Ethical Principles of Integrating ChatGPT Into IoT–Based Software Wearables: A Fuzzy-TOPSIS Ranking and Analysis Approach

Abstract

The rapid development of the internet of things (IoT) prompts organizations and developers to seek innovative approaches for future IoT device development and research. Leveraging advanced artificial intelligence (AI) models such as ChatGPT holds promise in reshaping the conceptualization, development, and commercialization of IoT devices. Through real-world data utilization, AI enhances the effectiveness, adaptability, and intelligence of IoT devices and wearables, expediting their production process from ideation to deployment and customer assistance. However, integrating ChatGPT into IoT–based devices and wearables poses ethical concerns including data ownership, security, privacy, accessibility, bias, accountability, cost, design, quality, storage, model training, explainability, consistency, fairness, safety, transparency, trust, and generalizability. Addressing these ethical principles necessitates a comprehensive review of the literature to identify and classify relevant principles. The author identified 14 ethical principles from the literature using a systematic literature review (SLR) with a criteria of frequency ≥ 50% based on similarities. Four categories emerge based on the identified ethical principles, culminating in the application of Fuzzy-TOPSIS for analyzing, categorizing, ranking, and prioritizing these ethical principles. From the Fuzzy-TOPSIS technique results, the principle of data security and privacy is the highly ranked ethical principle for IoT–based software wearable devices with the ranking value of “0.925” as a consistency coefficient index. This method, well-established in computer science, effectively navigates fuzzy and uncertain decision-making scenarios. The pioneer outcomes of this study provide a taxonomy-based valuable insight for software manufacturers, facilitating the analysis, ranking, categorization, and prioritization of ethical principles amid the integration of ChatGPT in IoT–based devices and wearables’ research and development.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1. Introduction

In the contemporary technological landscape, wearable devices powered by internet of things (IoT) software have infiltrated various facets of our daily lives. These encompass a diverse array of products including augmented reality glasses, fitness trackers, smart clothing, hearables, patches, implantables, and smartwatches [1]. Offering functionalities spanning communication, entertainment, navigation, and health monitoring, these gadgets cater to the multifaceted needs of end-users. With the continuous evolution of wearable hardware capabilities, there is a growing demand for sophisticated software solutions aimed at enhancing user experiences and delivering heightened utility [2–6].

Similarly, conversational interfaces have also gained immense popularity due to their intuitive and natural conversational flow. Recently, the large language model (LLM) a subset of deep learning (DL) models, has significantly advanced natural language processing (NLP) and found widespread applications across various domains, generating coherent and relevant responses. ChatGPT, a cutting-edge LLM developed by OpenAI, has demonstrated remarkable proficiency in understanding and generating human-like text, including entire papers, paragraphs, and sentences. Integrating ChatGPT into IoT–based devices and wearables holds promise for enabling continuous communication, enhancing user experience, and facilitating customization, thus opening up new avenues for innovation and interaction [7, 8].

Using a substantial amount of textual data, the LLM–like ChatGPT models were trained which is based on a transformer design and was further referred to as GPT, i.e., generative pretrained transformer. For providing likeable and pertinent conversations, the ChatGPT models have already been enhanced by employing chat-based data in an argument. Moreover, for an extensive variety of applications, this model, i.e., ChatGPT is suitable because of its versatility and human-like prose, such as artificial intelligence (AI) assistance, creation of autonomous contents, user supports, and other chatbots. The OpenAI experts are regularly updated and improved so as to maintain the model at the forefront of NLP research and development by using the advanced data and training approaches [9–11].

LLMs are a type of DL model that has had a considerable influence on NLP in recent years, with applications in a variety of academic fields. ChatGPT is the most commonly used chatbot nowadays facilitated by the latest version GPT-4 and designed by OpenAI, which concentrates on conversations as a result. ChatGPT is trained with massive texts and codes, which include an extensive selection of individual requests and answers that are effectively essential to such concerns. ChatGPT is now relevant to chatbots and virtual assistants owing to these training data [12–14].

LLM also play a significant role in enhancing context awareness within IoT–based devices and wearable technology. They possess the capability to aid in the analysis of data generated by such devices, empowering users to make informed decisions grounded in factual evidence. Moreover, LLMs are proficient in evaluating and analyzing sensor data, thereby facilitating the delivery of timely and relevant services or information. By assessing context-based data encompassing factors such as customer location, actions, and environmental variables, LLMs enable a more nuanced understanding of the surrounding context, ultimately enhancing the effectiveness and utility of IoT applications and wearables [15, 16].

NLP–capable wearables and IoT devices are simplifying information transmission between humans and machinery in a manner that seems more straightforward and understandable for each party. Just imagine being capable of talking to your smart home appliances to operate them without using a physical interface or a smartphone app. Since NLP has been applied to these devices, they can comprehend and react to human negotiations. This also applies to other IoT wearables and devices, such as vehicles, factories, smart machineries, and robots [12, 17].

By the integration of NLP, i.e., ChatGPT into IoT–based devices and wearables’ development and research, an increasingly natural and intimate relationship between users and their IoT–based wearables is made possible. IoT–based devices and wearables with NLP integration are, to put it simple, complimentary technologies which offer easy and more natural conversations among machines and consumers, hence opening new areas for sophisticated interactions of AI. NLP is also the basis of advanced exchange, i.e, response inquiries and interaction of natural languages into IoT–based devices and wearables’ development and research [17, 18].

ChatGPT integration into IoT–based devices and wearables might additionally enhance standardization and personalization of distant rehabs, allowing adaptive schedule modifications, real-time tracking, automated plan generation, and other inputs during guidelines and training. These advancements of ChatGPT into IoT–based devices and wearables’ development and research might boost the accessibility and efficacy of rehab services for a huge amount of users which have otherwise restricted access to them [14, 18].

ChatGPT may afford contextual user understandings, preservation analysis, customized recommendations, and interaction of natural language along with other advantages. ChatGPT may also improve the efficiency and access control of IoT–based devices and wearables’ development and research. They are also important instruments for enhancing the access control and intelligence of IoT–based devices and wearables due to their capability of analysis, versatility, and natural language ability [19, 20].

Researchers may utilize ChatGPT for replicating discussions and interactions between people from various cultural backgrounds, as well as to investigate the influence of linguistic variables such as accent, jargon, and terminology on language usage. Furthermore, ChatGPT can be used to produce synthetic knowledge for training additional models, and its effectiveness can be measured against data produced by humans [13, 19].

In this article, the author developed two research questions to help in addressing, analyzing, evaluating, ranking, and prioritizing these important concerns. These questions are as follows:

• RQ.1: What are the ethical principles of integrating ChatGPT into IoT–based devices and wearables’ development and research?

• RQ.2: How to rank, prioritize, analyze, and categorize these identified critical IoT software devices and wearable principles?

The author used the fuzzy “technique for order preference by similarity to ideal solution (TOPSIS)” technique for ChatGPT principles analysis, prioritization, ranking, and evaluation. It is an extension of the conventional TOPSIS approach utilizing fuzzy set theory, which deals with vague and uncertain data. A decision matrix is developed in a conventional TOPSIS manner to assess the performance of different choices across multiple criteria. The ideal and anti-ideal solutions are resolute, and the alternatives, i.e,. principles of ChatGPT, are evaluated and ranked according to their nearness to the ideal solutions and distance from the anti-ideal solutions [21–23].

The structure of this research article is outlined as follows.

In Section 2, a comprehensive systematic literature review (SLR)–based literature review is presented, delving into the ethical considerations surrounding the integration of ChatGPT into IoT–based devices and wearables. Subsection 2.1 describes the literature findings, while Subsection 2.2 highlights the problem statement and Subsection 2.3 emphasized the significance of research.

Section 3 provides an overview of the methodology employed in integrating ChatGPT with IoT–based devices and wearables. Within this section, Subsection 3.1 offers a comprehensive exploration of the fuzzy set theory, while Subsection 3.2 elucidates the Fuzzy-TOPSIS technique and its application.

Section 4 delves into the findings of the SLR in Subsection 4.1 (RQ1) and presents the results derived from the application of Fuzzy-TOPSIS in Subsection 4.2 (RQ2), which prioritizes, evaluates, ranks, and categorizes these ethical principles, further proposing a taxonomy-based analysis of the SLR–based literature and Fuzzy-TOPSIS results.

Finally, in Section 5, the article concludes by discussing future directions in this field. Subsection 5.1 is about acknowledgment of research support, and Section 6 emphasized the detailed bibliography (references).

2. Literature Review

Although, considering the myriad benefits, IoT–based devices and wearable still encounter ethical challenges when integrating advanced AI and LLM–like ChatGPT. These ethical concerns have been thoroughly explored in the literature by numerous researchers. For instance, Nahavandi, et al. [24] reviewed the AI methods for applications of wearable devices specifically in the field of sports, medical, and industrial domains. Furthermore, they highlighted the concerns of data collection, transmission, security, interoperability, storage, and quality. However, they did not actually detail the strategies to overcome these challenges. Similarly, Seneviratne et al. [25] surveyed IoT–based wearables and addressed security, computing power, and energy challenges, discussing computation offloading and in-device machine learning (ML) techniques. Yet, concrete solutions were not provided. Gill and Kaur [26] delved into the background and development of ChatGPT and further examined the future direction, foundation, applications, and challenges of ChatGPT. However, they did not elucidate how to deal with these challenges. Lund et al. [27] discusses recent developments in ML, NLP, and AI. Furthermore, it provides a comprehensive overview of the ethical considerations of using ChatGPT and similar technologies in research and development. However, they did not explain and answer about the impact of ethical consideration of using ChatGPT in research and development.

The authors Seng et al. [28] provided a comprehensive review and a detailed survey of the classification of research prototypes and smart wearables using AI and ML technologies. The authors also discussed about the conventional ML techniques and DL approaches. They found significant technical issues in communication and networking aspects for AI smart IoT–based wearables, including security, battery consumption, computational complexity, data storage, training and inference, communication, and routing overheads. The authors only identified future direction, but did not propose any solution to these challenges.

The authors Pasricha and Wolf [29] offered comprehensive perspectives on the design of semiconductor chips, growing use of AI technologies, and IoT wearables and applications and their ethical challenges including security, safety, e-waste, transparency, bias, trust while using AI in IoT systems. The authors also addressed these ethical issues from different points of view such as programing ethics, effective integration in security mechanisms, improved AI algorithm design, data transparency, and regulation with public policies. which are essential for establishing a framework for comprehending ethical manufacturing of computing systems.

The authors Kim et al. [30] presented AI–driven virtual emotion system in 5G networks, i.e., 5G-I-VEmoSYS, consisting of AI–VEmoFLOW, AI–VEmoBAR, and AI–VEmoMAP and carefully addressed the critical challenges for possible AI–based attacks.

The author Marengo [31] discussed the detailed SLR and guidelines of recent advancements and challenges in AI–driven IoT devices and wearables, concentrating on data privacy and analysis of AI–driven data. Furthermore, they analyzed the integration of AI into IoT devices and wearables in industrial, urban resource management, and healthcare, which enhance operational effectiveness, service personalization, and information-driven conclusions.

The authors Shi et al. [32] elaborates data-driven AI and knowledge-based AI and highlights IoT architecture’s three layers, namely, application, network, and sensing layers for convergence of AI in IoT. The authors discussed about the design, security, and privacy of IoT when integrating AI that will drive novel tendencies of manufactural revolutions of IoT devices.

The author Sepasgozar et al. [33] reviewed the integration of IoT and AI in the development of smart home appliances and identified gaps in terms of aged care system growth and restricted energy efficient integrated systems. Furthermore, they emphasized the challenges of power consumption and costs of IoT devices and sensors.

The authors’ Binder and Mezhuyev [34] underlined significant advantages, e.g., offering initial drafts and process speediness when creating IoT systems and device specifications using LLM specifically ChatGPT. In addition, they highlighted the challenges allied with specific domain knowledge, ambiguity, data quality, security, and workflow integrations for ensuring the significance and quality of generated IoT system specifications.

The authors Neha et al. [35] reviewed ChatGPT applications in medical healthcare comprehensively via patient symptom analysis and history of disease and addressed critical issues such as data privacy, trust, and accuracy while integrating ChatGPT in healthcare.

The authors Chakraborty et al. [36] deliberated the novel data-driven paradigm modification in the field of healthcare and medicines from conventional ML to novel DL approaches. The author also emphasized the key limitations, e.g., data quality and availability, data bias and trust, and data security when integrating DL–enabled ChatGPT technology in IoT–based sensors and devices while dealing with medicine and healthcare of patients.

The authors Oliveira et al. [37] proposed a novel framework and innovative design with improved efficiency for the convergence of AI technologies in IoIT, focusing on security concerns, resource limitation, AI and ML integration, and scalability.

The authors Sun et al. [38] surveyed and analyzed comprehensively the recent state of security in IoT devices and wearables, Moreover, they proposed a systematic security protection framework and discoursed the development and integration of IoT with AI, ML, big data, and cloud computing. The authors also discussed the challenges of data security, unauthorized access, limited resources, and generalizability issues while integrating IoT with these technologies.

The authors Zhang et al. [39] introduces a novel AI and IoT–based data-driven framework “MetaCity” and discusses the data integration, privacy, and technical dependencies issues when merging IoT technologies with AI data-driven techniques.

The authors Nguyen et al. [40] proposed the integration of edge computing, AI, and blockchain with IoT systems and also analyzed the security and privacy challenges which IoT devices faced during integration.

From the literature, it is evident that researchers have discussed various principles while integrating AI and LLM especially ChatGPT into IoT–based software wearables devices. However, none have employed the Fuzzy-TOPSIS technique to rank, prioritize, analyze, and categorize these principles, Furthermore, to know the level and potential of these principles of ChatGPT integration into IoT–based devices and wearables, a comparative overview of the most recent and relevant state-of-the-art (SOTA) work is shown in Table 1, which provides the overview, methodology, key findings, and limitations in the existing research area. This pioneering methodology could provide valuable insights and guidance for addressing the ethical concerns effectively.

Table 1

A comparative analysis of state-of-the-art (SOTA) for ChatGPT and IoT integration.

Paper	Author	Year	Methodology	Key findings	Key limitations
[41]	Yao et al.	2024	Systematic survey of LLM vulnerabilities and attacks in security, privacy, integrity, reliability, and confidentiality	Identified threats, i.e., adversarial attacks, data poisoning, and prompt injections	Limited empirical validation of modification techniques
[42]	Zong et al.	2025	Integration of LLM and IoT by combining machine learning, NLP, and IoT systems to develop intelligent applications	Determines enhanced automation and efficiency in IoT devices using LLMs	Security and privacy, lack of generalizability and scalability issues, computationally expensive, and latency issue
[43]	Wang et al.	2024	An enhanced BERT-of- Theseus “BT-TPF” model for novel intrusion detection in light-weight IoT devices for accuracy, efficiency, and protections	Enhanced IoT device performance via intrusion detection with a light-weight design appropriate for memory-constrained	Generalizability and scalability issues for all IoT scenarios
[17]	Kumar and Kumar	2023	NLP–based anomaly detection techniques for IoT security with a review of several IoT security practices	NLP–based threat modeling and anomaly detections. Further identification of IoT security challenges	-No case study-Lack of experimental work-A review-based and conceptual study-Limited real-world applicability
[44]	Chen et al.	2024	A combined framework for federated learning and LLM	Preservative privacy and decreasing communication overhead. Fine-tune LLM federated learning settings	Limited scalability, high computational costs, and insufficient heterogeneity management
[29]	Pasricha and Wolf	2023	IoT and AI ethical design framework	Environmental and societal challenges of designing computing technologies	Lacks case studies, empirical data, and implementation details on IoT systems and their issues
[26]	Gill and Kaur	2023	Analysis of ChatGPT capabilities, challenges, and vision IoT via review and survey study	ChatGPT integration with IoT and cyber-physical systems. Highlights the security and automation of IoT devices	Lack of experimental work, only conceptual analysis and theoretical work
[34]	Binder and Mezhuyev	2024	ChatGPT framework for the specification of IoT devices and wearables	ChatGPT framework assists automatic generations of IoT design and documentation via NLP	-Extra comprehensive testing required-Limited validations of IoT devices
[45]	Ali et al.	2025	Evaluation of detailed analysis of AI–driven machine learning and deep learning security techniques in smart cities	Emphasized IoT and AI integration and proposed a multilayered framework for security in smart areas	-Not applicable to simple IoT devices and wearables-Specific area in-depth analysis limitation
[46]	Kassab and DeFranco	2023	Conceptual framework and comprehensive literature review	Improve user automation and interactions via ChatGPT integration into IoT–based devices	Theoretical framework with limited empirical validation of data

2.1. Literature Findings

In this subsection, the authors present the 14 identified principles from gray literature as well as peer-reviewed articles. The details of these derived findings are discussed in [26, 28, 34, 47–56].

2.1.1. Data Security and Privacy

ChatGPT and wearable devices have access to so much of their information that customer’s security and privacy might be jeopardized. The secure retention and ethical use of customer data involve the development of relevant regulations and legislation. Furthermore, intolerance and misinformation are the only examples of the potentially harmful content produced by ChatGPT. So, it is critical to create controls to prevent the development of this kind of content.

2.1.2. Data Storage

ChatGPT is basically a LLM, and it might be difficult to fit it into memory-constrained IoT devices and wearables. Furthermore, maintaining a LLM on IoT wearable devices could put a pressure on the existing storage capacity of the wearables.

2.1.3. Data Bias

The amount and diversity of data used to train ChatGPT can have an impact on its performance. The use of biased data for training to develop predictions may have negative consequences in domains including healthcare, information technology, business, and many others.

2.1.4. Design and Comfort

Wearables sometimes have tiny screens, which limits the amount of area accessible for showing text. In addition, different interaction modalities, such as voice, touch, and actions may be supported. As a result of the delay in the connection between the device and the server where the model is stored, achieving real-time interaction with ChatGPT may be difficult, and delayed replies can annoy customers while also impeding the natural progression of speech, thus lowering the overall comfort and usefulness of the interconnected system.

2.1.5. Transparent Communication

Overcoming transparent communication entails enhancing the integration to deliver concise and clear messaging, retaining confidence among customers, and optimizing the user experience by avoiding uncertainty in the interpretation of natural language contributions under the surroundings of IoT–based devices and wearables.

2.1.6. Battery Consumption

ChatGPT models require a lot of computing resources due to the quantity of information they carry and might have a negative impact on the ecosystem. The energy consumption and efficacy of ChatGPT have a lot of space for enhancement.

2.1.7. Explainability

The ChatGPT model is complex and difficult to grasp. This may make it difficult to derive the model’s decision-making procedure and identify any errors. Since these IoT gadgets are frequently used in important or intimate situations, customers must understand the reasoning behind the AI–generated solutions.

2.1.8. Quality Issue

ChatGPT can produce excellent language; however, it may also deliver poor-quality or inappropriate responses. Persistent monitoring, understanding, and development are required to maintain ChatGPT’s function of delivering quality material.

2.1.9. Compatibility

Although compatibility problems can occur in a variety of settings, when implementing advanced AI models such as ChatGPT, they frequently pertain to difficulties integrating the model with disparate platforms, systems, or technology. Compatibility concerns must be resolved with a complete strategy that includes thorough documentation, reliable testing methods, and a dedication to continuing updates and maintenance.

2.1.10. User’s Understanding

AI–based wearables must have an interface for users that is simple, adaptable, and intuitive because it is intended to be utilized in a variety of settings. However, due to the tiny size of the device, the constrained screen area, and the requirement for a basic design that does not overwhelm the user, creating a user interface for a wearable device is difficult. As an outcome, certain wearable technology may have a cumbersome or challenging user interface that irritates consumers.

2.1.11. Trust

ChatGPT has shown great aptitude in making humanistic written form, yet it has also been demonstrated to produce errors or offer incorrect info on a few occasions. Maintaining trust in discoveries in science is dependent on the quality and reliability of AI–generated data.

2.1.12. Generalizability

ChatGPT is usually trained on very large datasets, and it is often erroneous and unable to generalize to new data. The development of innovative training methods is crucial to improving ChatGPT generalization.

2.1.13. Cost Issue

AI–based wearable gadgets can be pricey, which limits adoption among consumers. Wearable device prices might rise due to high production costs and the requirement for sophisticated components such as sensors and central processing units (CPUs). This can be a major obstacle to access for people who cannot afford the newest and powerful wearable technologies.

2.1.14. Data Hallucination

Even if it is improbable, an AI model may forecast that something will happen. For instance, even when there is no rain in the forecast, an AI weather prediction model may indicate that it will rain tomorrow. A number of variables, such as hallucination in the training set, overfitting, or restrictions in the model design, may contribute to this problem.

In general, these principles outlined in Figure 1 and elaborated upon previously pose significant hurdles for developers aiming to craft IoT devices and wearables that meet consumer demands and expectations. While extensively discussed in literature by numerous researchers, these concerns have not been systematically analyzed and prioritized using methodologies such as the Fuzzy-TOPSIS technique to identify the most critical principle. However, by addressing these paramount issues, developers can potentially develop IoT devices and wearables offering substantial benefits to users, thereby fostering the growth and advancement of the wearable technology sector through the integration of ChatGPT.

[figure(s) omitted; refer to PDF]

2.2. Problem Statement

From the detailed SLR study, the author found that ChatGPT improves effectiveness in acquiring information and interaction in the development and research of IoT–based software wearable devices. Its ability to generate authentic and contextually relevant language makes it an appealing research tool. However, there are certain ethical principles encompassing data ownership, security, privacy, accessibility, bias, accountability, cost, design, quality, storage, model training, explainability, consistency, fairness, safety, transparency, trust, and generalizability while integrating ChatGPT in IoT–based software wearables devices. Many researchers deliberated these principles in different articles, but yet, none of them have employed the Fuzzy-TOPSIS method to categorize, rank, evaluate, and prioritize these principles to know the level and potential of these principles and assist researchers and developers while integrating ChatGPT into IoT–based software wearable devices. This proposed pioneer methodology could provide valuable insights and guidance for addressing these ethical principles effectively.

2.3. Significance of Research

The author identified from the SLR study that ChatGPT integration in IoT–based software wearable devices carries great significance and potential and can enhance user’s interface, conversational control, security, natural language understanding, data analysis, synthesis, and much more based on natural language interactions. However, still ChatGPT has some principles that will affect its integration in IoT–based software wearable devices properly, including data ownership, security, privacy, accessibility, bias, accountability, cost, design, quality, storage, model training, explainability, consistency, fairness, safety, transparency, trust, and generalizability. Our proposed research aims to fil the given gap and will assist software developers, researchers, and users by elaborating the main fundamentals, i.e., critical challenges, ethical principles of using ChatGPT in IoT, level-based categorization, prioritization, ranking, and analysis of these identified principles. Our proposed research will further establish the impact of ChatGPT into IoT–based wearable devices’ development and research, by developing a taxonomy through a detailed SLR study and Fuzzy-TOPSIS global ranking, analysis, and prioritization technique. By contributing a body of knowledge and serving as solutions for integrating ChatGPT consideration in the development and research of IoT–based software wearables and devices, we believe that the findings of this study will be beneficial to the community of mainstream academic development and research.

3. Methodology

To obtain our desired goal of integrating ChatGPT into IoT–based device and wearable principles and opportunities identification, analysis, prioritization, ranking, and evaluation, the author designed the detailed procedure of the “fuzzy analytic hierarchy process (AHP)” including the overall six phases shown in Figure 2.

[figure(s) omitted; refer to PDF]

3.1. Fuzzy Set Theory

For eliminating fuzziness or uncertainty that could arise in the mind of humans while dealing with a particular concern in “multicriteria decision-making (MCDM),” the fuzzy theory set was developed [57], and introduced by Prakash and Barua, which is extremely important in illustrating the vague and uncertain data in MCDM [58, 59]. The most common mathematical methods for various decision-making criteria and vagueness problems based on humans are fuzzy logic and fuzzy sets.

To understand triangular fuzzy numbers (TFNs), each decision-maker in the decision-making process conducts a pairwise assessment and assigns comparison ratings to various criteria for evaluation of relative significance. Following that, the priority weights for each criterion are determined using the fuzzy AHP method. The experts applied the linguistic scale of importance to compare the criteria, since human opinion is classified as verbal expressions, which are further stated as verbal words or linguistic terms. The fuzzy AHP technique is made up of a linguistics scale variable with values expressed as words rather than crisp numeric values. The TFN is represented and defined by three values ($l$, $m$, and $u$), which express the values as the least (lower value), the most promising or most likely (mean value), and the highest (upper value), defining the fuzzy occurrences and uncertainty related with a decision demonstrated in Figure 3, where $\mathrm{\mu }\mathrm{F}x$ is the relationship and association function that graphs the object values from 0 to 1 within the fuzzy set [60, 61].

[figure(s) omitted; refer to PDF]

3.2. Fuzzy-TOPSIS Approach

The TOPSIS technique, presented by Hwang in 1981, is the most well-established approach that deals with fuzzy logic for addressing fuzzy and uncertain problems in MCDM. The principles of TOPSIS were established on the basis of alternations selected by each expert’s decision-making, which are the ones that are the farthest to that of fuzzy negative ideal solutions (FNISs) and the nearest to that of fuzzy positive ideal solutions (FPISs). Positively, the solution can increase the benefits’ criteria while lowering the cost criteria, while negatively, it can increase the benefits criteria while decreasing the cost criteria. However, there are several limitations to employing the TOPSIS conventional approach for MCDM, such as computing uncertain data in a fuzzy platform. Ambiguity and fuzziness are important factors in decision-making challenges.

Consequently, Fuzzy-TOPSIS is an effective strategy for resolving such sorts of ambiguity in data that arise from multiple expert decision-makers. In this research paper, the author used the same technique to identify, evaluate, rank, and prioritize the principles of integrating ChatGPT in IoT–based software wearables and devices, by considering the opinion and grading of each expert for the identified 14 principles and their alternatives in the decision-making process. The final results of the most critical principle, i.e., “best alternative” among these 14 principles will be the one closest to the FPIS and farthest from the FNIS. Several researchers have used this approach in their research articles [61–68].

For the identification, prioritization, ranking, analysis, and evaluation of principles associated with software wearable devices, the author offers well-planned Fuzzy-TOPSIS protocols. Following are the comprehensive steps for using the Fuzzy-TOPSIS technique.

1. Initially, each criterion that was established and their alternatives were rated by decision-makers using the triangular fuzzy scales demonstrated in Table 2.

2. In this step, a fuzzy decision matrix is constructed incorporating fuzzy numbers to measure the performance of each expert in an uncertain decision-making process. Therefore, let us assume that “$n$” is the decision-making criteria against $″m$” possible alternatives. As a result, the fuzzy rating is calculated for all criteria, which are denoted by $″D$” and described as $d_{ij}$ = ($a_{ij}$, $b_{ij}$, $c_{ij}$). The resultant fuzzy decision matrix $″D$” is shown in equation (1) for the “$n$” criterion and $″m$” alternatives, where $m$ = ($A_1$, $A_2$, $A_3$, …, $A_m$), and $n$ = ($C_1$, $C_2$, $C_3$,…, $C_n$). Furthermore, to represent the decision matrix, each number as a fuzzy number is shown in equation (2).$$\begin{array}{}\text{(1)}& \tilde{D}=\begin{array}{cccc}{d}_{11}& d_{12}& \cdots & d_{1n}\\ d_{21}& d_{22}& \cdots & d_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ d_{m1}& d_{m2}& \cdots & d_{mn}\end{array},\text{(2)}& \tilde{D}=\begin{array}{cccc}{a}_{11},b_{11},c_{11}& a_{12},b_{12},c_{12}& \cdots & a_{1n},b_{1n},c_{1n}\\ a_{21},b_{21},c_{21}& a_{22},b_{22},c_{22}& \cdots & a_{2n},b_{2n},c_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ a_{m1},b_{m1},c_{m1}& a_{m2},b_{m2},c_{m2}& \cdots & a_{mn},b_{mn},c_{mn}\end{array}.\end{array}$$

3. Next, we calculate the aggregate fuzzy rating for each criterion and alternative that will further result in the combined decision matrix. Now, for $″K$” decision-making experts, the fuzzy grading significance of weight is $X_{ij}$ = ($l_{ijk}$, $m_{ijk}$, $u_{ijk}$), where $i$ = (1, 2, 3, 4…$m$), $j$ = (1, 2, 3, 4…$n$), and the resulting fuzzy cumulative rates $X$ (ab) of the findings with respect to each criterion, together with all possible alternatives, are shown as follows.

  $X_{ij}$ = ($l_{ij}$, $m_{ij}$, $u_{ij}$), and aggregate fuzzy weights of each criteria are $W_{ij}$ = ($w_{j1}$, $w_{j2}$, $w_{j3}$), with ($w_{j1}$) as an aggregate fuzzy weight with respect to the $j$th criteria.$$\begin{array}{c}\text{(3)}& {\mathrm{a}}_{\text{ij}}={\mathrm{K}}^{\min }{\mathrm{l}}_{\text{ijk}},\mathrm{b}=\frac{1}{\mathrm{K}}{\displaystyle \sum }_{\mathrm{k}=1}^{\mathrm{K}}{\mathrm{m}}_{\text{ijK}},\\ \mathrm{c}={\mathrm{K}}^{\max }{\mathrm{u}}_{\text{ijK}},\\ \text{(4)}& {\mathrm{w}}_{\mathrm{j}1}={\mathrm{K}}^{\min }{\mathrm{w}}_{\text{jk}1},\mathrm{b}=\frac{1}{\mathrm{K}}{\displaystyle \sum }_{\mathrm{k}=1}^{\mathrm{K}}{\mathrm{m}}_{\text{jK}2},\\ \mathrm{c}={\mathrm{K}}^{\max }{\mathrm{u}}_{\text{jK}3}.\end{array}$$

  Where TFN components $l_{ijk}$ = lower value, $m_{ijk}$ = most likely value, and $u_{ijk}$ = upper, i.e., maximum value for the $j$th criteria with each $i$th alternative provided by the $k$th decision-maker.

4. This step consists of creating a matrix with normalization using the weights of criteria and the decision matrix jointly, where $P$ denotes the resulting normalized fuzzy decision matrix and $P_{ij}$ represents the performance rating of the $j$th criteria with the $i$th alternative for $i$ = (1, 2, 3, 4,…, $m$), and $j$ = (1, 2, 3, 4,…, $n$).$$\begin{array}{}\text{(5)}& \tilde{P}={{\tilde{P}}_{ij}}_{m\times n},\text{(6)}& {\tilde{\mathrm{p}}}_{\text{ij}}=\frac{{\mathrm{l}}_{\mathrm{j}}^{-}}{{\mathrm{u}}_{\text{ij}}},\frac{{\mathrm{m}}_{\mathrm{j}}^{-}}{{\mathrm{m}}_{\text{ij}}},\frac{u_{\mathrm{j}}^{-}}{{\mathrm{l}}_{\text{ij}}}.\end{array}$$

  Here, ${\mathrm{l}}_{\mathrm{j}}^{-}=\min {\mathrm{l}}_{\text{ij}}$ (called worst criteria).$$\begin{array}{}\text{(7)}& {\tilde{\mathrm{p}}}_{\text{ij}}=\frac{{\mathrm{l}}_{\text{ij}}}{{\text{uj}}^{\mathrm{\ast }}},\frac{{\mathrm{m}}_{\text{ij}}}{{\text{uj}}^{\mathrm{\ast }}},\frac{{\mathrm{u}}_{\text{ij}}}{{\text{uj}}^{\mathrm{\ast }}}.\end{array}$$

  Here, ${\text{uj}}^{\mathrm{\ast }}=\max {\mathrm{u}}_{\text{ij}}$ (called best criteria).

  In the above equations, $l_{ij}$, $m_{ij}$, ${andu}_{ij}$ are fuzzy performance ratings of the $j$th criteria for each $i$th alternative, $u$^∗$j$ is the highest upper value, while ${\mathrm{l}}_{\mathrm{j}}^{-}$ and ${\mathrm{m}}_{\mathrm{j}}^{-}$ are the lowest lower and middle values, respectively.

5. In this step, for the weighted normalized matrix $″V$” of the $j$th criteria for each $i$th alternative, we multiply fuzzy weight $″W_j$” with normalized fuzzy values “$P_{ij}$” from the fuzzy decision matrix. Here,$$\begin{array}{c}\text{(8)}& V_{ij}=P_{ij}\times W_j,V=V_{ij}n\times m,\end{array}$$

  where $i$ = (1, 2, 3, 4, … ‘$m$’) and $j$= (1, 2, 3, 4,… ‘$n$’).

6. Next, as shown in the following, we will determine the FNIS and FPIS’ solutions.$$\begin{array}{c}\text{(9)}& A^{-}=V_1^{-},V_2^{-},V_3^{-},\cdots ,V_n^{-},V_j^{-}=\min V_{i\times j},\\ \text{(10)}& A^{+}=V_1^{+},V_2^{+},V_3^{+},\cdots ,V_n^{+},V_j^{+}=\max V_{i\times j}.\end{array}$$

  Here, $i$ = (1, 2, 3, 4, …, m) and $j$ = (1, 2, 3, 4,…, $n$).

7. Using the Euclidean distance formula, the author will calculate the distances between all alternatives against each criterion from FPIS and FNIS. The alternative with the greatest remoteness from that of FNIS and the smallest distance from FPIS shall be the optimum among all the alternatives. The ${\mathrm{d}}_{\mathrm{i}}^{+}$ and ${\mathrm{d}}_{\mathrm{i}}^{-}$ distances from that of the FPIS and FNIS are determined for every alternative $i$ = (1, 2, 3, 4, …, up to $m$) as follows:$$\begin{array}{}\text{(11)}& {\tilde{\mathrm{d}}}_{\mathrm{i}}^{-}={\displaystyle \sum }_{\mathrm{j}=1}^{\mathrm{n}}{{\mathrm{v}}_{\text{ij}}-{\mathrm{v}}_{\mathrm{j}}^{-}}^{2}1/2,\text{(12)}& {\tilde{\mathrm{d}}}_{\mathrm{i}}^{+}={\displaystyle \sum }_{\mathrm{j}=1}^{\mathrm{n}}{{\mathrm{v}}_{\text{ij}}-{\mathrm{v}}_{\mathrm{j}}^{+}}^{2}1/2.\end{array}$$

  Here, $i$ = (1, 2, 3, 4, …, $m$).

8. The closeness of the coefficient index, or “CCi” value, is generated after all of the alternate FPIS and FNIS values have been calculated; the CCi values demonstrate both FPIS and FNIS concurrently by using the following formula:$$\begin{array}{}\text{(13)}& {\text{cc}}_{\mathrm{i}}=\frac{{\mathrm{d}}_{\mathrm{i}}^{-}}{{\mathrm{d}}_{\mathrm{i}}^{-}+{\mathrm{d}}_{\mathrm{i}}^{+}},\mathrm{i}=1,2,3,4,\cdots \mathrm{m},{\mathrm{c}}_{\mathrm{i}}ϵ0,1,\end{array}$$

  where ${\mathrm{d}}_{\mathrm{i}}^{-}$ represent the distance from FNIS and ${\mathrm{d}}_{\mathrm{i}}^{+}$ represents the distance from FPIS.

9. Finally, we will rate out all of the alternatives. The optimal alternative will be the one with the shortest distance from FPIS and the greatest distance from FNIS.

Table 2

Triangular fuzzy scale.

AHP scales	Linguistic scales	Fuzzy triangular scales	Fuzzy triangular reciprocal scales
1	Equally imperative (EI)	(1, 1, 1)	(1, 1, 1)
3	Weakly imperative (WI)	(2, 3, 4)	(0.25, 0.33, 0.5)
5	Fairly imperative (FI)	(4, 5, 6)	(0.16, 0.2, 0.25)
7	Strongly imperative (SI)	(6, 7, 8)	(0.125, 0.14, 0.16)
9	Absolutely more imperative (AMI)	(9, 9, 9)	(0.11, 0.11, 0.11)

4. Results

4.1. Research Question 1

For answering RQ1, a total of 14 IoT–based software devices and wearables’ principles were identified from the detailed SLR study of 77 final most relevant articles, an appendix list of these articles and their corresponding references are included in the Appendix, selected from a total of initial 434 articles using Scopus, Web of Science, Google Scholar, and Science Direct keeping the frequency ≥ 50% [23]. These final 14 identified principles are shown in Table 3. After the inclusion and exclusion criteria, the author excluded 357 papers on the basis of duplications and no relevancy. The detailed SLR analysis and phases of planning, conducting, and reporting reviews are shown in Figure 4. Furthermore, the author mostly selected the recent work of SLR from 2023 to 2025. This selected 77 final literature assessment flow of integrating ChatGPT in IoT–based devices and wearables principles is shown in Figure 5.

Table 3

Identified IoT–based wearables and device principles from SLR.

S.no	Principles	Frequency ($n$ = 77)	%
WDC-1	Data security and privacy	67	87
WDC-2	Data storage	64	83
WDC-3	Data bias	61	79
WDC-4	Design and comfort	60	78
WDC-5	Transparent communication	59	77
WDC-6	Battery consumption	58	75
WDC-7	Explainability	57	74
WDC-8	Quality issue	55	71
WDC-9	Compatibility	53	69
WDC-10	User understanding	51	66
WDC-11	Trust	50	65
WDC-12	Generalizability	48	62
WDC-13	Cost issue	44	57
WDC-14	Data hallucination	42	55

Abbreviation: WDC = wearable device challenge.

[figure(s) omitted; refer to PDF]

From the literature, the author identified a total of 14 IoT–based devices and wearables’ principles from the most related research articles. These principles are based on the frequency of ≥ 50%, with results including data security and privacy, data storage, data bias, design and comfort, transparent communication, battery consumption, explainability, quality issue, compatibility, user understanding, trust, generalizability, data hallucination, and cost issue, while integrating ChatGPT in IoT–based devices and wearables. Furthermore, we have categorized these principles into four categories on the basis of their relevancy, where each principle is categorized as “WDC” as displayed in Table 4.

Table 4

Categorization of the identified IoT–based wearable and device principles.

P.no	Category	Principles
WDC-2	Category-1	Data storage
WDC-3		Data bias
WDC-14		Data hallucination
WDC-8		Quality issue
WDC-5	Category-2	Transparent communication
WDC-1		Data security and privacy
WDC-6	Category-3	Battery consumption
WDC-12		Generalizability
WDC-9		Compatibility
WDC-7		Explainability
WDC-13		Cost issue
WDC-10	Category-4	User understanding
WDC-4		Design and comfort
WDC-11		Trust issue

4.2. Research Question 2

For answering RQ2, the Fuzzy-TOPSIS approach was applied to these identified principles to prioritize, evaluate, and rank each principle. This technique is also used by many other researchers while in uncertain decision-making situations [66, 67, 69]. The author applied the same steps one by one already discussed in Section 3 in detail.

1. For the decision-making, prioritization, and ranking of these identified software wearable device principles, we have the criteria as categories (category-1, category-2, category-3, and category-4), and the alternatives as all the 14 principles mentioned above.

2. The author computes the performance matrix using equation (1) based on various expert opinions. Each choice has a variable assigned to it since all replies are given in linguistic terms at the outset. Let us say the values of the linguistic phrases “just equal (JE)” and “absolutely more important (AMI)” are (1, 1, 1) and (9, 9, 9) correspondingly. The resulting matrix of fuzzified decision matrix for each labeled criteria and performances with respect to category-1 are shown in Tables 5 and 6.

3. Equations (3) and (4) are applied to determine the “geometric mean” and “aggregate fuzzy weights” for every criterion, and Tables 7 and 8 show the geometric mean and combined matrix for overall identified principles with the criteria following the aggregate fuzzy weights, respectively.

4. This phase involved calculating the normalized matrix from the combined decision matrix, using equations (5)–(7) given in Table 9, and the resultant normalized fuzzy weights are displayed in Table 10.

5. Next, we multiply each criterion’s weight by the weight of each alternative to create the weighted fuzzy normalized matrix. Using equation (8), the resulting normalized weighted decision-making matrix is displayed in Table 11.

6. The results are calculated using equations (9) and (10) for FPIS and FNIS, which are shown in Table 12.

7. In this stage, we calculated the distances from both FPIS and FNIS for all alternatives by applying equations (11) and (12). For example, for principle WDC-1, the values of ${\mathrm{d}}^{+}$ and ${\mathrm{d}}^{-}$ for the criterion category-3 are shown as follows:$$\begin{array}{c}\text{(14)}& {\mathrm{d}}^{+}=\text{SQRT}1/3\mathrm{\ast }{\mathrm{E}42-\mathrm{E}57}^2+{\mathrm{F}42-\mathrm{F}57}^2+{\mathrm{G}42-\mathrm{G}57}^2=0.61,{\mathrm{d}}^{-}=\text{SQRT}1/3\mathrm{\ast }{\mathrm{E}42-\mathrm{E}56}^2+{\mathrm{F}42-\mathrm{F}56}^2+{\mathrm{G}42-\mathrm{G}56}^2=0.1.\end{array}$$

  Applying a similar process and formula, we computed every alternative distance from that of FPIS and FNIS and then calculated the results of ${\mathrm{d}}^{+}$ and ${\mathrm{d}}^{-}$ by addition of all the alternatives with regard to every criterion as given in Tables 13 and 14. From Figures 6 and 7 we get the result for WDC-1 as “0.1” from FPIS and the largest distance of “1” from FNIS. The combined results of both FPIS and FNIS with ranks are shown in Figure 8, which clearly showed that the IoT–based software wearable device principle “data security and privacy (WDC-1)” is the top most principle with the highest ranking and the most critical one among the other alternatives, identical to that identified from the literature.

8. In this stage, we used equation (13) to calculate the close distance of coefficient index values (CCi) for all of the choices. Table 15 displays the final estimated CCi results for each of the 14 principles while integrating ChatGPT in IoT–based software wearable devices.

Table 5

Fuzzified decision matrix for the labeled criterion.

C	C-1			C-2			C-3			C-4
C-1	1	1	1	4	5	6	6	7	8	6	7	8
C-2	0.17	0.2	0.25	1	1	1	6	7	8	2	3	4
C-3	0.13	0.143	0.17	0.125	0.143	0.17	1	1	1	6	7	8
C-4	0.13	0.143	0.17	0.25	0.33	0.5	0.125	0.143	0.17	1	1	1

Table 6

Performance fuzzified matrix with respect to category-1.

C-1	C-1			C-2			C-3			C-4
WDC-1	9	9	9	6	7	8	9	9	9	6	7	8
WDC-2	1	1	1	1	1	1	6	7	8	6	7	8
WDC-3	6	7	8	2	3	4	1	1	1	6	7	8
WDC-4	1	1	1	6	7	8	9	9	9	4	5	6
WDC-5	6	7	8	9	9	9	4	5	6	6	7	8
WDC-6	4	5	6	4	5	6	2	3	4	6	7	8
WDC-7	4	5	6	6	7	8	4	5	6	6	7	8
WDC-8	2	3	4	1	1	1	2	3	4	2	3	4
WDC-9	4	5	6	4	5	6	4	5	6	1	1	1
WDC-10	1	1	1	6	7	8	1	1	1	4	5	6
WDC-11	4	5	6	1	1	1	4	5	6	4	5	6
WDC-12	4	5	6	2	3	4	6	7	8	2	3	4
WDC-13	6	7	8	2	3	4	1	1	1	2	3	4
WDC-14	6	7	8	6	7	8	2	3	4	1	1	1

Table 7

Geometric mean of each criterion.

	Geometric mean of all criteria
C-1	3.46410162	3.956321	4.42672768
C-2	1.18920712	1.43156912	1.68179283
C-3	0.55334096	0.614788153	0.686589048
C-4	0.25	0.287190894	0.343294525
Sum	5.4566497	6.289869167	7.138404083
Inverse values	0.183262635	0.158985819	0.140087334
Increasing order	0.14008733	0.158985819	0.183262635

Table 8

Aggregate fuzzy weights of every criterion.

	Fuzzy weights			Average weights	Normalize values
C-1	0.485277	0.628999	0.811254	1.925529	0.626675
C-2	0.166593	0.227599	0.30821	0.702402	0.228601
C-3	0.077516	0.097743	0.125826	0.301085	0.09799
C-4	0.035022	0.045659	0.062913	0.143594	0.046734
Sum	0.764407	1	1.308203	3.07261	1

Table 9

Combined decision matrix of all the principles of (with respect to) each criterion.

Weights	C-1			C-2			C-3			C-4
	0.51	0.6	0.9	0.21	0.21	0.31	0.08	0.1	0.2	0.042	0.05	0.07
Combined decision matrix
WDC-1	4	5	8	6	7.5	9	1	7	9	2	6	8
WDC-2	2	3.5	8	1	3	9	1	6	8	4	6.5	8
WDC-3	2	5	8	2	5	8	1	3.5	9	4	5.5	8
WDC-4	1	5	8	4	6	8	1	3	9	2	5.5	8
WDC-5	2	5	8	1	3	9	2	5.5	8	4	5.5	8
WDC-6	2	4	6	2	5.5	8	2	4.5	8	2	5.5	8
WDC-7	2	4	6	1	2.5	9	2	5.5	8	4	6	8
WDC-8	2	3.5	6	1	2	9	4	4.5	8	2	3.5	6
WDC-9	2	3.5	6	1	4.5	8	2	4.5	8	1	3	9
WDC-10	1	4	8	4	6	8	1	1	9	2	4.5	8
WDC-11	2	3.5	6	1	2.5	9	4	5.5	8	2	4.5	8
WDC-12	4	5	6	2	4.5	6	6	7	8	2	4.5	6
WDC-13	2	5.5	8	2	4.5	8	1	1	9	2	3.5	6
WDC-14	2	5	8	2	5.5	8	2	4	8	1	1	9

Table 10

Fuzzy normalized decision matrix.

Weights	C-1			C-2			C-3			C-4
	0.51	0.6	0.9	0.21	0.21	0.31	0.08	0.1	0.2	0.042	0.05	0.07
Fuzzy normalized decision matrix
WDC-1	0.5	0.6	1.0	0.7	0.8	1.0	0.1	0.1	1.0	0.2	0.7	0.9
WDC-2	0.3	0.4	1.0	0.1	0.3	1.0	0.1	0.2	1.0	0.4	0.7	0.9
WDC-3	0.3	0.6	1.0	0.2	0.6	0.9	0.1	0.3	1.0	0.4	0.6	0.9
WDC-4	0.1	0.6	1.0	0.4	0.7	0.9	0.1	0.3	1.0	0.2	0.6	0.9
WDC-5	0.3	0.6	1.0	0.1	0.3	1.0	0.1	0.2	0.5	0.4	0.6	0.9
WDC-6	0.3	0.5	0.8	0.2	0.6	0.9	0.1	0.2	0.5	0.2	0.6	0.9
WDC-7	0.3	0.5	0.8	0.1	0.3	1.0	0.1	0.2	0.5	0.4	0.7	0.9
WDC-8	0.3	0.4	0.8	0.1	0.2	1.0	0.1	0.2	0.3	0.2	0.4	0.7
WDC-9	0.3	0.4	0.8	0.1	0.5	0.9	0.1	0.2	0.5	0.1	0.3	1.0
WDC-10	0.1	0.5	1.0	0.4	0.7	0.9	0.1	1.0	1.0	0.2	0.5	0.9
WDC-11	0.3	0.4	0.8	0.1	0.3	1.0	0.1	0.2	0.3	0.2	0.5	0.9
WDC-12	0.5	0.6	0.8	0.2	0.5	0.7	0.1	0.1	0.2	0.2	0.5	0.7
WDC-13	0.3	0.7	1.0	0.2	0.5	0.9	0.1	1.0	1.0	0.2	0.4	0.7
WDC-14	0.3	0.6	1.0	0.2	0.6	0.9	0.1	0.3	0.5	0.1	0.1	1.0

Table 11

Normalized weighted fuzzy decision matrix.

	C-1			C-2			C-3			C-4
Normalized weighted fuzzy decision matrix
WDC-1	0.3	0.4	0.9	0.1	0.2	0.3	0.0	0.0	0.2	0.0	0.0	0.1
WDC-2	0.1	0.3	0.9	0.0	0.1	0.3	0.0	0.0	0.2	0.0	0.0	0.1
WDC-3	0.1	0.4	0.9	0.0	0.1	0.3	0.0	0.0	0.2	0.0	0.0	0.1
WDC-4	0.1	0.4	0.9	0.1	0.1	0.3	0.0	0.0	0.2	0.0	0.0	0.1
WDC-5	0.1	0.4	0.9	0.0	0.1	0.3	0.0	0.0	0.1	0.0	0.0	0.1
WDC-6	0.1	0.3	0.7	0.0	0.1	0.3	0.0	0.0	0.1	0.0	0.0	0.1
WDC-7	0.1	0.3	0.7	0.0	0.1	0.3	0.0	0.0	0.1	0.0	0.0	0.1
WDC-8	0.1	0.3	0.7	0.0	0.0	0.3	0.0	0.0	0.1	0.0	0.0	0.0
WDC-9	0.1	0.3	0.7	0.0	0.1	0.3	0.0	0.0	0.1	0.0	0.0	0.1
WDC-10	0.1	0.3	0.9	0.1	0.1	0.3	0.0	0.1	0.2	0.0	0.0	0.1
WDC-11	0.1	0.3	0.7	0.0	0.1	0.3	0.0	0.0	0.1	0.0	0.0	0.1
WDC-12	0.3	0.4	0.7	0.0	0.1	0.2	0.0	0.0	0.0	0.0	0.0	0.0
WDC-13	0.1	0.4	0.9	0.0	0.1	0.3	0.0	0.1	0.2	0.0	0.0	0.0
WDC-14	0.1	0.4	0.9	0.0	0.1	0.3	0.0	0.0	0.1	0.0	0.0	0.1

Table 12

FNI and FPI solutions.

${\mathrm{A}}^{+\text{ve}}$	${\mathrm{A}}^{-\mathrm{v}\mathrm{e}}$
0.3	0.1
0.4	0.2
0.9	0.5
0.1	0.0
0.2	0.0
0.3	0.1
0.0	0.0
0.1	0.0
0.2	0.0
0.0	0.0
0.0	0.0
0.1	0.0

Table 13

Distance from a fuzzy positive ideal solution.

Distance from FPIS					DI+
WDC-1	0.0	0.0	0.0	0.0	0.1
WDC-2	0.1	0.1	0.1	0.1	0.4
WDC-3	0.1	0.1	0.1	0.1	0.3
WDC-4	0.1	0.0	0.0	0.0	0.2
WDC-5	0.1	0.1	0.1	0.1	0.3
WDC-6	0.2	0.2	0.1	0.1	0.5
WDC-7	0.2	0.2	0.2	0.1	0.6
WDC-8	0.2	0.2	0.2	0.1	0.6
WDC-9	0.2	0.2	0.1	0.1	0.6
WDC-10	0.1	0.1	0.0	0.0	0.3
WDC-11	0.2	0.2	0.2	0.1	0.6
WDC-12	0.1	0.1	0.1	0.1	0.5
WDC-13	0.1	0.1	0.1	0.1	0.3
WDC-14	0.1	0.1	0.1	0.1	0.2

⁺Indicates distance from positive ideal solution.

Table 14

Distance from fuzzy negative ideal solution.

Distance from fuzzy negative ideal solution					DI−
WDC-1	0.3	0.3	0.3	0.1	1.00
WDC-2	0.3	0.3	0.3	0.1	0.89
WDC-3	0.3	0.3	0.3	0.1	0.92
WDC-4	0.3	0.3	0.3	0.1	0.93
WDC-5	0.3	0.3	0.3	0.1	0.92
WDC-6	0.1	0.1	0.1	0.1	0.52
WDC-7	0.1	0.1	0.1	0.1	0.52
WDC-8	0.1	0.1	0.1	0.1	0.50
WDC-9	0.1	0.1	0.1	0.1	0.50
WDC-10	0.3	0.3	0.3	0.1	0.91
WDC-11	0.1	0.1	0.1	0.1	0.51
WDC-12	0.2	0.2	0.1	0.1	0.56
WDC-13	0.3	0.3	0.3	0.1	0.93
WDC-14	0.3	0.3	0.3	0.1	0.92

⁻Shows distance from fuzzy negative ideal solution.

Table 15

Closeness coefficient index (CCi) values and ranks for every principle.

WDC. no	Principles	CCi	Global ranks
WDC-1	Data security and privacy	0.93	1
WDC-2	Data storage	0.54	8
WDC-3	Data bias	0.66	5
WDC-4	Design and comfort	0.71	2
WDC-5	Transparent communication	0.61	7
WDC-6	Battery consumption	0.25	10
WDC-7	Explainability	0.18	11
WDC-8	Quality issue	0.14	14
WDC-9	Compatibility	0.17	12
WDC-10	User understanding	0.65	6
WDC-11	Trust	0.15	13
WDC-12	Generalizability	0.34	9
WDC-13	Cost issue	0.66	4
WDC-14	Data hallucination	0.67	3

Abbreviations: GR = global ranking; WDC = wearable device challenge.

[figure(s) omitted; refer to PDF]

In Figure 9 of global ranking and prioritization, the most critical software wearable device principle while integrating ChatGPT is “data security and privacy (WDC-1)” with a largest ranking of CCi value of “0.925,” because ChatGPT and IoT–based wearable devices hold private information of users directly open to anyone to access, which put users data at risk. The second principle for IoT–based devices and wearables while integrating ChatGPT is “design and comfort (WDC-4)” with a global ranking of CCi value of “0.707,” as users have diverse demands, so the design and other features of software wearable devices while integrating ChatGPT can be main concerns for developers. The third principle for IoT–based devices and wearables while integrating ChatGPT is “data hallucination (WDC-14)” with a global ranking of CCi value of “0.665,” as sometimes ChatGPT produce false information while asking for results, known as data hallucination, i.e., not all results of ChatGPT are accurate.

[figure(s) omitted; refer to PDF]

As identified that the principle of “design and comfort” is the fourth one with a frequency of 78% from the identified selected articles of SLR shown in Table 3, but from Fuzzy-TOPSIS results, this principle is on the second position with CCi value 0.707; similarly, the principle of “data hallucination” is on the 14^th position from relevant literature, but from Fuzzy-TOPSIS results, this is the third most critical barrier while integrating ChatGPT into IoT–based devices and wearables with a CCi value 0.665, which means that data hallucination is also a major issue for IoT–based devices and wearables while integrating ChatGPT. From the literature as well as from the Fuzzy-TOPSIS ranking approach, it is also clear that “data security and privacy” is the most critical principle for software wearable devices while integrating ChatGPT into IoT–based devices and wearables with the frequency of 87% from the relevant literature and CCi value of 0.925. The proposed taxonomy-based analysis of SLR findings and Fuzzy-TOPSIS results is shown in Figure 10 in detail. So, the developers and researchers must focus on these principles while integrating ChatGPT into IoT–based devices and wearables’ development and research for better outcomes.

[figure(s) omitted; refer to PDF]

5. Conclusion and Future Work

In this paper, the author identified 14 ethical principles from literature via SLR while integrating ChatGPT into IoT–based devices and wearables keeping the frequency criteria ≥ 50% for each principle. Next, we applied the technique of Fuzzy-TOPSIS to evaluate, rank, and prioritize the identified ethical principles, which were also used by so many other researchers while making uncertain decision-making [23, 60, 66, 67, 70].

For this, the author categorized these principles into four categories on the basis of similarities. The findings of the Fuzzy-TOPSIS approach describe that the category-2 principle of data security and privacy is the top most ethical principle for software wearable devices while integrating ChatGPT with a CCi value of 0.925. Our final results provide a taxonomy based on the prioritization of software wearable device principles for manufacturer industries to deliberate the maximum ranking and analysis of IoT–based software wearable device principles ethically.

In addition, the author planned to empirically evaluate our findings by a detailed questionnaire survey from the experts of the same area. Next, we will employ the interpretive structure modeling (ISM) approach to develop a level-based decision model and analyze the relationship among ethical principles of IoT–based device and wearables while integrating ChatGPT. We will further conduct the cross impact matrix multiplication applied to classification (MICMAC) technique for creating a cluster-based decision model. These models will aim to help the software organizations and developers ethically integrate ChatGPT into IoT–based devices and wearables by following the identified principles by considering the motivators and highlighting the demotivators. The proposed pioneer findings will be established as a benchmark for integrating ChatGPT into IoT–based devices and wearables’ development and research with an emphasis on ethical deliberations.

Finally, while IoT–based software wearable devices present several ethical principles, they also offer significant opportunities for customers and industries. By focusing on these ethical principles, its prioritization, and capitalizing on these opportunities, we can ethically integrate ChatGPT into IoT–based devices and wearables and the IoT–based software wearable devices can remain to develop and progress to meet the customers’ requirements and preferences (See Table A1).

Funding

The research article “Ethical Principles of Integrating ChatGPT into IoT–based Software Wearables: A Fuzzy-TOPSIS Ranking and Analysis Approach, manuscript ID: 6660868,” was supported by the Key Program of the National Natural Science Foundation of China (Grant no. 62432010).

Acknowledgments

We are very much thankful to Dalian University of Technology, Liaoning, China for the support in our research.

Appendix

Appendix A: List of Final 77 Selected Articles for Research

Table A1

The final 77 selected articles used for this research study.

P.IDs	Papers
p1	Wang, Q., et al., Blockchain for the IoT and industrial IoT: A review. Internet of Things, 2020. 10: p. 100081
p2	Pino, A. F. S., et al., Systematic literature review on mechanisms to measure the technological maturity of the Internet of Things in enterprises. Internet of Things, 2024. 25: p. 101082
p3	Shaikh, T. A., T. Rasool, and P. Verma, Machine intelligence and medical cyber-physical system architectures for smart healthcare: Taxonomy, challenges, opportunities, and possible solutions. Artificial Intelligence in Medicine, 2023. 146: p. 102692
p4	Nikpour, M., et al., Intelligent energy management with IoT framework in smart cities using intelligent analysis: An application of machine learning methods for complex networks and systems. Journal of Network and Computer Applications, 2025. 235: p. 104089
p5	Liu, H., et al., A review of the smart world. Future Generation Computer Systems, 2019. 96: p. 678–691
p6	Nahavandi, D., et al., Application of artificial intelligence in wearable devices: Opportunities and challenges. Computer Methods and Programs in Biomedicine, 2022. 213: p. 106541
p7	Ali, J., et al., A deep dive into cybersecurity solutions for AI–driven IoT–enabled smart cities in advanced communication networks. Computer Communications, 2025. 229: p 108000
p8	Jahid, A., M. H. Alsharif, and T. J. Hall, The convergence of blockchain, IoT and 6G: Potential, opportunities, challenges and research roadmap. Journal of Network and Computer Applications, 2023. 217: p. 103677
p9	Ullah, I., et al., Integration of data science with the intelligent IoT (IIoT): Current challenges and future perspectives. Digital Communications and Networks, 2024
p10	Ghosh, A., D. Chakraborty, and A. Law, Artificial Intelligence in Internet of Things. CAAI Transactions on Intelligence Technology, 2018. 3
p11	Marengo, A., Navigating the nexus of AI and IoT: A comprehensive review of data analytics and privacy paradigms. Internet of Things, 2024. 27: p. 101318
p12	Ramadan, M. N. A., et al., AI–powered IoT and UAV systems for real-time detection and prevention of illegal logging. Results in Engineering, 202424: p. 103277
p13	Veiga, T., et al., Towards containerized, reuse-oriented AI deployment platforms for cognitive IoT applications. Future Generation Computer Systems, 2023. 142: p. 4–13
p14	Raman, R., P. Calyam, and K. Achuthan, ChatGPT or Bard: Who is a better Certified Ethical Hacker? Computers & Security, 2024. 140: p. 103804
p15	Qiu, Y. and Y. Jin, ChatGPT and fine-tuned BERT: A comparative study for developing intelligent design support systems. Intelligent Systems with Applications, 2024. 21: p. 200308
p16	Gheewala, S., et al., Exploiting deep transformer models in textual review based recommender systems. Expert Systems with Applications, 2024. 235: p. 121120
p17	Gill, S. S. and R. Kaur, ChatGPT: Vision and challenges. Internet of Things and Cyber-Physical systems, 2023. 3: p. 262–271
p18	Khennouche, F., et al., Revolutionizing generative pretrained: Insights and challenges in deploying ChatGPT and generative chatbots for FAQs. Expert Systems with Applications, 2024. 246: p. 123224
p19	Bergomi, L., et al., Reshaping free-text radiology notes into structured reports with generative question answering transformers. Artificial Intelligence in Medicine, 2024.154: p. 102924.
p20	Dwivedi, Y. K., et al., Opinion Paper: “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policy. International Journal of Information Management, 2023. 71: p. 102642
p21	Casheekar, A., et al., A contemporary review on chatbots, AI–powered virtual conversational agents, ChatGPT: Applications, open challenges and future research directions. Computer Science Review, 2024. 52: p. 100632
p22	Kuhail, M. A., et al., “Will I be replaced?” Assessing ChatGPT’s effect on software development and programmer perceptions of AI tools. Science of Computer Programming, 2024. 235: p. 103111
p23	Al-Hawawreh, M., A. Aljuhani, and Y. Jararweh, ChatGPT for cybersecurity: practical applications, challenges, and future directions. Cluster Computing, 2023. 26 (6): p. 3421–3436
p24	Kassab, M. and J.F. DeFranco, Unleashing the Potential: Integrating ChatGPT and the Internet of Things for Enhanced User Experiences and Automation. Computer, 2023. 56 (12): p. 91–94
p25	Xia, Y., et al., Enhancing intelligent IoT services development by integrated multi-token code completion. Computer Communications, 2023. 212: p. 313–323
p26	Ahmad, A., et al., Enhancing ChatGPT’s Querying Capability with Voice-Based Interaction and CNN-Based Impaired Vision Detection Model. Computers, Materials and continua, 2024. 78 (3): p. 3129–3150
p27	Mahmood, A., et al., User Interaction Patterns and Breakdowns in Conversing with LLM-Powered Voice Assistants. International Journal of Human-Computer Studies, 2025. 195: p. 103406
p28	Huynh-The, T., et al., Artificial intelligence for the metaverse: A survey. Engineering Applications of Artificial Intelligence, 2023. 117: p. 105581
p29	Luo, H., J. Luo, and A.V. Vasilakos, BC4LLM: A perspective of trusted artificial intelligence when blockchain meets large language models. Neurocomputing, 2024. 599: p. 128089
p30	Ray, P. P., ChatGPT: A comprehensive review on background, applications, key challenges, bias, ethics, limitations and future scope. Internet of Things and Cyber-Physical Systems, 2023. 3: p. 121–154
p31	Seneviratne, S., et al., A survey of wearable devices and challenges. IEEE Communications Surveys & Tutorials, 2017. 19 (4): p. 2573–2620
p32	Pasricha, S. and M. Wolf, Ethical Design of Computers: From Semiconductors to IoT and Artificial Intelligence. IEEE Design & Test, 2023
p33	Chakraborty, C., et al., From machine learning to deep learning: Advances of the recent data-driven paradigm shift in medicine and healthcare. Current Research in Biotechnology, 2024. 7: p. 100164
p34	Neha, F., et al., ChatGPT: Transforming Healthcare with AI. AI, 2024. 5 (4): p. 2618–2650
p35	Binder, M. and V. Mezhuyev, A framework for creating an IoT system specification with ChatGPT. Internet of Things, 2024. 27: p. 101218
p36	Sepasgozar, S., et al., A Systematic Content Review of Artificial Intelligence and the Internet of Things Applications in Smart Home. Applied Sciences, 2020. 10 (9): p. 3074
p37	Shi, F., et al., Recent Progress on the Convergence of the Internet of Things and Artificial Intelligence. IEEE Network, 2020. 34 (5): p. 8–15
p38	Marengo, A., Navigating the nexus of AI and IoT: A comprehensive review of data analytics and privacy paradigms. Internet of Things, 2024.27: p. 101318.
p39	Kim, H., et al., Research challenges and security threats to AI–driven 5G virtual emotion applications using autonomous vehicles, drones, and smart devices. IEEE network, 2020. 34 (6): p. 288–294
p40	Oliveira, F., et al., Internet of Intelligent Things: A convergence of embedded systems, edge computing and machine learning. Internet of Things, 2024. 26: p. 101153
p41	Wang, M., et al., Security and privacy in 6G networks: New areas and new challenges. Digital Communications and Networks, 2020. 6 (3): p. 281–291
p42	Zhang, Y., et al., MetaCity: Data-driven sustainable development of complex cities. The Innovation, 2025: p. 100775
p43	Sun, P., et al., A survey on privacy and security issues in IoT–based environments: Technologies, protection measures and future directions. Computers & Security, 2025. 148: p. 104097
p44	Le, T. and S. Shetty, Artificial intelligence-aided privacy preserving trustworthy computation and communication in 5G-based IoT networks. Ad Hoc Networks, 2022. 126: p. 102752
p45	Gerke, S., T. Minssen, and G. Cohen, Ethical and legal challenges of artificial intelligence-driven healthcare, in Artificial intelligence in healthcare. 2020, Elsevier. p. 295–336
p46	Keshta, I., AI–driven IoT for smart health care: Security and privacy issues. Informatics in Medicine Unlocked, 2022. 30: p. 100903
p47	Nguyen, T., H. Nguyen, and T. Nguyen Gia, Exploring the integration of edge computing and blockchain IoT: Principles, architectures, security, and applications. Journal of Network and Computer Applications, 2024. 226: p. 103884
p48	Kakhi, K., et al., The internet of medical things and artificial intelligence: trends, challenges, and opportunities. Biocybernetics and Biomedical Engineering, 2022. 42 (3): p. 749–771
p49	Ismail Keshta, AI–driven IoT for smart health care: Security and privacy issues, Informatics in Medicine Unlocked, Volume 30, 2022, 100903, ISSN 2352–9148, https://doi.org/10.1016/j.imu.2022.100903
p50	Alwahedi, F., et al., Machine learning techniques for IoT security: Current research and future vision with generative AI and large language models. Internet of Things and Cyber-Physical Systems, 2024. 4: p. 167–185
p51	Mohanta, B. K., et al., Survey on IoT security: Challenges and solution using machine learning, artificial intelligence and blockchain technology. Internet of Things, 2020. 11: p. 100227
p52	Cambiaso, E., et al., Advancements on IoT and AI applied to Pneumology. Microprocessors and Microsystems, 2024. 108: p. 105062
p53	Veiga, T., et al., Towards containerized, reuse-oriented AI deployment platforms for cognitive IoT applications. Future Generation Computer Systems, 2023. 142: p. 4–13
p54	Casheekar, A., et al., A contemporary review on chatbots, AI–powered virtual conversational agents, ChatGPT: Applications, open challenges and future research directions. Computer Science Review, 2024. 52: p. 100632
p55	Seng, K. P., et al., Machine learning and AI technologies for smart wearables. Electronics, 2023. 12 (7): p. 1509
p56	Lund, B. D., et al., ChatGPT and a new academic reality: Artificial Intelligence–written research papers and the ethics of the large language models in scholarly publishing. Journal of the Association for Information Science and Technology, 2023. 74 (5): p. 570–581
p57	Wang, Y., et al., A Survey on ChatGPT: AI–Generated Contents, Challenges, and Solutions. IEEE Open Journal of the Computer Society, 2023. 4: p. 280–302
p58	Almuhanna, H., et al., AI Asthma Guard: Predictive Wearable Technology for Asthma Management in Vulnerable Populations. Applied System Innovation, 2024. 7 (5): p. 78
p59	Yao, Y., et al., A survey on large language model (LLM) security and privacy: The Good, The Bad, and The Ugly. High-Confidence Computing, 2024. 4 (2): p. 100211
p60	Soliman, M. M., et al., Artificial intelligence-powered Metaverse: Analysis, challenges and future perspectives. Artificial Intelligence Review, 2024. 57 (2): p. 36
p61	Kumar, Y. and V. Kumar, Security in IoT systems using natural language processing: Future challenges and directions. Internet Technology Letters, 2023. 6 (4): p. e411
p62	Zhang, Q., et al., LLM-TSFD: An industrial time series human-in-the-loop fault diagnosis method based on a large language model. Expert Systems with Applications, 2025. 264: p. 125861
p63	Kalutharage, C. S., X. Liu, and C. Chrysoulas, Neurosymbolic learning and domain knowledge-driven explainable AI for enhanced IoT network attack detection and response. Computers & Security, 2025. 151: p. 104318
p64	Okey, O. D., et al., Investigating ChatGPT and cybersecurity: A perspective on topic modeling and sentiment analysis. Computers & Security, 2023. 135: p. 103476
p65	Qiu, Y. and Y. Jin, ChatGPT and fine-tuned BERT: A comparative study for developing intelligent design support systems. Intelligent Systems with Applications, 2024. 21: p. 200308
p66	Lauriola, I., A. Lavelli, and F. Aiolli, An introduction to Deep Learning in Natural Language Processing: Models, techniques, and tools. Neurocomputing, 2022. 470: p. 443–456
p67	Straub, V. J., et al., Artificial intelligence in government: Concepts, standards, and a unified framework. Government Information Quarterly, 2023. 40 (4): p. 101881
p68	Cao, K., et al., DMSeqNet-mBART: A state-of-the-art Adaptive-DropMessage enhanced mBART architecture for superior Chinese short news text summarization. Expert Systems with Applications, 2024. 257: p. 125095
p69	Zohourian, A., et al., IoT–PRIDS: Leveraging packet representations for intrusion detection in IoT networks. Computers & Security, 2024. 146: p. 104034
p70	Msigwa, C., D. Bernard, and J. Yun, IoT–driven wearable devices enhancing healthcare: ECG classification with cluster-based GAN and meta-features. Internet of Things, 2024. 28: p. 101405
p71	Min, B., et al., Recent advances in natural language processing via large pretrained language models: A survey. ACM Computing Surveys, 2023. 56 (2): p. 1–40
p72	Annepaka, Y. and P. Pakray, Large language models: a Survey of their development, capabilities, and applications. Knowledge and Information Systems, 2024
p73	Chen, Z., et al., Evolution and Prospects of Foundation Models: From Large Language Models to Large MultimodalModels. Computers, Materials and Continua, 2024. 80 (2): p. 1753–1808
p74	Sworna, N. S., et al., Towards development of IoT–ML–driven healthcare systems: A survey. Journal of Network and Computer Applications, 2021. 196: p. 103244
p75	Arslan, S., M. Ozkaya, and G. Kardas, Modeling Languages for Internet of Things (IoT) Applications: A Comparative Analysis Study. Mathematics, 2023. 11 (5): p. 1263
p76	Poongodi, T., et al., Wearable devices and IoT. A handbook of Internet of Things in biomedical and cyber-physical system, 2020: p. 245–273
p77	Mocrii, D., Y. Chen, and P. Musilek, IoT–based smart homes: A review of system architecture, software, communications, privacy and security. Internet of Things, 2018. 1: p. 81–98

References

[1] A. Zohourian, S. Dadkhah, H. Molyneaux, E. C. P. Neto, A. A. Ghorbani, "IoT-PRIDS: Leveraging 6 Acket Representations for Intrusion Detection in IoT Networks," Computers & Security, vol. 146,DOI: 10.1016/j.cose.2024.104034, 2024.

[2] T. Poongodi, R. Krishnamurthi, R. Indrakumari, P. Suresh, B. Balusamy, "Wearable Devices and IoT," Intelligent Systems Reference Library, pp. 245-273, DOI: 10.1007/978-3-030-23983-1_10, 2020.

[3] S. Arslan, M. Ozkaya, G. Kardas, "Modeling Languages for Internet of Things (IoT) Applications: A Comparative Analysis Study," Mathematics, vol. 11 no. 5,DOI: 10.3390/math11051263, 2023.

[4] D. Mocrii, Y. Chen, P. Musilek, "IoT-based Smart Homes: A Review of System Architecture, Software, Communications, Privacy and Security," Internet of Things, vol. 1-2, pp. 81-98, DOI: 10.1016/j.iot.2018.08.009, 2018.

[5] C. Msigwa, D. Bernard, J. Yun, "IoT-Driven Wearable Devices Enhancing Healthcare: ECG Classification With Cluster-Based GAN and Meta-Features," Internet of Things, vol. 28,DOI: 10.1016/j.iot.2024.101405, 2024.

[6] N. S. Sworna, A. K. M. M. Islam, S. Shatabda, S. Islam, "Towards Development of IoT-ML Driven Healthcare Systems: A Survey," Journal of Network and Computer Applications, vol. 196,DOI: 10.1016/j.jnca.2021.103244, 2021.

[7] B. Min, H. Ross, E. Sulem, "Recent Advances in Natural Language Processing via Large Pre-Trained Language Models: A Survey," ACM Computing Surveys, vol. 56 no. 2,DOI: 10.1145/3605943, 2023.

[8] Z. Chen, L. Xu, H. Zheng, "Evolution and Prospects of Foundation Models: From Large Language Models to Large Multimodal Models," Computers, Materials & Continua, vol. 80 no. 2, pp. 1753-1808, DOI: 10.32604/cmc.2024.052618, 2024.

[9] Y. Annepaka, P. Pakray, "Large Language Models: A Survey of Their Development, Capabilities, and Applications," Knowledge and Information Systems, vol. 67 no. 3, pp. 2967-3022, DOI: 10.1007/s10115-024-02310-4, 2024.

[10] K. Cao, W. Cheng, Y. Hao, "DMSeqNet-mBART: A State-Of-The-Art Adaptive-DropMessage Enhanced mBART Architecture for Superior Chinese Short News Text Summarization," Expert Systems with Applications, vol. 257,DOI: 10.1016/j.eswa.2024.125095, 2024.

[11] V. J. Straub, D. Morgan, J. Bright, H. Margetts, "Artificial Intelligence in Government: Concepts, Standards, and a Unified Framework," Government Information Quarterly, vol. 40 no. 4,DOI: 10.1016/j.giq.2023.101881, 2023.

[12] I. Lauriola, A. Lavelli, F. Aiolli, "An Introduction to Deep Learning in Natural Language Processing: Models, Techniques, and Tools," Neurocomputing, vol. 470, pp. 443-456, DOI: 10.1016/j.neucom.2021.05.103, 2022.

[13] Y. Qiu, Y. Jin, "ChatGPT and Finetuned BERT: A Comparative Study for Developing Intelligent Design Support Systems," Intelligent Systems With Applications, vol. 21,DOI: 10.1016/j.iswa.2023.200308, 2024.

[14] O. D. Okey, E. U. Udo, R. L. Rosa, D. Z. Rodriguez, J. Kleinschmidt, "Investigating ChatGPT and Cybersecurity: A Perspective on Topic Modeling and Sentiment Analysis," Computers & Security, vol. 135,DOI: 10.1016/j.cose.2023.103476, 2023.

[15] C. S. Kalutharage, X. Liu, C. Chrysoulas, "Neurosymbolic Learning and Domain Knowledge-Driven Explainable AI for Enhanced IoT Network Attack Detection and Response," Computers & Security, vol. 151,DOI: 10.1016/j.cose.2025.104318, 2025.

[16] Qi Zhang, C. Xu, J. Li, Y. Sun, J. Bao, D. Zhang, "LLM-TSFD: An Industrial Time Series Human-In-The-Loop Fault Diagnosis Method Based on a Large Language Model," Expert Systems With Applications, vol. 264,DOI: 10.1016/j.eswa.2024.125861, 2025.

[17] Y. Kumar, V. Kumar, "Security in IoT Systems Using Natural Language Processing: Future Challenges and Directions," Internet Technology Letters, vol. 6 no. 4,DOI: 10.1002/itl2.411, 2023.

[18] M. M. Soliman, E. Ahmed, A. Darwish, A. E. Hassanien, "Artificial Intelligence Powered Metaverse: Analysis, Challenges and Future Perspectives," Artificial Intelligence Review, vol. 57 no. 2,DOI: 10.1007/s10462-023-10641-x, 2024.

[19] Y. Wang, Y. Pan, M. Yan, Z. Su, T. H. Luan, "A Survey on ChatGPT: AI–Generated Contents, Challenges, and Solutions," IEEE Open Journal of the Computer Society, vol. 4, pp. 280-302, DOI: 10.1109/ojcs.2023.3300321, 2023.

[20] H. Almuhanna, M. Alenezi, M. Abualhasan, S. Alajmi, R. Alfadhli, A. S. Karar, "AI Asthma Guard: Predictive Wearable Technology for Asthma Management in Vulnerable Populations," Applied System Innovation, vol. 7 no. 5,DOI: 10.3390/asi7050078, 2024.

[21] M. A. Akbar, K. Smolander, S. Mahmood, A. Alsanad, "Toward Successful DevSecOps in Software Development Organizations: A Decision-Making Framework," Information and Software Technology, vol. 147,DOI: 10.1016/j.infsof.2022.106894, 2022.

[22] S. Rafi, M. A. Akbar, W. Yu, A. Alsanad, A. Gumaei, M. U. Sarwar, "Exploration of DevOps Testing Process Capabilities: An ISM and Fuzzy TOPSIS Analysis," Applied Soft Computing, vol. 116,DOI: 10.1016/j.asoc.2021.108377, 2022.

[23] A. W. Khan, M. U. Khan, J. A. Khan, "Identification and Prioritization of Security Challenges of Big Data on Cloud Computing Based on SLR: A Fuzzy-TOPSIS Analysis Approach," Journal of Software: Evolution and Process, vol. 33 no. 12,DOI: 10.1002/smr.2387, 2021.

[24] D. Nahavandi, R. Alizadehsani, A. Khosravi, U. R. Acharya, U. Rajendra, "Application of Artificial Intelligence in Wearable Devices: Opportunities and Challenges," Computer Methods and Programs in Biomedicine, vol. 213,DOI: 10.1016/j.cmpb.2021.106541, 2022.

[25] S. Seneviratne, Y. Hu, T. Nguyen, "A Survey of Wearable Devices and Challenges," IEEE Communications Surveys & Tutorials, vol. 19 no. 4, pp. 2573-2620, DOI: 10.1109/comst.2017.2731979, 2017.

[26] S. S. Gill, R. Kaur, "ChatGPT: Vision and Challenges," Internet of Things and Cyber-Physical Systems, vol. 3, pp. 262-271, DOI: 10.1016/j.iotcps.2023.05.004, 2023.

[27] B. D. Lund, T. Wang, N. R. Mannuru, B. Nie, S. Shimray, Z. Wang, "ChatGPT and a New Academic Reality: Artificial Intelligence‐written Research Papers and the Ethics of the Large Language Models in Scholarly Publishing," Journal of the Association for Information Science and Technology, vol. 74 no. 5, pp. 570-581, DOI: 10.1002/asi.24750, 2023.

[28] K. P. Seng, L. M. Ang, E. Peter, A. Mmonyi, A. Mmonyi, "Machine Learning and AI Technologies for Smart Wearables," Electronics, vol. 12 no. 7,DOI: 10.3390/electronics12071509, 2023.

[29] S. Pasricha, M. Wolf, "Ethical Design of Computers: From Semiconductors to IoT and Artificial Intelligence," 2022.

[30] H. Kim, J. Ben-Othman, L. Mokdad, J. Son, C. Li, "Research Challenges and Security Threats to AI-Driven 5G Virtual Emotion Applications Using Autonomous Vehicles, Drones, and Smart Devices," IEEE network, vol. 34 no. 6, pp. 288-294, DOI: 10.1109/mnet.011.2000245, 2020.

[31] A. Marengo, "Navigating the Nexus of AI and IoT: A Comprehensive Review of Data Analytics and Privacy Paradigms," Internet of Things, vol. 27,DOI: 10.1016/j.iot.2024.101318, 2024.

[32] F. Shi, H. Ning, W. Huangfu, "Recent Progress on the Convergence of the Internet of Things and Artificial Intelligence," IEEE Network, vol. 34 no. 5,DOI: 10.1109/mnet.011.2000009, 2020.

[33] S. Sepasgozar, R. Karimi, L. Farahzadi, "A Systematic Content Review of Artificial Intelligence and the Internet of Things Applications in Smart Home," Applied Sciences, vol. 10 no. 9,DOI: 10.3390/app10093074, 2020.

[34] M. Binder, V. Mezhuyev, "A Framework for Creating an IoT System Specification with ChatGPT," Internet of Things, vol. 27,DOI: 10.1016/j.iot.2024.101218, 2024.

[35] F. Neha, D. Bhati, D. K. Shukla, Md Amiruzzaman, "ChatGPT: Transforming Healthcare With AI," AIMplus, vol. 5 no. 4, pp. 2618-2650, DOI: 10.3390/ai5040126, 2024.

[36] C. Chakraborty, M. Bhattacharya, S. Pal, S.-S. Lee, "From Machine Learning to Deep Learning: Advances of the Recent Data-Driven Paradigm Shift in Medicine and Healthcare," Current Research in Biotechnology, vol. 7,DOI: 10.1016/j.crbiot.2023.100164, 2024.

[37] F. Oliveira, D. G. Costa, F. Assis, I. Silva, "Internet of Intelligent Things: A Convergence of Embedded Systems, Edge Computing and Machine Learning," Internet of Things, vol. 26,DOI: 10.1016/j.iot.2024.101153, 2024.

[38] P. Sun, Yi Wan, Z. Wu, Z. Fang, Qi Li, "A Survey on Privacy and Security Issues in IoT-Based Environments: Technologies, Protection Measures and Future Directions," Computers & Security, vol. 148,DOI: 10.1016/j.cose.2024.104097, 2025.

[39] Y. Zhang, Y. Lin, G. Zheng, "MetaCity: Data-Driven Sustainable Development of Complex Cities," Innovation, vol. 6 no. 2,DOI: 10.1016/j.xinn.2024.100775, 2025.

[40] T. Nguyen, H. Nguyen, T. Nguyen Gia, T. Nguyen, "Exploring the Integration of Edge Computing and Blockchain IoT: Principles, Architectures, Security, and Applications," Journal of Network and Computer Applications, vol. 226,DOI: 10.1016/j.jnca.2024.103884, 2024.

[41] Y. Yao, J. Duan, K. Xu, Y. Cai, Z. Sun, Y. Zhang, "A Survey on Large Language Model (Llm) Security and Privacy: The Good, the Bad, and the Ugly," High-Confidence Computing, vol. 4 no. 2,DOI: 10.1016/j.hcc.2024.100211, 2024.

[42] M. Zong, A. Hekmati, M. Guastalla, Y. Li, B. Krishnamachari, "Integrating Large Language Models With Internet of Things: Applications," Discover Internet of Things, vol. 5 no. 1,DOI: 10.1007/s43926-024-00083-4, 2025.

[43] Z. Wang, J. Li, S. Yang, X. Luo, D. Li, S. Mahmoodi, "A Lightweight IoT Intrusion Detection Model Based on Improved BERT-Of-Theseus," Expert Systems with Applications, vol. 238,DOI: 10.1016/j.eswa.2023.122045, 2024.

[44] C. Chen, X. Feng, Y. Li, "Integration of Large Language Models and Federated Learning," Patterns, vol. 5 no. 12,DOI: 10.1016/j.patter.2024.101098, 2024.

[45] J. Ali, S. Kumar Singh, W. Jiang, "A Deep Dive into Cybersecurity Solutions for AI-Driven IoT-Enabled Smart Cities in Advanced Communication Networks," Computer Communications, vol. 229,DOI: 10.1016/j.comcom.2024.108000, 2025.

[46] M. Kassab, J. F. DeFranco, "Unleashing the Potential: Integrating ChatGPT and the Internet of Things for Enhanced User Experiences and Automation," Computer, vol. 56 no. 12, pp. 91-94, DOI: 10.1109/mc.2023.3297788, 2023.

[47] I. Keshta, "AI-Driven IoT for Smart Health Care: Security and Privacy Issues," Informatics in Medicine Unlocked, vol. 30,DOI: 10.1016/j.imu.2022.100903, 2022.

[48] S. Gerke, T. Minssen, G. Cohen, "Ethical and Legal Challenges of Artificial Intelligence-Driven Healthcare," Artificial Intelligence in Healthcare, pp. 295-336, 2020.

[49] M. Wang, T. Zhu, T. Zhang, J. Zhang, S. Yu, W. Zhou, "Security and Privacy in 6G Networks: New Areas and New Challenges," Digital Communications and Networks, vol. 6 no. 3, pp. 281-291, DOI: 10.1016/j.dcan.2020.07.003, 2020.

[50] E. Cambiaso, S. Narteni, I. Baiardini, F. Braido, A. Paglialonga, M. Mongelli, "Advancements on IoT and AI Applied to Pneumology," Microprocessors and Microsystems, vol. 108,DOI: 10.1016/j.micpro.2024.105062, 2024.

[51] T. Veiga, H. A. Asad, F. A. Kraemer, K. Bach, K. Bach, "Towards Containerized, Reuse-Oriented AI Deployment Platforms for Cognitive IoT Applications," Future Generation Computer Systems, vol. 142,DOI: 10.1016/j.future.2022.12.029, 2023.

[52] A. Casheekar, A. Lahiri, K. Rath, K. S. Prabhakar, K. Srinivasan, K. Srinivasan, "A Contemporary Review on Chatbots, AI-Powered Virtual Conversational Agents, ChatGPT: Applications, Open Challenges and Future Research Directions," Computer Science Review, vol. 52,DOI: 10.1016/j.cosrev.2024.100632, 2024.

[53] K. Kakhi, R. Alizadehsani, H. M. D. Kabir, A. Khosravi, S. Nahavandi, U. R. Acharya, "The Internet of Medical Things and Artificial Intelligence: Trends, Challenges, and Opportunities," Biocybernetics and Biomedical Engineering, vol. 42 no. 3, pp. 749-771, DOI: 10.1016/j.bbe.2022.05.008, 2022.

[54] T. Le, S. Shetty, "Artificial Intelligence-Aided Privacy Preserving Trustworthy Computation and Communication in 5G-Based IoT Networks," Ad Hoc Networks, vol. 126,DOI: 10.1016/j.adhoc.2021.102752, 2022.

[55] F. Alwahedi, A. Aldhaheri, M. A. Ferrag, A. Battah, N. Tihanyi, "Machine Learning Techniques for IoT Security: Current Research and Future Vision with Generative AI and Large Language Models," Internet of Things and Cyber-Physical Systems, vol. 4, pp. 167-185, DOI: 10.1016/j.iotcps.2023.12.003, 2024.

[56] B. K. Mohanta, D. Jena, U. Satapathy, S. Patnaik, "Survey on IoT Security: Challenges and Solution Using Machine Learning, Artificial Intelligence and Blockchain Technology," Internet of Things, vol. 11,DOI: 10.1016/j.iot.2020.100227, 2020.

[57] B. Cayir Ervural, S. Zaim, O. F. Demirel, Z. Aydin, D. Delen, "An ANP and Fuzzy TOPSIS-Based SWOT Analysis for Turkey’s Energy Planning," Renewable and Sustainable Energy Reviews, vol. 82, pp. 1538-1550, DOI: 10.1016/j.rser.2017.06.095, 2018.

[58] C. Prakash, M. k Barua, Barua, "Integration of AHP-TOPSIS Method for Prioritizing the Solutions of Reverse Logistics Adoption to Overcome Its Barriers under Fuzzy Environment," Journal of Manufacturing Systems, vol. 37, pp. 599-615, DOI: 10.1016/j.jmsy.2015.03.001, 2015.

[59] C. Prakash, M. K. Barua, "An Analysis of Integrated Robust Hybrid Model for Third-Party Reverse Logistics Partner Selection Under Fuzzy Environment," Resources, Conservation and Recycling, vol. 108, pp. 63-81, DOI: 10.1016/j.resconrec.2015.12.011, 2016.

[60] M. Nazim, C. Wali Mohammad, M. Sadiq, "A Comparison Between Fuzzy AHP and Fuzzy TOPSIS Methods to Software Requirements Selection," Alexandria Engineering Journal, vol. 61 no. 12, pp. 10851-10870, DOI: 10.1016/j.aej.2022.04.005, 2022.

[61] V. Singh, V. Kumar, V. B. Singh, "A Hybrid Novel Fuzzy AHP-TOPSIS Technique for Selecting Parameter-Influencing Testing in Software Development," Decision Analytics Journal, vol. 6,DOI: 10.1016/j.dajour.2022.100159, 2023.

[62] S. Madhavi, N. C. Santhosh, S. Rajkumar, R. Praveen, "Pythagorean Fuzzy Sets-Based VIKOR and TOPSIS-Based Multi-Criteria Decision-Making Model for Mitigating Resource Deletion Attacks in WSNs," Journal of Intelligent and Fuzzy Systems, vol. 44 no. 6, pp. 9441-9459, DOI: 10.3233/jifs-224141, 2023.

[63] J. Jana, S. K. Roy, "Linguistic Pythagorean Hesitant Fuzzy Matrix Game and Its Application in Multi-Criteria Decision Making," Applied Intelligence, vol. 53 no. 1,DOI: 10.1007/s10489-022-03442-2, 2023.

[64] A. Saoud, M. Lachgar, M. Hanine, "decideXpert: Collaborative System Using AHP-TOPSIS and Fuzzy Techniques for Multicriteria Group Decision-Making," SoftwareX, vol. 29,DOI: 10.1016/j.softx.2024.102026, 2025.

[65] W. Kriswardhana, B. Toaza, E. Kiss, Domokos, Duleba, Szabolcs, "Analytic Hierarchy Process in Transportation Decision-Making: A Two-Staged Review on the Themes and Trends of Two Decades," Expert Systems With Applications, vol. 7, 2025.

[66] J. Liu, Qi Zhang, M. Xie, M. Lin, Z. Xu, "A Blockchain Platform Selection Method With Heterogeneous Multi-Criteria Decision-Making Based on Hybrid Distance Measures and an AHP-EWM Weight Method," Expert Systems with Applications, vol. 256,DOI: 10.1016/j.eswa.2024.124910, 2024.

[67] A. A. Magabaleh, L. L. Ghraibeh, A. Y. Audeh, "Systematic Review of Software Engineering Uses of Multi-Criteria Decision-Making Methods: Trends, Bibliographic Analysis, Challenges, Recommendations, and Future Directions," Applied Soft Computing, vol. 163,DOI: 10.1016/j.asoc.2024.111859, 2024.

[68] K. Vasantha Lakshmi, K. N. Udaya Kumara, "A Novel Randomized Weighted Fuzzy AHP by Using Modified Normalization with the TOPSIS for Optimal Stock Portfolio Selection Model Integrated with an Effective Sensitive Analysis," Expert Systems With Applications, vol. 243,DOI: 10.1016/j.eswa.2023.122770, 2024.

[69] S. F. Ismael, A. H. Alias, A. A. Zaidan, "Toward Sustainable Transportation: A Pavement Strategy Selection Based on the Extension of Dual-Hesitant Fuzzy Multicriteria Decision-Making Methods," IEEE Transactions on Fuzzy Systems, vol. 31 no. 2, pp. 380-393, DOI: 10.1109/tfuzz.2022.3168050, 2023.

[70] Y. B. Abushark, A. I. Khan, F. J. Alsolami, A. Almalawi, Md M. Alam, A. Agrawal, Kumar, Rajeev, Khan, R. Ahmad, "Usability Evaluation through Fuzzy AHP-TOPSIS Approach: Security Requirement Perspective," Computers, Materials & Continua, vol. 68, pp. 1203-1218, 2021.