizing lecturer training on AI integration in lessons is a must. Wahyuni (2022) proposes the integration of chatbots in the lessons as an alternative and collaborative solution for English language learners, and highlights the benefits of incorporating chatbots in English language learning, particularly in terms of providing feedback and obtaining responses during interaction. The impact of chatbots on learners’ confidence and specific language acquisition is in close relationship with enhanced fluency levels and conversational support.

Recognizing the potential risks involved in heavily relying on ChatGPT should be emphasized; both students and instructors must exercise caution when incorporating AI tools for learning purposes. Jalil et al. (2023) focus on ChatGPT and its performance in answering curriculum-based questions. The study reveals that while the responses generated by ChatGPT were not always 100% accurate, they did provide valuable insights; the framing of the problem significantly influences the accuracy of the responses and explanations. By providing additional information, the answers improved progressively, showcasing a trajectory from initial incorrect responses to modified versions until the desired answer was achieved.

2.3 Chatbots as Collaborative Solutions for English Language Learning

Prompt engineering is a relatively new discipline that refers to the practice of developing and optimizing prompts to effectively utilize large language models, particularly in natural language processing tasks (Giray, 2023). Prompt engineering has shown potential in enhancing the performance of chatbots withing educational environments. For achieving optimal outcomes in tourism ESP subjects, the inclusion of prompt engineering as a component of competence training content is recommended. Following with Jalil et al. (2023), incorrect responses were characterized by the chatbot's lack of knowledge (due to its database only covering until 2021 in version 3.5). This lack of information led to a lack of definitions for certain terms or a misunderstanding of certain concepts. In other instances, ChatGPT incorrectly focused on irrelevant aspects of the question instead of addressing what was important. There were also cases where ChatGPT made erroneous assumptions due to a lack of understanding of the requested information. In these situations, ChatGPT's responses were incorrect. This research warns us about the need to make our students aware of these limitations and emphasize the use of chatbots as assistants, and insist on good prompting engineering training, as said above. Given the possibility of obtaining incomplete or incorrect answers, our students should develop the ability to make the most of these tools without taking unnecessary risks. They should also be cautious about relying on limited, stereotyped, or non-inclusive responses.

ChatGPT can perform effectively in tasks that involve reasoning, which can lead to an improvement in learners’ reasoning abilities, ultimately reflecting in their conversational skills. However, the way the prompt is framed remains one of the main challenges, and the impact of prompt engineering needs to be taken into account. For Qin et al. (2023) the datasets encompass a wide range of tasks including reasoning, natural language inference, question answering, dialogue, summarization, named entity recognition, and sentiment analysis. Topsakal and Topsakal (2022) have specifically evaluated the performance of ChatGPT in generating dialogues for language learning purposes; they have leveraged the capabilities of ChatGPT to generate interactive dialogues that can be integrated with platforms like Google DialogFlow, successfully testing ChatGPT's ability to create dialogues in a format compatible with DialogFlow, which facilitates its practical usage.

Hallucination problems may arise, and reasoning is not reliable. However, ChatGPT performance can be enhanced  through interactive engagement with users; other possibilities, such as multitasks capacities, are still under study.

2.4 Ensuring Ethical Uses and Integrity in Professional Competences

These statements suggest the need to provide students with training in crafting suitable prompts to enhance ChatGPT's performance. To enable language models to showcase their abilities, sophisticated prompt engineering is required. However, previous language models only allow for a single exploration, meaning that the target outcome can vary significantly with minor changes in the prompt instruction.

The practical implications of conversational AI models such as ChatGPT extend to various industries, offering the potential to enhance customer service, improve language translation, and generate creative content. However, their deployment also raises significant ethical and social concerns, including the potential perpetuation of biases and stereotypes as well as their impact on employment. To address these issues, it is crucial to develop these systems in an ethical and socially responsible manner, which involves utilizing diverse and representative training data and conducting regular bias testing (Mattas, 2023). Moreover, the incorporation of gamification elements and surprise rewards can enhance the engagement and enjoyment of the learning process. These elements can also encourage learners to explore different features of the software and participate in new activities, promoting a more interactive and immersive learning experience.

Susnjak (2022) highlights ChatGPT's ability to mimic human-like text generation and perform complex cognitive tasks. However, this raises concerns about the potential threat it poses to the integrity of online exams. The paper suggests solutions such as invigilated and oral exams, advanced proctoring techniques, and AI-text output detectors, although they are not foolproof. Further research is needed to fully understand the implications of large language models like ChatGPT and develop effective strategies to prevent cheating. Educators and institutions must be aware of the possibility of ChatGPT being used for cheating and take measures to ensure fairness and validity in online exams for all students.

The potential adverse societal consequences of ChatGPT, including the propagation of fake news, plagiarism, and concerns regarding social security are emphasized by Guo et al. (2023). To assess these impacts, the authors gathered a substantial dataset consisting of tens of thousands of comparison responses encompassing a wide range of topics, including open-domain, financial, medical, legal, and psychological domains. Through comprehensive human evaluations and linguistic analyses, the study aims to elucidate the inherent disparities between human-generated content and that generated by ChatGPT.

Humans can understand hidden meanings based on their common sense and knowledge, while ChatGPT relies solely on the literal words of the question. In terms of objectivity, ChatGPT provides objective answers, whereas humans often express subjective opinions. In terms of safety and neutrality, ChatGPT tends to generate safer, more balanced, and neutral texts compared to humans. When it comes to specificity, human answers are often more detailed and include citations from sources like legal provisions, books, and papers, particularly in fields such as medicine, law, and technology. For instance, Kung et al. (2023)’s exam of the implications of qualitative and quantitative feedback on ChatGPT's performance and clinical reasoning abilities. AI models like ChatGPt show potential in assisting with text production, but cannot replace human grading and evaluation.

ChatGPT's responses are typically more formal, while humans may use more colloquial language. Finally, ChatGPT's responses are less expressive in terms of emotion, whereas humans often convey their feelings using punctuation and grammar features. These differences underscore the unique characteristics of human and ChatGPT responses across various dimensions. However, further development is needed to overcome limitations and enhance the model's capabilities.

3 Methodology. Literature Review and Bibliometric Analysis

Prior to this study, there is a notable dearth of original contributions concerning the practical implementation of these technologies within the realm of tourism education. The existing body of literature lacks substantive empirical data to substantiate the efficacy and utility of chatbots and AI in the pedagogical context, particularly within the unique context of English learning in tourism studies.

Recognizing the conspicuous gaps in the literature regarding the application of chatbots and AI in the field of education, this deficiency is especially pertinent within the domain of tourism. Our methodology for this contribution commences with an exhaustive review of the existing literature in the field. This literature review aims to elucidate the theoretical underpinnings and potential benefits of integrating chatbots and AI in educational settings within the specific context of tourism studies.

In addition to our comprehensive literature review, we employ a bibliometric analysis as a crucial methodological approach. Bibliometric analysis is instrumental in identifying the most influential authors, institutions, and publications within the field of chatbots and AI in education. By integrating bibliometric analysis into our methodology, we seek to enhance our understanding of the current state of research, although we still require adequate feedback to refine our objectives in the present article.

This bibliometric analysis serves as the cornerstone upon which we construct our research framework and proposal for the incorporation of chatbots as a valuable tool to provide insights into the prevailing trends in English language education within the tourism context.

It is essential to highlight the notable absence of original contributions by the author in terms of personal publications, practical experience, and empirical findings related to the integration of chatbots and AI in English language education in the context of tourism. However, through this study, we aim to address this gap and contribute to a more comprehensive and data-driven assessment of the integration of chatbots and AI in the unique field of English for tourism education.

4 Results. Innovating English Lessons with AI Chatbots: Challenges and Opportunities

With the advent of generative AI, the landscape in which university graduates will embark on their future professions is being reconsidered. It is necessary to reassess lecturing methods, taking this situation into account, and integrating the use of these tools during the lesson, aiming to anticipate situations that require the use of the English language, occasionally incorporating generative AI tools.

By leveraging the educational framework offered by the Bachelor's Degree in Tourism at Malaga University, a novel perspective on engaging in the professional realm of tourism is introduced, as the subjects on English are focused on acquiring a high level of proficiency in a specialised and demanding context, serving as a bridge between the academic realm and the graduates’ future professional lives. While these subjects must adhere to the parameters outlined in the degree verification report, the delivery of its contents allows for the integration of cutting-edge technology, hereby enhancing the effectiveness of the learning process and, above all, improving the future English proficiency of the graduates in their professional tourism environments.

5 Conclusions

The importance of incorporating technology in lecturing English in the field of tourism at the University cannot be overstated. University lecturing involves developing our students’ competencies as well as their knowledge of content so that they can effectively perform the professional tasks they will face. The use of chatbots and generative AI apps enables the application of these new tools to focus on the competencies taught in university degrees. The English lecturing team specialised in tourism at the University of Malaga is aware of this new paradigm and is actively exploring the integration of these tools, while considering specific criteria:

1. The professional trajectory of students in the field of tourism can experience substantial enhancements through the utilization of applications aimed at assisting them in its design. Websites such as Europassport have emerged as indispensable platforms for personal promotion, enabling the creation of curriculum vitae and the generation of accompanying documents including cover letters, interview preparation materials, language passports, and more. Professional social networks like LinkedIn also serve as highly effective platforms for candidate and company interaction, service promotion, and more. In recent years, applications like Canvas or Trello have contributed to enhancing professional performance. In addition to these existing tools, uncoded generative AI has emerged, allowing professionals to rely on them as assistants for generating content such as text, images, videos, audio, and more. These apps require proper handling when requesting information through descriptors or prompts, especially when dealing with conversational chatbots. Lesson training for proficient and professional utilization of these prompts is essential for adapting to the latest technologies and achieving optimal professional performance.

2. In the tourism industry, the texts generated are characterized by their technical nature, whether in marketing, commercial letters (complaints, apologies, information, confirmation, etc.), reports, memos, minutes, etc. As a textual typology, technical discourse is recognized for its clear messaging, use of specific formulas, concise sentences, or the incorporation of bullet points, among other features. If a tourism professional utilizes a chatbot like ChatGPT, Bing, Bard, Perplexity, etc. they must ensure that the resulting text maintains the specific characteristics required for that type of text generation. This is one of the reasons for which we must convey to our students that chatbots are not substitutes for our work. Consistently using this type of tool will gradually alleviate the more tedious and repetitive aspects of certain functions, such as generating technical documents.

3. The potential of integrating AI chatbots in these ESP subjects is worth exploring. Students in the Bachelor’s Degree in Tourism, particularly in their English subjects that cover tourism management content and skills, can explore various applications of AI in their respective contexts. One such application involves the use of AI apps to comprehend user profiles and, subsequently, provide tailored products, services, or content based on their individual needs, interests, and preferences.

For tourism professionals, it is crucial to utilize appropriate prompts to gather user profile information that aligns with the specific demands of each situation, being, in fact a matter of highly competent communication skills. It may be emphasized that the present contribution represents one of the pioneering efforts in this area, with limited prior feedback or experiences documented by the author herself.