Chatbot Development Research Articles

Adaptive Chatbots: Real-Time Sentiment Analysis for Customer Support

In the era of digital transformation and increasing online interactions, customer support is a critical aspect of business success. This paper investigates the development of adaptive customer support chatbots that use real-time sentiment analysis to generate contextually appropriate responses. By leveraging advanced sentiment detection techniques, the system aims to enhance user interaction, satisfaction, and overall customer service experience. This innovation is particularly relevant in today's fast-paced, digitally connected world where personalized and empathetic customer service can significantly impact brand loyalty and customer retention. The proposed approach addresses the growing demand for more intelligent and emotionally aware chatbots, aligning with current trends in artificial intelligence and consumer expectations.

AI Chatbot integration in SME marketing platforms: Improving customer interaction and service efficiency

This review paper examines the integration of AI chatbots in small and medium-sized enterprise (SME) marketing platforms, focusing on their impact on customer interaction, service efficiency, and future implications. AI chatbots, powered by advancements in natural language processing and machine learning, have revolutionized customer service by automating routine tasks, providing 24/7 availability, and personalizing interactions based on user data. The paper explores historical developments, current trends, and applications of chatbots in SMEs, highlighting benefits such as cost efficiency and improved service processes. Furthermore, it discusses emerging technologies in chatbot development and outlines long-term benefits for SMEs, including enhanced customer satisfaction and operational efficiencies. Practical recommendations for SMEs considering chatbot integration are provided to optimize implementation and leverage this technology effectively in their marketing strategies. Keywords: AI Chatbots, SME Marketing, Customer Interaction, Service Efficiency, Automation, Future Trends.

Ethical Considerations in Human-Centered AI: Advancing Oncology Chatbots through Large Language Models.

The integration of chatbots in oncology underscores the pressing need for human-centered AI that addresses patient and family concerns with empathy and precision. Human-centered AI emphasizes ethical principles, empathy, and user-centric approaches, ensuring technology aligns with human values and needs. This review critically examines the ethical implications of employing Large Language Models (LLMs) like GPT-3 and GPT-4 in oncology chatbots. It examines how these models replicate human-like language patterns, impacting the design of ethical AI systems. The paper identifies key strategies for ethically developing oncology chatbots, focusing on potential biases arising from extensive datasets and neural networks. Specific datasets, such as those sourced from predominantly Western medical literature and patient interactions, may introduce biases by over-representing certain demographic groups. Moreover, the training methodologies of LLMs, including fine-tuning processes, can exacerbate these biases, leading to outputs that may disproportionately favor affluent or Western populations while neglecting marginalized communities. By providing examples of biased outputs in oncology chatbots, the review highlights the ethical challenges LLMs present and the need for mitigation strategies. The study emphasizes integrating human-centric values into AI to mitigate these biases, ultimately advocating for the development of oncology chatbots that are aligned with ethical principles and capable of serving diverse patient populations equitably.

Designing chatbots in language classrooms: an empirical investigation from user learning experience

Amidst the technological evolution shaping the landscape of education, this research critically examines the imperative factors influencing the design of language-teaching chatbots in Thai language classrooms. Employing a comprehensive two-pronged methodology, our study delves into the intricacies of chatbot design by engaging with a diverse participant pool of pre-service teachers, in-service teachers, and educators. The study included the chatbot design opinion survey, a validated scale assessing attitudes toward chatbot design, and semi-structured interviews with teachers, educators, and experts to offer qualitative insights. Our findings reveal five key factors influencing chatbot design: learner autonomy and self-directed learning, content and interaction design for language skill development, implementation and usage considerations, alternative learning approaches and flexibility, and content presentation and format. Concurrently, thematic analysis of interviews results in five overarching themes: enhancing interactive language learning, motivational learning experience, inclusive language learning journey, blended learning companion, and communicative proficiency coaching. These findings inform the development of effective chatbots for Thai language classrooms within the evolving landscape of technology-driven education. These findings inform the development of effective chatbots for language classrooms within the evolving landscape of technology-driven education.

Arabic Chatbots Challenges and Solutions: A Systematic Literature Review

Since the beginning of Natural Language Processing, researchers have been highly interested in the idea of equipping machines with the ability to think, understand, and communicate in a human-like manner. However, despite the significant progress in this field, particularly in English, Arabic research remains in its early stages of development. This study presents a Systematic Literature Review on challenges faced in Arabic chatbot development and the proposed solutions. Utilizing the search terms "ARABIC," "CHATBOT," "CHALLENGES," and "SOLUTION," (including synonyms) we systematically surveyed studies published between 2000 and 2023 from Scopus, Science Direct, Web of Science, PubMed, SpringerLink, IEEE Xplore, ACM, Ebesco, and ICI. Moreover, besides Google Scholar and ResearchGate, we employed a manual snowballing technique to discover supplementary relevant research by examining the references of all chosen primary studies. The included studies were assessed for eligibility based on the quality assessment checklist we developed. Out of 3,891 studies, only 64 were deemed eligible. Many challenges were identified, including the scarcity of well-structured datasets. To overcome this (n=35) studies manually collected and preprocessed data. Additionally, Arabic language complexity led (n=53) researchers to adopt pre-scripted rules approaches, followed by generative approaches (n=10), and hybrid approach (n=1). Furthermore, (n=27) studies employed human-based evaluation metrics to assess the chatbot performance. while, (n=11) studies haven’t used any metrics. Based on conducted research, a critical research priority is providing Arabic with high-quality resources, such as an Arabic dataset that includes dialectal variations and incorporates empathy, lexicon corpora, and also a word normalization library. These resources will enable the chatbot to interact more naturally and humanely. Additionally, hybrid approaches have shown promising results, particularly in low-resource languages, such as Arabic. Therefore, more focus should be dedicated into implementing hybrid approaches in chatbot development. Furthermore, evaluating the chatbot performance is still an open domain for further research and contribution, highlighting the need for innovative standardized evaluation methods.

Effectiveness of chatbots in increasing uptake, intention, and attitudes related to any type of vaccination: A systematic review and meta-analysis.

This systematic review and meta-analysis analyzed and summarized the growing literature on the effectiveness of chatbot-delivered interventions in increasing uptake, intention, and attitudes related to any type of vaccination. We identified randomized controlled studies (RCTs), quasi-experimental studies, and non-experimental studies from the following platforms: PubMed, Web of Science, MEDLINE, Global Health, APA PsycInfo, and EMBASE databases. A total of 12 eligible studies published from 2019 to 2023 were analyzed and summarized. In particular, one RCT showed that a chatbot-delivered tailored intervention was more effective than a chatbot-delivered non-tailored intervention in promoting seasonal influenza vaccine uptake among older adults (50.5% versus 35.3%, p = 0.002). Six RCTs were included in the meta-analysis to evaluate the effectiveness of chatbot interventions to improve vaccination attitudes and intentions. The pooled standard mean difference (SMD) of overall attitude change was 0.34 (95% confidence intervals [CI]: 0.13, 0.55, p = 0.001). We found a non-significant trivial effect of chatbot interventions on improving intentions of vaccination (SMD: 0.11, 95% CI: -0.13, 0.34, p = 0.38). However, further evidence is needed to draw a more precise conclusion. Additionally, study participants reported high satisfaction levels of using the chatbot and were likely to recommend it to others. The development of chatbots is still nascent and rooms for improvement exist.

It’s better than nothing: The influence of service failures on user reusage intention in AI chatbot

Artificial intelligence (AI) chatbot have become increasingly popular as a tool for improving employee productivity over the last few years. In the early stages of AI chatbot development, exploring the impact of AI chatbot service failures on user reusage intention is useful for coordinating human–computer interaction and optimizing AI chatbot service mechanisms. The extant literature on AI service failures focuses on service recovery and anthropomorphism. There is less literature comparing different types of service failures and their effects. The article includes three studies. First, a randomized group experiment was conducted with 120 respondents. The results showed significant differences in the impact of different AI chatbot service failures on user reusage intentions. Second, an online questionnaire was completed by 386 respondents, the results found specific impact mechanisms of service failures on user reusage intentions. Third, an interview survey was conducted with 15 customers using AI chatbots to verify the findings of Study 1 and Study 2. Furthermore determine the boundary conditions for the unsupported hypotheses through meta-inference. The research enriches the literature on relationship marketing and expands the attribution theory of service failures. In addition, which provides theoretical basis and practical support for companies to reduce adverse effects of service failures.

Penerapan Uji-T Independen untuk Sistem Chatbot Gaotek

Artificial intelligence has become a primary focus in the development of computer systems that mimic human brain functions. One application is the development of Chatbots to assist in various fields. In this context, the research aims to evaluate the effectiveness of two types of Chatbots, namely Chatbot Gemini and ChatGPT, in providing customer service at GAOTek Inc. Through the application of Independent T-tests, the research findings indicate a significant difference in the performance of the two Chatbots. ChatGPT exhibits a lower average score compared to Chatbot Gemini, demonstrating significant efficiency superiority. The Independent T-test statistical results show the rejection of the null hypothesis, indicating a significant difference between the two groups. The alternative hypothesis is accepted, confirming a distinction in the performance of the two Chatbots. Thus, ChatGPT is declared as the most efficient Chatbot in the context of customer service at GAOTek Inc. These results are expected to provide valuable insights for the development and implementation of Chatbots in enhancing customer service effectiveness across various industries, particularly at GAOTek Inc.

Human–AI communication in initial encounters: How AI agency affects trust, liking, and chat quality evaluation

Artificial intelligence (AI) agency plays an important role in shaping humans’ perceptions and evaluations of AI. This study seeks to conceptually differentiate AI agency from human agency and examine how AI’s agency manifested on source and language dimensions may be associated with humans’ perceptions of AI. A 2 (AI’s source autonomy: autonomous vs human-assisted) × 2 (AI’s language subjectivity: subjective vs objective) × 2 (topics: traveling vs reading) factorial design was adopted ( N = 376). The results showed autonomous AI was rated as more trustworthy, and AI using subjective language was rated as more trustworthy and likable. Autonomous AI using subjective language was rated as the most trustworthy, likable, and of the best quality. Participants’ AI literacy moderated the interaction effect of source autonomy and language subjectivity on human trust and chat quality evaluation. Results were discussed in terms of human–AI communication theories and the design and development of AI chatbots.

Integrating discourse features and response assessment for advancing empathetic dialogue

Empathetic response generation is a crucial task in natural language processing, enabling emotionally resonant machine–human interactions. In this paper, we introduce the InfRa (Integrating Discourse Features and Response Assessment) model to address limitations in traditional methods for this task, such as the lack of deep dialogue comprehension and response control. InfRa integrates discourse features to augment structural dialogue understanding, with a novel edge pruning and mutual information learning module to further refine the representation. The model also employs a response evaluation module for dynamic optimization, ensuring emotional and semantic consistency between the generated response and its context. Our experiments demonstrate that InfRa outperforms existing baselines, reducing the Perplexity (PPL) score by approximately 9 points and excelling in all three fine-grained aspects of human evaluation. This research not only advances the development of empathetic chatbots but also provides valuable insights for broader text generation tasks.

Cultural heritage digital twin: modeling and representing the visual narrative in Leonardo Da Vinci’s Mona Lisa

In this paper, Artificial Intelligence/Knowledge Representation methods are used for the digital modeling of cultural heritage elements. Accordingly, the new concept of digital cultural heritage twin is presented as composed of a physical component and an immaterial component of the cultural entity. The former concerns the physical aspects, i.e. style, name of the artist, execution time, dimension, etc. The latter represents the emotional and intangible aspects transmitted by the entity, i.e. emotions, thoughts, opinions. In order to digitally model the physical and immaterial components of the twin, the Narrative Knowledge Representation Language has been formally introduced and described. It is particularly suitable for representing the immaterial aspects of the cultural entity, as it is capable of modeling in a simple but rigorous and efficient way complex situations and events, behaviours, attitudes, etc. As an experiment, NKRL has been adopted for representing some of the most relevant intangible items of the visual narrative underlying the hidden painting that lies beneath the Mona Lisa (La Gioconda) image painted by Leonardo Da Vinci on the same poplar panel. Real-time application of the resulting knowledge base opens up novel possibilities for the development of virtual objects, chatbots and expert systems, as well as the definition of semantic search platforms related to cultural heritage.

Analysis of the Effectiveness of Model, Data, and User-Centric Approaches for Chat Application: A Case Study of BlenderBot 2.0

BlenderBot 2.0 represents a significant advancement in open-domain chatbots by incorporating real-time information and retaining user information across multiple sessions through an internet search module. Despite its innovations, there are still areas for improvement. This paper examines BlenderBot 2.0’s limitations and errors from three perspectives: model, data, and user interaction. From the data perspective, we highlight the challenges associated with the crowdsourcing process, including unclear guidelines for workers, insufficient measures for filtering hate speech, and the lack of a robust process for verifying the accuracy of internet-sourced information. From the user perspective, we identify nine types of limitations and conduct a thorough investigation into their causes. For each perspective, we propose practical methods for improvement and discuss potential directions for future research. Additionally, we extend our analysis to include perspectives in the era of large language models (LLMs), further broadening our understanding of the challenges and opportunities present in current AI technologies. This multifaceted analysis not only sheds light on BlenderBot 2.0’s current limitations but also charts a path forward for the development of more sophisticated and reliable open-domain chatbots within the broader context of LLM advancements.

Scenario-Based Approach to AI’s Agency to Perform Human-Specific Tasks

Abstract The development of service robots and chatbots has changed the way companies interact with their customers. Several automated tasks have been taken over by AI, such as communication with customers, processing of orders and also other automated tasks. To measure the impact AI will have on our society, we have analyzed AI’s agency by testing different scenarios of AI’s ability to perform human-specific tasks such as having own political opinions, religious beliefs and other situations which go beyond the simple execution of tasks. The results show that consumers perceive a low probability that AI has such thoughts. However, AI with stronger anthropomorphic characteristics is more likely to have agency to perform these types of tasks. According to our study, a human-like appearance enhances the perception that AI can have political views, religious beliefs and an own identity. Moreover, people consider that AI with anthropomorphic characteristics should be responsible for their mistakes and punished if they do something wrong. In spite of this, AI, independent of its anthropomorphic characteristics, is not believed to have the ability to solve mankind problems such as reducing pollution, stabilizing the world economy or to ensure world peace. Our research also shows that men are more inclined to attribute agency to AI compared to women, which perceive a lower AI agency.

Pengembangan Layanan dan Pengenalan Akademik di Lingkungan Kampus berbasis Aplikasi Chatbot bagi Calon Mahasiswa Baru

Technology has a crucial role in information dissemination, creating a need for interactive and user-friendly communication tools. One of tools that can be used in this research is chatbot. Chatbots serve as effective intermediaries in the information exchange process, providing a simple and interactive interface. This research aims to develop a Button-Based Chatbot using the Smojo AI platform, with an easy-to-use interface for delivering college preparation information to prospective students. The development of the chatbot employs the SDLC Waterfall method. The chatbot is named "KaKa: Campus Companion" and it provides information on interest tests recommendations, effective study methods, admission pathways to public universities, and scholarship recommendations. The results of this research demonstrate the successful development of a chatbot that delivers simple and interactive information. The implementation of the SDLC Waterfall method in chatbot development ensures a well-structured and coordinated approach, focusing on each menu within the chatbot. Based on the test results, which have been carried out both white-box and black-box testing, the application can run correctly 100% and UAT testing with an accuracy level of 91%.

The way you assess matters: User interaction design of survey chatbots for mental health

The global pandemic has pushed human society into a mental health crisis, prompting the development of various chatbots to supplement the limited mental health workforce. Several organizations have employed mental health survey chatbots for public mental status assessments. These survey chatbots typically ask closed-ended questions (Closed-EQs) to assess specific psychological issues like anxiety, depression, and loneliness, followed by open-ended questions (Open-EQs) for deeper insights. While Open-EQs are naturally presented conversationally in a survey chatbot, Closed-EQs can be delivered as embedded forms or within conversations, with the length of the questionnaire varying according to the psychological assessment. This study investigates how the interaction style of Closed-EQs and the questionnaire length affect user perceptions regarding survey credibility, enjoyment, and self-awareness, as well as their responses to Open-EQs in terms of quality and self-disclosure in a survey chatbot. We conducted a 2 (interaction style: form-based vs. conversation-based) × 3 (questionnaire length: short vs. middle vs. long) between-subjects study (N=213) with a loneliness survey chatbot. The results indicate that the form-based interaction significantly enhances the perceived credibility of the assessment, thereby improving response quality and self-disclosure in subsequent Open-EQs and fostering self-awareness. We discuss our findings for the interaction design of psychological assessment in a survey chatbot for mental health.

Accelerating editorial processes in scientific journals: Leveraging AI for rapid manuscript review

IntroductionArtificial intelligence (AI) integration in business is increasingly recognized for enhancing productivity and efficiency. AI automates routine tasks and accelerates decision-making processes, optimizing operations and data analysis. This paper discusses the emergence of AI chat assistants in various sectors, focusing on models like GPT-3, Watson Assistant, Dialogflow, Alexa, Siri, Cortana, and Google Bard. These AI chat assistants exemplify the transformative impact of AI technologies in everyday life. Materials and methodsThe study explores AI's role in chatbot development, examining how AI algorithms and machine learning facilitate human-like conversations in diverse industries. The method includes a detailed analysis of AI-powered chatbots' customer service, healthcare, and e-commerce applications. The manuscript also introduces an AI-assisted peer review methodology, employing ChatGPT to analyze and comment on scientific manuscripts, aiming to enhance the efficiency and quality of the academic review process. ResultsThe AI-assisted review of selected manuscripts demonstrates the potential of ChatGPT in providing insightful feedback and suggestions. The study showcases how AI can contribute to improved manuscript quality, increased efficiency in review processes, and enhanced accessibility. The results highlight the importance of addressing ethical considerations, including privacy and bias, in deploying AI technologies. ConclusionsThis research illustrates the significant role of AI in revolutionizing business processes and academic practices. AI chat assistants and AI-assisted peer review methodologies represent pivotal advancements in their respective fields. The study underscores the need for continuous evaluation of the ethical implications of AI technologies to ensure their responsible and equitable use.

“The ChatGPT bot is causing panic now – but it’ll soon be as mundane a tool as Excel”: analysing topics, sentiment and emotions relating to ChatGPT on Twitter

AbstractChatGPT, a sophisticated chatbot system by OpenAI, gained significant attention and adoption in 2022 and 2023. By generating human-like conversations, it attracted over 100 million monthly users; however, there are concerns about the social impact of ChatGPT, including panic, misinformation and ethics. Twitter has become a platform for expressing views on ChatGPT and popular NLP approaches like topic modelling, sentiment analysis and emotion detection are commonly used to study public discourses on Twitter. While these approaches have limitations, an analytical process of existing best practices captures the evolving nature of these views. Previous studies have examined early reactions and topics associated with ChatGPT on Twitter but have not fully explored the combination of topics, sentiment and emotions, nor have they explicitly followed existing best practices. This study provides an overview of the views expressed on Twitter about ChatGPT by analysing 88,058 tweets from November 2022 to March 2023 to see if panic and concern were replicated in Twitter discourses. The topics covered human-like text generation, chatbot development, writing assistance, data training, efficiency, impact on business and cryptocurrency. Overall, the sentiment was predominantly positive, indicating that concerns surrounding ChatGPT were not widely replicated. However, sentiment fluctuated, with a decline observed around the launch of ChatGPT Plus. The discourse saw consistent patterns of trust and fear, with trust maintaining a steady presence until a decline potentially influenced by concerns about biases and misinformation. We discuss how our findings build upon existing research regarding ChatGPT by providing trajectories of topics, sentiment and emotions.

Rubric Development and Validation for Assessing Tasks' Solving via AI Chatbots

This research aimed to develop and validate a rubric to assess Artificial Intelligence (AI) chatbots' effectiveness in accomplishing tasks, particularly within educational contexts. Given the rapidly growing integration of AI in various sectors, including education, a systematic and robust tool for evaluating AI chatbot performance is essential. This investigation involved a rigorous process including expert involvement to ensure content validity, as well as the application of statistical tests for assessing internal consistency and reliability. Factor analysis also revealed two significant domains, "Quality of Content" and "Quality of Expression", which further enhanced the construct validity of the evaluation scale. The results from this investigation robustly affirm the reliability and validity of the developed rubric, thus marking a significant advancement in the sphere of AI chatbot performance evaluation within educational contexts. Nonetheless, the study simultaneously emphasizes the requirement for additional validation research, specifically those entailing a variety of tasks and diverse AI chatbots, to further corroborate these findings. The ramifications of this research are profound, offering both researchers and practitioners engaged in chatbot development and evaluation a comprehensive and validated framework for the assessment of chatbot performance.

Assessing the Suitability of Artificial Intelligence-Based Chatbots as Counseling Agents for Patients with Brain Tumor: A Comprehensive Survey Analysis

The internet, particularly social media, has become a popular resource for learning about health and investigating one's own health conditions. The development of AI chatbots has been fueled by the increasing availability of digital health data and advances in natural language processing techniques. While these chatbots are more accessible than before, they sometimes fail to provide accurate information. We used representative chatbots currently available (ChatGPT-3.5, Bing Chat, and Google Bard) to answer questions commonly asked by brain tumor patients. The simulated situations with questions were made and selected by the brain tumor committee. These questions are commonly asked by brain tumor patients. The goal of the study was introduced to each chatbot, the situation was explained, and questions were asked. All responses were collected without modification. The answers were shown to the committee members, and they were asked to judge the responses while blinded to the type of chatbot. There was no significant difference in accuracy and communication ability among the three groups (p=0.253, 0.090, respectively). For empathy, Bing Chat and Google Bard were superior to ChatGPT (p=0.004, 0.002, respectively). The purpose of this study was not to assess or verify the relative superiority of each chatbot. Instead, the aim was to identify the shortcomings and changes needed if AI chatbots are to be used for patient medical purposes. AI-based chatbots are a convenient way for patients and the general public to access medical information. Under such circumstances, medical professionals must ensure that the information provided to chatbot users is accurate and safe.

Development and Pilot-Testing of an Optimized Conversational Agent or "Chatbot" for Peruvian Adolescents Living With HIV to Facilitate Mental Health Screening, Education, Self-Help, and Linkage to Care: Protocol for a Mixed Methods, Community-Engaged Study.

Adolescents living with HIV are disproportionally affected by depression, which worsens antiretroviral therapy adherence, increases viral load, and doubles the risk of mortality. Because most adolescents living with HIV live in low- and middle-income countries, few receive depression treatment due to a lack of mental health services and specialists in low-resource settings. Chatbot technology, used increasingly in health service delivery, is a promising approach for delivering low-intensity depression care to adolescents living with HIV in resource-constrained settings. The goal of this study is to develop and pilot-test for the feasibility and acceptability of a prototype, optimized conversational agent (chatbot) to provide mental health education, self-help skills, and care linkage for adolescents living with HIV. Chatbot development comprises 3 phases conducted over 2 years. In the first phase (year 1), formative research will be conducted to understand the views, opinions, and preferences of up to 48 youths aged 10-19 years (6 focus groups of up to 8 adolescents living with HIV per group), their caregivers (5 in-depth interviews), and HIV program personnel (5 in-depth interviews) regarding depression among adolescents living with HIV. We will also investigate the perceived acceptability of a mental health chatbot, including barriers and facilitators to accessing and using a chatbot for depression care by adolescents living with HIV. In the second phase (year 1), we will iteratively program a chatbot using the SmartBot360 software with successive versions (0.1, 0.2, and 0.3), meeting regularly with a Youth Advisory Board comprised of adolescents living with HIV who will guide and inform the chatbot development and content to arrive at a prototype version (version 1.0) for pilot-testing. In the third phase (year 2), we will pilot-test the prototype chatbot among 50 adolescents living with HIV naïve to its development. Participants will interact with the chatbot for up to 2 weeks, and data will be collected on the acceptability of the chatbot-delivered depression education and self-help strategies, depression knowledge changes, and intention to seek care linkage. The study was awarded in April 2022, received institutional review board approval in November 2022, received funding in December 2022, and commenced recruitment in March 2023. By the completion of study phases 1 and 2, we expect our chatbot to incorporate key needs and preferences gathered from focus groups and interviews to develop the chatbot. By the completion of study phase 3, we will have assessed the feasibility and acceptability of the prototype chatbot. Study phase 3 began in April 2024. Final results are expected by January 2025 and published thereafter. The study will produce a prototype mental health chatbot developed with and for adolescents living with HIV that will be ready for efficacy testing in a subsequent, larger study. DERR1-10.2196/55559.