Henoch Juli Christanto*
| Christine Dewi | Stephen Aprius Sutresno | Andri Dayarana K. Silalahi
© 2024 The authors. This article is published by IIETA and is licensed under the CC BY 4.0 license (http://creativecommons.org/licenses/by/4.0/).
OPEN ACCESS
This paper introduces the concepts of Chat Generative Pre-Trained Transformer (GPT) and artificial intelligence (AI). Chat GPT utilizes the GPT language model, which is trained using deep learning techniques and the transformer algorithm. It leverages the transformer's ability to understand human language and generate natural responses in conversations. ChatGPT is utilized in various contexts such as virtual assistants, chatbots, and interactive platforms to improve user interactions with technology. Our efforts also explore the wider domain of artificial intelligence, encompassing machine learning, deep learning, and natural language processing. The advancements in artificial intelligence (AI) technology have had a significant impact on various industries. The study emphasizes the significance of ongoing enhancement, safeguarding, confidentiality, and ethical deliberations in the creation and implementation of ChatGPT and AI chatbots. Ongoing research endeavors to improve the dependability and credibility of AI chatbot systems, despite obstacles such as bias and comprehensibility AI chatbots, can facilitate tailored and efficient human-machine interactions by giving priority to ethical considerations and promoting collaboration. In contemporary research initiatives, the integration of ChatGPT and AI technologies is of great significance, as it presents unique prospects for exploration and invention. ChatGPT, due to its capacity to understand and produce written content, functions as a potent instrument for enhancing communication, resolving issues, and disseminating knowledge in several fields. Hence, it is imperative for researchers to fully grasp the capabilities and consequences of AI, particularly on platforms like ChatGPT, to optimally harness the entire potential of these technologies in their respective fields.
artificial intelligence, ChatGPT, ethical considerations, natural language processing, transformer algorithm
The introduction to ChatGPT and AI involves the use of technology that encompasses ChatGPT and AI [1]. ChatGPT is an example of artificial intelligence technology based on the GPT language model. The GPT language model is one of several generative models that have been trained using deep learning techniques, particularly the transformer algorithm [2]. ChatGPT is an exceptionally proficient language model for producing conversational replies. Trained on a large amount of data, this model utilizes the generative capabilities of the transformer architecture to understand and generate text that is like human speech or interaction.
The natural language processing (NLP) tests that have been performed have demonstrated that the transformer architecture is extremely successful. For this discussion, GPT is an implementation of the transformer architecture, which is well-known for its "pre-trained" approach. The GPT model has undergone extensive training using a vast corpus of textual data, which has endowed it with a robust comprehension of human language. Chat GPT utilizes the linguistic comprehension of the GPT model to produce more authentic and fluent replies during discussions. ChatGPT can understand the context of texts and provide suitable responses when users ask questions or make assertions. This helps create a more realistic and responsive experience when interacting with ChatGPT [3, 4].
In the current landscape, the relevance of ChatGPT and AI technologies cannot be overstated. Several factors contribute to the importance of studying and understanding these technologies as follows: (1) Rapid Technological Advancements: AI, including models like ChatGPT, is undergoing rapid advancements, with new capabilities and applications emerging at an unprecedented pace. Academics, practitioners, and policymakers need to stay current on these advancements to successfully take advantage of the full potential of AI and navigate the ramifications of this technology. (2) Integration into Everyday Life: AI technologies, including ChatGPT, are increasingly integrated into various aspects of everyday life. Understanding how these technologies function, their limitations, and their impact on society is essential for individuals and organizations alike. (3) Ethical and Societal Implications: The deployment of AI raises complex ethical and societal implications, including concerns about bias, privacy, job displacement, and algorithmic accountability. Examining these issues in the context of ChatGPT and similar AI models is critical for ensuring responsible development and deployment practices. ChatGPT systems are frequently employed in diverse applications.
The ability of ChatGPT to understand human language naturally and generate meaningful responses has made it a popular solution in supporting user interactions with technology. Furthermore, the introduction also covers the concept of AI in general. AI is a field that focuses on creating machines capable of carrying out tasks that usually necessitate human intelligence. The methodologies employed in AI exhibit diversity, encompassing machine learning, deep learning, and natural language processing techniques akin to those utilized in ChatGPT. Machine learning is an expansive area of study that concentrates on creating algorithms that allow computers to acquire knowledge from data. On the other hand, deep learning is a specific branch of machine learning that employs neural networks with numerous layers to represent intricate patterns. NLP is a discipline that primarily concerns itself with the interface between computers and human languages. It utilizes machine learning and deep learning methods to understand and produce information on human language. The advancement of artificial intelligence (AI) and the investigation of diverse practical uses are greatly facilitated by the contributions from each field.
In recent years, advancements in AI technology, particularly in the field of deep learning, have opened new opportunities and had a significant impact across various sectors, such as automation, voice recognition, data analysis, and more [5, 6]. AI continues to evolve, and research is ongoing to improve its performance and explore further applications. Overall, ChatGPT and AI are two important aspects in the development of information technology that have a significant impact on how we interact with computers and the systems around us. This introduction opens the door to great potential in delivering innovative solutions.
The objective of this study is to provide clarity and categorization to past research and to assist readers, researchers, and practitioners who wish to delve into this area of study. The subsequent outline illustrates the organization of the current document. The paper elucidates the versatility of ChatGPT across various domains within AI research showcasing its adaptability in generating responses, analyzing data, and facilitating communication. Through empirical examples and case studies, we demonstrate how ChatGPT can be effectively utilized to streamline processes and enhance productivity in diverse contexts.
Before delving into the existing research documented in the pertinent literature, Section 2 presents a comprehensive overview of ChatGPT models that are presently available and actively utilized. Section 3 presents the applications, advantages, and hazards of ChatGPT. A thorough examination, analysis, and discourse on the ChatGPT difficulty will be presented in Section 4, offering a more extensive understanding. Section 5 presents the final findings and potential future considerations.
ChatGPT is a specific iteration of the GPT model created by OpenAI. GPT models utilize the transformer architecture, a neural network design renowned for its exceptional performance in applications related to natural language processing. ChatGPT is specifically engineered to provide responses that closely conversationally resemble those of a human. (1) Generative: It can generate coherent and contextually relevant text. (2) Pre-trained: It has been trained on a vast amount of diverse internet text before being fine-tuned for specific tasks [7, 8]. (3) Transformer: It uses a transformer architecture, which is effective in capturing long-range dependencies in data. A related section of the study covers some of the important research in the field of introduction to ChatGPT and AI. Researcher [9] introduces the ChatGPT model which is a large generative language model optimized for conversational agent applications. This model was trained using a pre-training and fine-tuning approach with large conversational datasets taken from the Internet. Our work demonstrates the ability of the ChatGPT model to generate consistent and meaningful responses in conversations. Furthermore, other research explores the use of transfer learning in the context of ChatGPT [10]. The authors propose an approach to leverage existing knowledge in the ChatGPT model to accelerate and enhance learning on specific tasks, such as translating text or providing specific information. The results of this study show the potential of transfer learning in increasing the efficiency and performance of the ChatGPT model [11, 12]. Researcher [13] focuses on personalizing ChatGPT using user feedback. The author proposes a method that allows users to provide feedback on the responses generated by ChatGPT and use that feedback to improve future responses. The results of this research show that personalized responses can improve user experience and make interactions with ChatGPT more relevant [14, 15].
The study [16] addresses the problem of bias in the responses generated by ChatGPT. The author proposes an approach to identify and reduce bias in the ChatGPT model, considering the diversity of the training dataset and using customized adjustment techniques. The results of this study demonstrate efforts to reduce bias in conversational systems and ensure fair and inclusive responses. The study [17] extends the capabilities of ChatGPT by considering multimodal inputs, such as text, images, and sound. The authors propose an approach that integrates natural language processing with image and sound processing to enable ChatGPT to understand contexts more richly and provide more informative responses. The results of this study demonstrate the potential of using multimodal input to improve the interaction and response quality of ChatGPT.
2.1 History of ChatGPT and feature-based approaches
One feature-based approach that is commonly used is text feature extraction. These features can include simple statistics such as word count or sentence length, or more complex features such as the use of keywords or sentiment analysis [18]. Using these features, the ChatGPT model can gain insights into the structure and content of conversations which helps in generating more informative and relevant responses. Conventional feature-based models in NLP frequently depend on manually designed features to represent different characteristics of language. These qualities may encompass linguistic, syntactic, or semantic characteristics. Although these approaches can be helpful in specific situations, they necessitate domain expertise and may not capture intricate patterns as efficiently as deep learning models [19, 20].
Figure 1. The development of ChatGPT
The development of ChatGPT is a component of OpenAI's broader efforts in language modeling, particularly focusing on the GPT architecture, as depicted in Figure 1. Below is a concise summary of the significant events that preceded the development of ChatGPT: (1) GPT-2 (2019): OpenAI initially introduced GPT-2, a language model of considerable magnitude comprising 1.5 billion parameters. The GPT-2 model exhibited remarkable prowess in producing text that was both cogent and contextually pertinent. Because of apprehensions regarding its potential for generating deceptive content, OpenAI initially restricted public access to the complete model but subsequently made it available to the public. (2) GPT-3 (2020): Building on the success of GPT-2, OpenAI introduced GPT-3, a much larger model with 175 billion parameters, making it one of the most powerful language models at the time. GPT-3 showcased remarkable language generation abilities, enabling it to perform a wide range of tasks, from translation to code generation, with minimal task-specific training. (3) ChatGPT (2020): ChatGPT, a variant of GPT-3 implemented by OpenAI, was purposefully developed to handle conversational interactions. It was refined to deliver responses that were more comprehensive and pertinent to the context, using a chat-based interface. Although it exhibited commendable conversational capabilities, it also possessed certain drawbacks, including the generation of inaccurate or illogical responses. (4) ChatGPT API (2021): The ChatGPT API, which was introduced by OpenAI, enables programmers to incorporate ChatGPT into their services and applications. This resulted in increased accessibility of the model for a multitude of use cases, encompassing customer support, virtual assistants, and more. (5) ChatGPT Improvements: OpenAI continued to receive user feedback and made several updates to improve the performance and safety of ChatGPT. Iterative refinements addressed some of the model's limitations, including instances where it might produce incorrect or biased responses.
Moreover, the development of ChatGPT and its predecessors is part of OpenAI's ongoing efforts to advance natural language processing and AI capabilities [21]. The models are trained on large datasets, and their deployment involves considerations for ethical use, safety, and minimizing biases. The history of these models reflects a progression toward more powerful and versatile language models, with an emphasis on responsible AI development.
2.2 Deep learning approaches
Convolutional Neural Networks (CNNs) One approach in the development of GPT Chat and AI is to use CNNs, which is one of the methods in deep learning. CNNs have proven successful in a variety of NLP and language modeling tasks. CNNs are used to extract features from text data, both in the form of raw text and in the form of word vectors (word embeddings). The CNN architecture consists of convolution layers that apply filters to the input text to find relevant patterns and features. The convolution results of this filter are converted to a non-linear format using an activation function that follows a fixed distance, such as the Rectified Linear Unit (ReLU).
In addition to the convolution layer, CNNs also include a pooling layer which functions to reduce the dimensions of the data generated by the convolution layer. This helps in reducing the complexity and size of the data representation, thereby reducing the number of parameters that need to be trained. Furthermore, fully connected layers are used to link the merged results into the appropriate class or label. This layer performs classification and provides predictive output based on features extracted from text data. A deep learning approach using CNNs on GPT Chat and AI allows models to automatically learn relevant and complex features from input text. Thus, the model can understand the context of the conversation, recognize certain patterns, and generate better and more relevant responses. The application of CNNs in GPT Chat and AI is constantly evolving with the use of more sophisticated architectures and better techniques in natural language processing. This helps improve the interaction quality and responsiveness of the model, thereby providing a better user experience.
2.3 Local Information data collection
In ChatGPT, the dataset used for generating responses to user queries typically consists of pairs of inputs and corresponding outputs. Figure 2 describes some simple examples of dataset entries commonly used in ChatGPT [22, 23].
Figure 2. ChatGPT input and output example
However, ChatGPT can't access real-time data or provide information about specific, current local events, businesses, or details unless it has been explicitly mentioned in the training data available up to its last update. When faced with inquiries regarding local information, it is advisable to consult reputable local sources or utilize specialized local information services to obtain the most current and precise details [24].
We use a diverse approach that incorporates both qualitative and quantitative research approaches to accomplish the goals that we have set for our research activities. To begin, we carried out an exhaustive analysis of the current literature on ChatGPT and its applications in AI, during which we synthesized the most important findings and identified any knowledge gaps that were present. After that, we make use of empirical approaches such as case studies, experiments, and surveys to collect primary data on the utilization of ChatGPT and the perceptions of its users within the communities that are relevant to our research. In addition, we conduct qualitative analyses, such as content analysis and thematic coding, to shed light on the prominent themes, trends, and issues that are associated with the adoption and implementation of ChatGPT. The purpose of this mixed-method approach is to triangulate the findings and provide a nuanced knowledge of the opportunities and restrictions associated with using ChatGPT in AI research.
Overall, our methodology is designed to offer a comprehensive exploration of the role of ChatGPT in AI topics, drawing upon diverse sources of evidence to inform our analysis and recommendations.
3.1 ChatGPT applications, benefits, and risks
Some examples of the Applications of ChatGPT: (1) Customer Support: Implementing ChatGPT in customer support chatbots allows businesses to handle customer queries, provide information, and assist with common issues, improving customer service efficiency [25, 26]. (2) ChatGPT can function as the conversational interface for virtual assistants, facilitating information provision, task assistance, query resolution, and reminder establishment. (3) Content Creation: Writers and content creators can use ChatGPT for brainstorming ideas, generating creative content, or overcoming writer's block by receiving suggestions and prompts. (4) Programming Assistance: Developers can leverage ChatGPT for coding-related assistance, including generating code snippets, helping with debugging, and offering explanations of coding concepts [27]. (5) Language Translation: ChatGPT can assist in language translation tasks by generating translations or helping users understand phrases in different languages [28]. (6) Educational Tools: ChatGPT can be integrated into educational platforms, offering personalized tutoring, answering student questions, and providing explanations for various subjects [29]. (7) Idea Generation: Entrepreneurs and innovators can use ChatGPT to generate ideas, explore possibilities, and discuss potential projects or innovations. (8) Interactive Entertainment: ChatGPT can be employed to create interactive storytelling experiences, chat-based games, or other forms of engaging and immersive content.
ChatGPT has several advantages as shown in Figure 3: (1) Natural Language Interaction: ChatGPT provides a more natural and conversational interaction, making it user-friendly and accessible for a wide range of applications. (2) Versatility: The model's generative capabilities make it versatile across different domains, allowing it to perform various tasks without the need for task-specific training. (3) Ease of Integration: OpenAI provides an API for ChatGPT, making it easy for developers to integrate the model into their applications and services. (4) Accessibility: ChatGPT can be a valuable tool for individuals with varying levels of technical expertise, helping, and information in a conversational manner [30].
Figure 3. ChatGPT advantages
Risks and Considerations of ChatGPT: (1) Biases: The model has the possibility of mirroring biases that exist in the data it was trained on, which could result in biased or unsuitable replies. Attempts are made to alleviate prejudices, but, they may persist. (2) Accuracy: ChatGPT generates responses based on patterns learned during training, and it may produce inaccurate or nonsensical answers. Users should be cautious and verify critical information. (3) Lack of Understanding: The model might not fully understand context or possess true comprehension. It relies on statistical patterns and may generate responses that sound plausible but are not accurate. (4) Ethical Use: To prevent the misuse of technology for malicious objectives, it must be utilized responsibly. This includes avoiding the generation of damaging content, the dissemination of misinformation, and engagement in improper activities. (5) Data Privacy: When integrating ChatGPT into applications, developers need to consider data privacy concerns [31, 32].
It's crucial to be aware of both the benefits and risks associated with ChatGPT and to use the technology responsibly, considering ethical guidelines and best practices in its implementation. OpenAI also actively seeks user feedback to address issues and improve the model over time.
The long-term impact of ChatGPT's use in various applications on the job market and human skill sets is multifaceted. While there may be challenges related to job displacement and shifting skill requirements, there are also opportunities for individuals to adapt, upskill, and thrive in an AI-driven economy. By investing in education, training, and reskilling initiatives, societies can better prepare individuals for the opportunities and challenges that lie ahead in the AI era [33].
The initial step in the development of GPT Chat was to identify the user needs and goals of the system. Will this system be used to assist in the completion of a specific task, provide information, or simply entertain users? This requirements analysis helps in determining the required features and functionality. To train the GPT Chat model, training data is required. Data can be collected from various sources, such as human conversation, text dialogue, discussion forums, or other sources of text data. The data should exhibit diversity and encompass a broad spectrum of subjects and situations to enhance the model's capacity to comprehend and address user inquiries. Once the data has been collected, the next step is to carry out data pre-processing. This procedure involves multiple steps, such as data cleansing to eliminate unnecessary characters, deduplication to remove duplicates, tokenization to split down the text into smaller units, and distraction removal to eliminate unnecessary punctuation [34].
Data pre-processing aims to prepare clean and structured training data for the model. Once data pre-processing is complete, GPT Chat models can be trained using a deep learning approach. This involves using a neural network architecture which in the case of GPT is usually a Transformer. The model is trained by providing the training data prepared in the previous step and optimizing the parameters based on the specified goals. This training process can take significant time and computational resources. After the model has been trained, it is important to carry out validation and evaluation to ensure good quality and performance [35].
This involves dividing the data into training sets and validation sets. The validation set is used to test the trained model and measure evaluation metrics such as accuracy, precision, recall, or other metrics as needed. If the model does not produce sufficient results, the previous steps must be repeated, including data processing and model adjustment. Based on the evaluation results, this step involves improving the model by adjusting parameters or architecture, expanding the training dataset, or using regularization techniques to improve performance. This process is repeated until the model reaches the desired quality level. Once the GPT Chat model has been trained and tested, the next step is to implement the system. This involves integrating the model with the relevant chat platform or application, connecting the system with the necessary data resources, and managing the flow of interaction with the user [36].
Figure 4. ChatGPT features
During this stage, the system can also be adjusted by monitoring and gathering user feedback for continuous improvement. After implementation, it is important to perform continuous maintenance and upgrades on the system. This involves monitoring system performance, troubleshooting any issues or bugs that arise, and gathering additional data to enrich the model. Continuous improvement and improvement can be done by repeating some of the previous steps, such as retraining the model with an updated dataset or updating the model architecture. Figure 4 describes the ChatGPT features.
The introduction of ChatGPT and other AI-driven chatbot technologies has brought about a revolution in human-machine interactions. Language models like GPT-3 have undergone rigorous training using vast amounts of textual data. As a result, they can produce coherent and contextually appropriate replies to user inputs. They have been utilized in diverse sectors, including customer service, virtual assistants, and content creation. ChatGPT and AI chatbots offer a significant benefit by efficiently managing large numbers of inquiries and delivering rapid responses, which improves efficiency and scalability [37].
They can assist users in information retrieval, answering frequently asked questions, and even engaging in casual conversations. This significantly reduces the workload for human operators and enables them to focus on more complex tasks. However, AI chatbots do come with their own set of challenges. An important issue to consider is the possibility of producing biased or unsuitable answers. Given that these algorithms acquire knowledge from text data created by humans, they have the potential to inadvertently sustain and reinforce preexisting biases and prejudices that are inherent in the training data. This has the potential to result in outputs that are discriminatory or offensive, highlighting the significance of ethical issues and careful monitoring throughout the training process. Another obstacle concerns the issue of explainability [38].
Continual endeavors are underway to create methods and structures that improve the comprehensibility and clarity of AI models. AI chatbots also raise substantial concerns about privacy and security. Interactions with these technologies frequently entail the exchange of personal or confidential data. Implementing strong security protocols is essential to safeguard user data and thwart any unauthorized intrusion. Like any technological advancements, ChatGPT and other language models may encounter a range of obstacles in the future. Table 1 displays various obstacles and issues that may arise when using ChatGPT. As AI systems like ChatGPT become more integrated into daily life, there is a risk of humans becoming overly reliant on them. Investigate how dependency on AI may impact decision-making processes, problem-solving abilities, and self-reliance among individuals and organizations [39].
With the rapid pace of AI innovation, intellectual property (IP) issues have become increasingly complex. Explore how IP laws and regulations, such as patents, copyrights, and trade secrets, impact AI development and deployment. Investigate the challenges of protecting AI algorithms, datasets, and trained models, as well as the implications of open-source initiatives and collaborative research efforts. Additionally, consider the role of licensing agreements and technology transfer mechanisms in facilitating responsible AI innovation while safeguarding IP rights.
Table 1. Potential challenges of ChatGP
|
No |
Challenge |
Explanation |
|
1 |
Bias and Fairness [2] |
Language models like ChatGPT can inadvertently perpetuate or amplify biases present in their training data. Addressing and mitigating biases to ensure fair and unbiased responses remains a significant challenge. |
|
2 |
Accuracy and Reliability [3] |
Ensuring the accuracy and reliability of information generated by language models is crucial. Efforts are needed to improve fact-checking capabilities and minimize the generation of inaccurate or misleading content. |
|
3 |
Ethical Use [4] |
Preventing the misuse of language models for unethical purposes, such as spreading misinformation, generating harmful content, or engaging in malicious activities, is an ongoing challenge. |
|
4 |
Handling Sensitive Topics [5] |
Language models need to be able to handle sensitive topics with care, providing accurate and respectful responses while avoiding harm or offense. |
|
5 |
Context Understanding [6] |
Improving the models' understanding of context and the ability to maintain coherent and relevant conversations, especially in longer interactions or complex dialogues, is an area for improvement. |
|
6 |
User Feedback Integration [7] |
Effectively incorporating user feedback to address issues, improve performance, and iteratively refine the models is crucial. OpenAI actively seeks user feedback as part of its model development process [40]. |
|
7 |
Data Privacy [8, 9] |
Maintaining and ensuring user privacy when interacting with language models is an ongoing concern. Implementing robust privacy measures is essential as these models become more widely used. |
|
8 |
Real-Time Information Handling [10, 11] |
Providing accurate and up-to-date information in real time is a challenge for language models since they are typically trained on static datasets. Integrating real-time data while maintaining reliability is an area for improvement [41]. |
|
9 |
Customization and Control [12, 13] |
Balancing the customization and control users have over the behavior of language models is essential. Striking the right balance between user preferences and ethical considerations is an ongoing challenge [42]. |
|
10 |
Resource Consumption [14] |
Large language models like ChatGPT can be resource-intensive both in terms of computational power and energy consumption. Finding ways to optimize efficiency while maintaining performance is a concern. Addressing these challenges involves ongoing research, collaboration, and a commitment to responsible AI development. OpenAI and other organizations continue to work on improving the capabilities of language models while addressing ethical considerations and potential risks. User feedback and community involvement are crucial in shaping the development and deployment of these technologies [43]. |
The impact of ChatGPT and AI chatbots on our interaction with technology is indisputable. New possibilities for virtual assistants, information retrieval systems, and customer support have arisen thanks to their capacity to understand and provide responses that resemble human speech. No matter how difficult it is to overcome problems like comprehensibility and bias, AI chatbot systems are always being improved to make them more trustworthy and reliable. As these technologies evolve, it is imperative to prioritize ethical considerations such as bias mitigation and transparency in their development and deployment. Close collaboration between AI researchers, developers, and policymakers is essential to ensure that AI chatbots are designed and implemented in a manner that aligns with societal values and safeguards user interests. Ultimately, the potential benefits of AI chatbots are extensive, ranging from improved customer experiences to increased efficiency in various domains. By capitalizing on the advantages of these technologies while acknowledging their drawbacks, we may utilize the potential of AI to develop human-machine interactions that are more dynamic, customized, and efficient.
Enhancements in the ability of chatbots to accurately comprehend and interpret user inputs, including handling of ambiguity, context, and colloquial language. Improvements in generating human-like responses that are contextually relevant, coherent, and engaging. Advancements in tailoring responses to individual users' preferences, interests, and conversation history, lead to more personalized and interactive interactions. Integration of multiple modalities, such as text, voice, images, and gestures, to enable richer and more expressive communication with users. Incorporation of mechanisms to address ethical concerns, such as bias mitigation, privacy protection, and transparency in AI chatbot interactions.
Future advancements and endeavors in the field of ChatGPT and AI are expected to focus on tackling current obstacles, enhancing capabilities, and investigating novel applications.
This research is supported by Atma Jaya Catholic University of Indonesia, Deakin University Australia, and the Vice-Rector of Research, Innovation, and Entrepreneurship at Satya Wacana Christian University.
[1] Halaweh, M. (2023). ChatGPT in education: Strategies for responsible implementation. Contemporary Educational Technology, 15(2): ep421. https://doi.org/10.30935/cedtech/13036
[2] Tlili, A., Shehata, B., Adarkwah, M.A., Bozkurt, A., Hickey, D.T., Huang, R., Agyemang, B. (2023). What if the devil is my guardian angel: ChatGPT as a case study of using chatbots in education. Smart Learning Environments, 10(1): 15. https://doi.org/10.1186/s40561-023-00237-x
[3] Rahman, M.M., Watanobe, Y. (2023). ChatGPT for education and research: Opportunities, threats, and strategies. Applied Sciences, 13(9): 5783. https://doi.org/10.3390/app13095783
[4] Ray, P.P. (2023). ChatGPT: A comprehensive review on background, applications, key challenges, bias, ethics, limitations and future scope. Internet of Things and Cyber-Physical Systems, 3: 121-154. https://doi.org/10.1016/j.iotcps.2023.04.003
[5] Shoufan, A. (2023). Exploring students' perceptions of ChatGPT: Thematic analysis and follow-up survey. IEEE Access, 11: 38805-38818. https://doi.org/10.1109/ACCESS.2023.3268224
[6] Dewi, C., Chen, R.C., Liu, Y.T., Yu, H. (2021). Various generative adversarial networks model for synthetic prohibitory sign image generation. Applied Sciences, 11(7): 2913. https://doi.org/10.3390/app11072913
[7] Dewi, C., Juli Christanto, H. (2022). Combination of deep cross-stage partial network and spatial pyramid pooling for automatic hand detection. Big Data and Cognitive Computing, 6(3): 85. https://doi.org/10.3390/bdcc6030085
[8] Agapiou, A., Lysandrou, V. (2023). Interacting with the artificial intelligence (AI) language model ChatGPT: A synopsis of Earth observation and remote sensing in archaeology. Heritage, 6(5): 4072-4085., https://doi.org/10.3390/heritage6050214
[9] Huallpa, J.J. (2023). Exploring the ethical considerations of using Chat GPT in university education. Periodicals of Engineering and Natural Sciences, 11(4): 105-115. https://doi.org/10.21533/pen.v11i4.3770
[10] Orrù, G., Piarulli, A., Conversano, C., Gemignani, A. (2023). Human-like problem-solving abilities in large language models using ChatGPT. Frontiers in Artificial Intelligence, 6: 1199350. https://doi.org/10.3389/frai.2023.1199350
[11] Gunawan, J. (2023). Exploring the future of nursing: Insights from the ChatGPT model. Belitung Nursing Journal, 9(1): 1-5. https://doi.org/10.33546/bnj.2551
[12] Febria, J., Dewi, C., Mailoa, E. (2021). Comparison of capacitated vehicle routing problem using initial route and without initial route for pharmaceuticals distribution. In 2021 2nd International Conference on Innovative and Creative Information Technology (ICITech), Salatiga, Indonesia, pp. 94-98. https://doi.org/10.1109/ICITech50181.2021.9590116
[13] Alanzi, T.M. (2023). Impact of ChatGPT on teleconsultants in healthcare: Perceptions of healthcare experts in Saudi Arabia. Journal of Multidisciplinary Healthcare, 2309-2321. https://doi.org/10.2147/JMDH.S419847
[14] Shi, X., Du, X., Song, X. (2023). Research on miniaturization trend of ChatGPT technology model. Journal of Artificial Intelligence and Technology, 3(3): 95-99. https://doi.org/10.37965/jait.2023.0298
[15] Dewi, C., Chen, A.P.S., Christanto, H.J. (2023). Recognizing similar musical instruments with YOLO models. Big Data and Cognitive Computing, 7(2): 94. https://doi.org/10.3390/bdcc7020094
[16] Iwuozor, K.O., Olaniyi, B.O., Anyanwu, V.U., Suleiman, M.A., Omoleye, W.S., Enahoro-Ofagbe, F.E., Katagum, A.A., Moronkola, I.A., Opeyemi, A.M. (2023). The effect of ChatGPT on sugar industry research. Sugar Tech, 25(6): 1278-1284. https://doi.org/10.1007/s12355-023-01300-0
[17] Zhang, T., Qian, L. (2023). ChatGPT related technology and its applications in the medical field. Advanced Ultrasound in Diagnosis & Therapy (AUDT), 7(2): 158-171. https://doi.org/10.37015/AUDT.2023.230028
[18] Carvalho, I., Ivanov, S. (2024). ChatGPT for tourism: Applications, benefits and risks. Tourism Review, 79(2): 290-303. https://doi.org/10.1108/TR-02-2023-0088
[19] Atlas, S. (2023). ChatGPT for higher education and professional development: A ChatGPT for higher education and professional development. A Guide to Conversational AI Guide to Conversational AI Terms of Use, vol. 1. https://digitalcommons.uri.edu/cba_facpubs/548
[20] Dai, G., Fan, J., Dewi, C. (2023). ITF-WPI: Image and text based cross-modal feature fusion model for wolfberry pest recognition. Computers and Electronics in Agriculture, 212: 108129. https://doi.org/10.1016/j.compag.2023.108129
[21] Yang, L., Wang, J. (2023). Factors influencing initial public acceptance of integrating the ChatGPT-type model with government services. Kybernetes. https://doi.org/10.1108/K-06-2023-1011
[22] Kung, T.H., Cheatham, M., Medenilla, A., Sillos, C., De Leon, L., Elepaño, C., Madriaga, M., Aggabao, R., Diaz-Candido, G., Maningo, J., Tseng, V. (2023). Performance of ChatGPT on USMLE: Potential for AI-assisted medical education using large language models. PLoS Digital Health, 2(2): e0000198. https://doi.org/10.1371/journal.pdig.0000198
[23] Dewi, C., Chen, R.C. (2019). Random forest and support vector machine on features selection for regression analysis. International Journal of Innovation, Computing, and Information Control, 15(6): 2027-2037.
[24] Roumeliotis, K.I., Tselikas, N.D. (2023). ChatGPT and open-ai models: A preliminary review. Future Internet, 15(6): 192. https://doi.org/10.3390/fi15060192
[25] Kalla, D., Smith, N. (2023). Study and analysis of chat GPT and its impact on different fields of study. International Journal of Innovative Science and Research Technology, 8(3). https://www.researchgate.net/publication/369539233
[26] Dewi, C., Chen, R.C. (2020). Decision making based on IoT data collection for precision agriculture. Huk M., Maleszka M., Szczerbicki E. (eds) Intelligent Information and Database Systems: Recent Developments. ACIIDS 2019. Studies in Computational Intelligence, 2020, vol. 830, pp. 31-42. https://doi.org/ 10.1007/978-3-030-14132-5_3
[27] Chen, R.C., Zhuang, Y.C., Chen, J.K., Dewi, C. (2022). Deep learning for automatic road marking detection with Yolov5. In 2022 International Conference on Machine Learning and Cybernetics (ICMLC), IEEE, Japan, pp. 170-174. https://doi.org/10.1109/ICMLC56445.2022.9941313
[28] Mijwil, M., Aljanabi, M., Ali, A.H. (2023). ChatGPT: Exploring the role of cybersecurity in the protection of medical information. Mesopotamian Journal of Cybersecurity, 2023: 18-21. https://doi.org/10.58496/MJCS/2023/004
[29] Wenzlaff, K., Spaeth, S. (2022). Smarter than humans? Validating how OpenAI’S ChatGPT model explains crowdfunding, alternative finance and community finance. Validating how OpenAI’s ChatGPT Model Explains Crowdfunding, Alternative Finance and Community Finance. https://doi.org/10.2139/ssrn.4302443
[30] Abdelkader, O.A. (2023). ChatGPT's influence on customer experience in digital marketing: Investigating the moderating roles. Heliyon, 9(8). https://doi.org/10.1016/j.heliyon.2023.e18770
[31] Mattas, P.S. (2023). ChatGPT: A study of AI language processing and its implications. International Journal of Research Publication and Reviews, 4(2): 435-440, https://doi.org/10.55248/gengpi.2023.4218
[32] Corsello, A., Santangelo, A. (2023). May artificial intelligence influence future pediatric research?-The case of ChatGPT. Children, 10(4): 757. https://doi.org/10.3390/children10040757
[33] Mizumoto, A., Eguchi, M. (2023). Exploring the potential of using an AI language model for automated essay scoring. Research Methods in Applied Linguistics, 2(2): 100050. https://doi.org/10.1016/j.rmal.2023.100050
[34] Sevgi, U.T., Erol, G., Doğruel, Y., Sönmez, O.F., Tubbs, R.S., Güngor, A. (2023). The role of an open artificial intelligence platform in modern neurosurgical education: A preliminary study. Neurosurgical Review, 46(1): 86. https://doi.org/10.1007/s10143-023-01998-2
[35] Crawford, J., Cowling, M., Allen, K.A. (2023). Leadership is needed for ethical ChatGPT: Character, assessment, and learning using artificial intelligence (AI). Journal of University Teaching & Learning Practice, 20(3): 02. https://doi.org/10.53761/1.20.3.02
[36] Wardat, Y., Tashtoush, M.A., AlAli, R., Jarrah, A.M. (2023). ChatGPT: A revolutionary tool for teaching and learning mathematics. Eurasia Journal of Mathematics, Science and Technology Education, 19(7): em2286. https://doi.org/10.29333/ejmste/13272
[37] Gill, S.S., Xu, M., Patros, P., Wu, H., Kaur, R., Kaur, K., Fuller, S., Singh, M., Arora, P., Parlikad, A.K., Stankovski, V., Abraham, A., Ghosh, S.A.K., Lutfiyya, H., Kanhere, S.S., Bahsoon, R., Rana, O.R., Dustdar, S., Sakellariou, R., Uhlig, S., Buyya, R. (2024). Transformative effects of ChatGPT on modern education: Emerging Era of AI Chatbots. Internet of Things and Cyber-Physical Systems, 4: 19-23. https://doi.org/10.1016/j.iotcps.2023.06.002
[38] Sullivan, M., Kelly, A., McLaughlan, P. (2023). ChatGPT in higher education: Considerations for academic integrity and student learning. Journal of Applied Learning & Teaching, 6(1): 1-10. https://doi.org/10.37074/jalt.2023.6.1.17
[39] Dewi, C., Chen, R.C., Liu, Y.T., Liu, Y.S., Jiang, L.Q. (2020). Taiwan stop sign recognition with customize anchor. In Proceedings of the 12th International Conference on Computer Modeling and Simulation, pp. 51-55. https://doi.org/10.1145/3408066.3408078
[40] Dewi, C., Chen, R.C., Yu, H. (2020). Weight analysis for various prohibitory sign detection and recognition using deep learning. Multimedia Tools and Applications, 79(43): 32897-32915. https://doi.org/10.1007/s11042-020-09509-x
[41] Farrokhnia, M., Banihashem, S.K., Noroozi, O., Wals, A. (2024). A SWOT analysis of ChatGPT: Implications for educational practice and research. Innovations in Education and Teaching International, 61(3): 460-474. https://doi.org/10.1080/14703297.2023.2195846
[42] Helberger, N., Diakopoulos, N. (2023). ChatGPT and the AI act. Internet Policy Review, 12(1). https://doi.org/10.14763/2023.1.1682
[43] AlAfnan, M.A., Dishari, S., Jovic, M., Lomidze, K. (2023). Chatgpt as an educational tool: Opportunities, challenges, and recommendations for communication, business writing, and composition courses. Journal of Artificial Intelligence and Technology, 3(2): 60-68. https://doi.org/10.37965/jait.2023.0184
