Challenges in implementing ChatGPT and generative artificial intelligence in various business sectors

From customer service to content development, ChatGPT and generative AI can alter corporate processes across industries. Implementing these technologies presents substantial hurdles that firms must overcome to maximize their potential. Generative AI systems need a lot of data to work, which raises privacy and security concerns and makes GDPR compliance difficult. Model biases can perpetuate prejudice or produce inconsistent consumer experiences, affecting brand reputation and trust. These technologies' scalability and adaptation across industries is another issue. Some sectors demand very precise contextual information, yet generative AI models are generally trained on broad, generic datasets, which may limit their efficacy. The significant computational costs and infrastructural requirements of installing these models strain resources, especially for smaller organizations. Companies typically lack the in-house skills to maintain and fine-tune complex AI systems. This chapter analyses these challenges and provides actionable insights to assist firms strategically negotiate the intricacies of using generative AI for sustainable and scalable impact.

Keywords: ChatGPT, Artificial Intelligence, Human, Large Language Model, Challenges, Business

Citation: Patil, D., Rane, N. L., & Rane, J. (2024). Challenges in implementing ChatGPT and generative artificial intelligence in various business sectors. In The Future Impact of ChatGPT on Several Business Sectors (pp. 107-145). Deep Science Publishing. https://doi.org/10.70593/978-81-981367-8-7_3 

3.1 Introduction

Generative artificial intelligence (AI), especially language models like ChatGPT, has transformed many industries. ChatGPT's natural language processing and understanding capabilities are unmatched in customer service, content creation, predictive analytics, and more (George & George, 2023; AlAfnan et al., 2023; Shihab et al., 2023). Automation, ideation, and business decision-support have been achieved with this technology. However, business environments present complex challenges that must be understood and addressed to implement these models effectively (Raj et al., 2023; Arman & Lamiyar, 2023; Chuma & De Oliveira, 2023; Rane et al., 2024a). To stay competitive and innovative, businesses must understand the challenges of integrating such technologies as well as their potential. Data privacy and security are industry priorities. Generative AI systems learn from massive data sets, raising questions about data ownership, protection, and use. GDPR and HIPAA, which strictly regulate how personal data should be handled, make it difficult for finance and healthcare companies to use AI models like ChatGPT. Due to ChatGPT's learning mechanisms, sensitive interactions may be retained, making compliance with generative AI difficult. Strong strategies that prioritize functionality and ethics are needed to balance regulatory requirements with the AI model's capabilities.

Technical and operational issues arise when integrating ChatGPT into workflows (Jarco & Sulkowski, 2023; Haleem et al., 2022; Deike, 2024). These models need a lot of computational power and expertise to train and tune for industry needs. To accurately answer customer questions, a banking AI model would need financial language and process training. Businesses must train and maintain these models, which requires ongoing costs and skilled staff. Integration of these AI systems into legacy infrastructures often requires complex adjustments to ensure compatibility, which increases implementation time and cost. SMEs may lack the resources to scale AI integration, making this issue particularly relevant. These AI models may produce biased and unpredictable results. ChatGPT, like other language models, learns from datasets with biases that, if unaddressed, may lead to prejudiced or inappropriate responses (Nugroho et al., 2023; Diantoro et al., 2024; Chakraborty et al., 2023). In sensitive fields like HR and law, bias can have serious reputational and legal consequences, making bias mitigation a priority. Human language and social nuances make complete neutrality difficult, even with AI model training advances. Businesses must monitor and refine these models to reduce biases, often requiring complex algorithmic adjustments and constant oversight.

Scalability and efficiency are additional challenges when deploying ChatGPT and generative AI in high-volume industries like retail and telecommunications (Javaid et al., 2023; Rane, 2023; Cribben & Zeinali, 2023; Rane et al., 2024b). These models can handle large-scale operations, but they must perform consistently under heavy workloads (Jusman et al., 2023; Kalla et al., 2023; Wu et al., 2023). The ChatGPT model used by a telecommunications provider to handle customer inquiries must respond quickly and accurately even during peak hours. This level of performance requires advanced infrastructure that can support the model's demands, which can be costly and technical. Human-AI collaboration remains difficult, especially in sectors that require human judgment. ChatGPT can generate drafts or make recommendations in healthcare and legal services, but human oversight is needed to ensure accuracy and compliance. Businesses must align AI autonomy and human supervision in workflows that maximize both. Clear roles and procedures are needed to avoid confusion and errors in this collaboration. Staff must also be trained to manage and use AI systems.

Moreover, generative AI implementation discussions have raised ethical and transparency concerns (Yu, 2023; Sharma & Yadav, 2022; Liu et al., 2023). Many users expect to know when they're talking to an AI, especially in customer service. Some companies struggle to be transparent without overwhelming customers or compromising service (Kocoń et al., 2023; Roumeliotis & Tselikas, 2023; Rahman & Watanobe, 2023). Businesses adopting generative AI must establish clear disclosure guidelines, maintain ethical standards, and build user trust. ChatGPT and generative AI are powerful tools for businesses, and overcoming these challenges is crucial for companies looking to reach their full potential. Businesses can lead innovation and service delivery by understanding and addressing AI implementation challenges.

• Contributions of this research include a literature review on the challenges businesses face when implementing ChatGPT and generative AI, including regulatory, operational, ethical, and technical issues.
• Keywords and Co-occurrence Analysis: Finds and maps the most relevant keywords, revealing trends and associations among core AI integration challenges.
• Cluster Analysis: Categorizes key challenges across business sectors to guide future research and development.

3.2 Co-occurrence and cluster analysis of the keywords

Fig. 3.1 shows the co-occurrence and cluster analysis of the keywords in the literature. The visual clusters and keyword associations show how generative AI concepts, including ChatGPT, are related to different domains and applications.

Main nodes and structure

The network diagram centers on ChatGPT and Generative AI as key nodes connecting domains. Three main clusters represent important thematic areas: education (blue), computational linguistics and AI systems (green), and healthcare and human-centric concerns (red). These clusters show where AI and ChatGPT are used and the challenges of integrating them into each sector. The connections between nodes show co-occurrences of terms and related discussions, suggesting how ChatGPT and generative AI affect different domains, where they may face unique challenges, and what considerations are necessary for successful implementation. Despite ChatGPT's widespread impact, sectoral needs, regulatory requirements, ethical considerations, and technological adaptations for each field present different challenges.

Blue Education Cluster

Generative AI and ChatGPT are increasingly used in education, as shown by the blue education cluster. Current AI applications in learning environments focus on students, teachers, academic integrity, assessment, and higher education. Educational institutions face unique challenges, especially in integrity and meaningful learning. ChatGPT could generate responses or complete assignments for students, undermining academic integrity. This concern is closely related to assessment and critical thinking, emphasizing the need for reliable student performance evaluation. E-learning, federated learning, and contrastive learning, which personalize education and improve learning, are also being investigated by teachers and educational institutions using AI. However, ensuring data privacy, adapting curriculum to include AI literacy, and addressing biases in AI-generated content are significant technological and ethical challenges.

Case-studies, information systems, and challenges suggest empirical studies and systematic frameworks to assess ChatGPT's educational efficacy and limitations. Higher education and teachers recommend focusing on student learning, faculty training, and institutional readiness to integrate these technologies. The infrastructure and expertise needed to integrate generative AI into education and the ethical implications of giving students access to powerful AI tools that could disrupt traditional education are the challenges.

Computational Linguistics and AI Systems Cluster (Green)

The green cluster emphasizes language model, computational linguistics, natural language processing, and learning systems, which are essential to generative AI system development and refinement. This cluster is ChatGPT's scientific and technological backbone, focusing on generative AI methodologies that work across domains. Language models like ChatGPT need computational linguistics and NLP research to improve performance and applicability. NLP processes human language so AI can understand, generate, and interact with users. AI must perform classification, semantics, and text processing accurately to be effective in this domain. Refining these models to handle human language nuances, adapt to diverse cultural contexts, and operate without biases is the technical challenge.

Learning systems and reinforcement learning show how ChatGPT and similar models are trained and improved. For AI systems to adapt to changing user and business needs, reinforcement learning iteratively improves AI through feedback. Reinforcement learning requires large datasets and risks reinforcing model biases and errors. Cybersecurity and high quality reflect technical challenges, particularly data integrity and AI output accuracy and reliability. Generative AI is being used in more sensitive areas like customer data management and business operations, raising cybersecurity concerns. AI models are difficult to implement without compromising security and quality, so the industry needs strong data protection and risk assessment mechanisms. GPT-4, Bard, and large language model demonstrate the rapid evolution and competitive landscape of generative AI research, where new models are developed and tested for optimal performance in various applications.

Fig. 3.1 Co-occurrence analysis of the trending keywords

Health and Human Concerns Cluster (Red)

The red cluster emphasizes healthcare, ethics, and human interactions, including human, humans, healthcare, medical education, and ethics technology. This diagram shows the ethical, procedural, and safety issues generative AI, particularly ChatGPT, faces in human-centered domains like healthcare. ChatGPT and generative AI in clinical practice, patient care, and medical education present significant challenges. AI can aid diagnosis and provide patient information, but its content must be accurate and reliable. Diagnostic accuracy, data accuracy, and controlled study emphasize the need for rigorous validation and testing before AI tools are widely used in clinical settings. Mistakes could harm patients, emphasizing the need for careful implementation and oversight.

Ethics technology and critical thinking reflect ongoing concerns about AI's morality in healthcare. AI can make autonomous decisions or help doctors diagnose and treat patients, raising accountability concerns, especially when AI systems make mistakes. Ethics technology requires ethical frameworks to protect patient rights, autonomy, and dignity when using AI in healthcare. Healthcare, unlike other industries, requires consistent, high-quality AI system performance due to life-or-death outcomes, according to reliability and review. The frequent use of procedures, publication, review, and electronic health record highlights AI integration challenges with medical processes and documentation standards. AI systems must interpret and generate healthcare-compatible data and follow industry standards. Controlled studies and reproducibility highlight the need for empirical evidence of AI's efficacy in healthcare, with a focus on rigorous trials to validate AI's role in patient care and treatment planning.

Connections and Overlapping Issues

Despite their thematic differences, the clusters are interconnected. For instance, chatbots and social media bridge the educational, technical, and healthcare sectors, demonstrating ChatGPT's potential for customer engagement, online education, and patient support. This overlap suggests that while sector-specific challenges exist, ethical and reliable information processing are universal. Risk assessment links the technical and human-centric clusters, emphasizing the need to assess ChatGPT deployment risks in sensitive applications. This shows the need for cross-disciplinary solutions that combine technology, ethics, and domain expertise. Decision making is another cluster theme, indicating ChatGPT's role in helping people make decisions in business, healthcare, and education. However, this role requires balancing AI's contributions with human judgment to make ethical and well-informed decisions. Automation, natural language processing, and human describe the ongoing challenge of seamless human-AI collaboration. AI systems struggle with human language and communication, especially in sectors that value empathy, trust, and ethics.

Fig. 3.2 shows how primary challenges affect AI technologies and which sectors are most affected. Understanding these relationships illuminates the challenges companies face when adopting transformative tools. Technical challenges, which are fundamental to AI adoption, include system architecture, algorithmic accuracy, and technology limitations in ChatGPT and generative AI models. Technical challenges affect ChatGPT and generative AI equally, indicating that technical hurdles are significant across both AI types. Because technical challenges set the baseline for all other issues, if AI's foundational systems don't perform optimally or meet benchmarks, they ripple across implementation efforts in any sector, limiting ChatGPT and generative AI's effectiveness and usability. Another major issue in the diagram is data privacy. Since generative AI requires large datasets for training and development, this challenge affects it more than ChatGPT. Healthcare and financial services, which handle sensitive data, find implementation difficult due to data security and privacy compliance. Data privacy issues emphasize the need for strong safeguards to prevent data breaches, unauthorized access, and misuse, which directly impacts both technologies and makes it difficult for healthcare organizations to fully adopt them without concerns about patient and customer data protection.

Fig. 3.2 Challenges in implementing ChatGPT in various business sectors

In the diagram, bias and fairness are major obstacles for AI technologies, but generative AI is more affected. AI can unintentionally reinforce harmful biases in training data, which could cause problems in financial services, which require high accuracy and fairness. Biases that could lead to unfair lending, credit scoring, or customer segmentation are especially problematic in the financial industry, which relies heavily on data-driven insights. To avoid perpetuating inequalities, ChatGPT and generative AI must mitigate bias. Another major issue that impacts ChatGPT and generative AI adoption is scalability. Because they require a lot of computing power and resources to scale, generative AI applications have more scalability issues than ChatGPT. AI systems must be scalable in retail and manufacturing, where data volumes are high and real-time processing is needed for inventory management and supply chain analysis. Scalability is especially important in retail, where slow or no large-scale AI implementation could hinder customer targeting and inventory planning.

ChatGPT and generative AI are affected by interpretability, a flow challenge, but generative AI is more affected. Users' ability to trust, understand, and explain AI models' decision-making is called interpretability. Making sense of AI outputs is crucial for quality control and operational efficiency in manufacturing. AI adoption is hindered by interpretability issues, as manufacturing and financial services need transparency and insight into model inner workings to avoid "black box" decisions. Without clear interpretability, organizations may be hesitant to use AI models, reducing AI's effectiveness and applicability across business use cases. Cost plagues ChatGPT and generative AI, with generative AI suffering more. The high costs of development, implementation, and maintenance can deter small and medium-sized businesses. Due to budget constraints, healthcare and education struggle to implement AI technologies. In healthcare, cost can prevent hospitals and clinics from using AI for patient care, data analysis, or administrative automation. Generative AI models require computational resources and ongoing investments, making it harder for financially constrained sectors to fully integrate these advanced technologies.

Both AI technologies face security risks, especially in sensitive sectors like the public sector. Security risks include data breaches, unauthorized access, and system hacking. Security risks raise concerns about AI system confidentiality and integrity, which hinders AI adoption in the public sector, which handles massive amounts of sensitive and personal data. Governments and public institutions must secure AI implementations to avoid financial and public trust issues. AI integration into public systems is hindered by security risks. Sector-specific adoption barriers highlight the challenges industries face in adopting ChatGPT and generative AI. Healthcare, education, financial services, and manufacturing each have regulatory, operational, and cultural factors that affect AI adoption. Healthcare has strict patient privacy regulations that limit AI use unless it meets them. The ethical and pedagogical effects of AI's role in automating learning and evaluating students present additional challenges for educators. These adoption barriers demonstrate that each industry needs a customized AI implementation strategy to address its needs and limitations.

Another major challenge is legal and regulatory compliance, which emphasizes the importance of industry regulations, data protection laws, and ethical standards across sectors. Due to its tendency to generate novel content, generative AI faces more intellectual property and content moderation issues. In education, AI applications must comply with regulations to meet standards and policies. Global markets with different AI laws and standards exacerbate legal and regulatory issues. Failure to comply can result in legal penalties, fines, and reputational damage, slowing AI adoption in highly regulated industries. AI implementation challenges are summarized in this Sankey diagram. Each major challenge-from technical limitations to legal compliance-flows through ChatGPT and generative AI, highlighting their dual impact. The diagram shows how these challenges affect different business sectors based on their needs, priorities, and constraints. This Sankey diagram shows the complex challenges businesses face when adopting ChatGPT and generative AI, emphasizing the need for tailored solutions, improved data security, bias mitigation, cost optimization, and sector-specific standards. 

3.3 Challenges in implementing ChatGPT and generative artificial intelligence in various business sectors

Table 3.1 shows the challenges in implementing ChatGPT and generative artificial intelligence in various business sectors (Zhong et al., 2023; Gilardi et al., 2023; Shen et al., 2023; Rane et al., 2024c). Since ChatGPT and generative AI have changed operations, customer interactions, and data management, implementing them across industries is difficult (Liu et al., 2023; Yeo et al., 2023; Aydın & Karaarslan, 2023; Rane et al., 2024d). Businesses struggle to integrate these tools due to data privacy, ethics, accuracy, and scalability concerns.

Data Security and Privacy Issues

ChatGPT and generative AI systems are rapidly adopted in finance, healthcare, and customer service, creating complex data privacy and security issues. Data collection, storage, and processing put businesses at risk of cybersecurity attacks, but these systems need lots of it. ChatGPT and other generative AI models use large language models (LLMs) fine-tuned on massive datasets, raising concerns about sensitive data being captured and exposed. Medical data breaches can have serious legal and financial consequences. Companies must invest heavily in data security and anonymization to reduce exposure risks, but GDPR and HIPAA compliance is difficult. Global data protection laws complicate compliance for businesses which must navigate multiple regulations.

Morality, bias

Ethics and bias issues grow in business generative AI. Large, diverse datasets may include human biases when training AI systems like ChatGPT (Zhou et al., 2023; Opara et al., 2023; Singh et al., 2023; Rane et al., 2024e). In human resources, law enforcement, and finance, algorithmic bias can cause discrimination. AI recruitment tools trained on biased data may perpetuate demographic inequalities. AI transparency and accountability raise ethical concerns, especially when AI-generated content or decisions affect customers and employees. Many generative AI models, called "black boxes," are opaque, making AI behavior hard to understand or control for businesses. Recent AI ethics research advises companies to use more explicable AI models. However, balancing generative AI performance and explainability remains difficult, making ethical risk management for businesses difficult.

Erroneous Results, Hallucinations

"AI hallucinations." occur when generative AI is implemented incorrectly. Generative models like ChatGPT may produce plausible but false content. Poor advice or actions can harm customers or stakeholders in high-stakes fields like finance, healthcare, and legal advisory. Recently introduced verification and fact-checking protocols address this. It's hard to verify AI-generated content, especially when the model answers complex questions. Businesses need strict quality control and human oversight to reduce hallucinated output risks. The balance between automation and human intervention limits the safety of generative AI in high-accuracy sectors.

High cost of operations and resources

Developing and maintaining generative AI systems is costly for startups and small businesses (Rane et al., 2024f; Rane & Shirke, 2024; Patil et al., 2024; Rane & Paramesha, 2024). Training and tuning large language models requires expensive hardware and lots of electricity. These systems need more infrastructure for high-volume queries or real-time interactions. These costs can be prohibitive, especially for data storage, security, and compliance. Operating costs make AI, data science, and cybersecurity specialists unaffordable for many businesses. Recent AI industry trends like cloud-based AI services have reduced these barriers, but high costs still plague many companies, especially in regions with poor digital infrastructure. Financial risks and benefits of AI implementation require planning and risk assessment.

Integration with Legacy Systems

Many businesses use legacy systems, making generative AI integration difficult. Banking, insurance, and manufacturing systems may not support advanced AI due to outdated architecture. These systems need expensive IT infrastructure upgrades or complete overhauls to integrate ChatGPT or other AI models, which is time-consuming and complicated. Companies integrating generative AI systems with digital tools and databases face interoperability issues. Integration requires expensive middleware or custom solutions due to API limitations, data silos, and fragmented IT ecosystems. Businesses must choose between maintaining their systems or investing heavily in AI-compatible architectures. The COVID-19 pandemic has accelerated system modernization, but many businesses lack the resources to go digital.

Lack of Skills and Knowledge

Generative AI requires rare skills. Rare machine learning, natural language processing, and data engineering skills are needed for AI model development, fine-tuning, and management. Generative AI solution builders, deployers, and managers are hard to find. Generated AI may be limited in domain-specific industries like healthcare and legal services due to the AI talent shortage. Complex generative AI applications increase demand for skilled AI practitioners, but education and training programs have not. While AI upskilling, tech partnerships, and literacy programs are promising, the gap remains. Companies must compete with tech giants for AI talent, who can pay more and provide more resources.

Compliance and Regulation

Rapid generative AI evolution has outpaced regulatory frameworks, leaving businesses uneasy about compliance. Companies struggle to define AI's uses, data handling, and liability as governments and regulators standardize it. Finance, healthcare, and legal services are especially susceptible to fines, reputational damage, and legal penalties for noncompliance. Multinational corporations struggle to implement consistent AI policies due to country-specific AI regulations. EU laws like the AI Act suggest AI governance is stricter. AI implementation is more complicated and expensive because businesses must update their policies and invest in compliance resources to comply with these new regulations. Businesses must balance innovation with changing legal standards due to regulatory uncertainty.

Trust and Adoption Barriers

Creating generative AI systems requires user and employee trust. Users distrust AI-driven interactions, especially when using automated customer service or buying. Customers may dislike generative AI for its impersonality or privacy concerns. Workers fear generative AI will replace them. Businesses must educate and disclose AI to gain trust. Show AI's value as an augmentation tool, not a replacement, to gain user buy-in. Recent trends like hybrid AI-human customer service models involve complex human interactions to build trust and maximize AI efficiency. Businesses must prove AI improves customer and employee experiences to gain popularity.

Environmental Impact and Sustainability

Generative AI's environmental impact is raised by the energy needed to train and run large language models. AI-implemented businesses have high carbon footprints due to computational energy consumption. This challenge is especially relevant for manufacturing and retail companies with carbon-reduction CSR initiatives. As sustainability becomes more important, companies must reduce their AI operations' environmental impact. AI researchers have created energy-efficient model architectures and renewable energy data centers to address these issues. Generative AI's environmental and business benefits are hard to manage. Sustainable data management and green technologies are needed to deploy an eco-friendly AI model as businesses become more environmentally conscious.

Federated Learning, Privacy-Preserving AI

AI research is focused on privacy-preserving methods that let models learn from decentralized data sources without accessing sensitive data. Federated learning grows. It localizes data on the source device and trains AI models on multiple devices or servers. This method works in healthcare and finance, where sensitive data is heavily regulated. Training models on distributed devices improves data privacy and security while using AI insights. Federated learning often uses differential privacy techniques to obscure data points with noise. This approach helps businesses implement GDPR and CCPA-compliant AI, reducing privacy risks.

Model Interpretability/XAI

AI decisions must be understood in highly regulated fields like finance, healthcare, and law, so XAI research is essential. However, "black box" deep learning models like ChatGPT can produce complex outputs. Researchers are developing frameworks to make AI decisions more transparent for businesses to track. SHAP and LIME explain model behavior to increase transparency. Explainability builds user trust by ensuring AI-driven customer service decisions follow ethics and company policies. Increasing XAI research should boost user trust and address industry ethics.

Reducing AI Hallucinations and Improving Accuracy

Research focuses on "AI hallucinations," when generative AI models produce inaccurate or fabricated responses. Recent advances have improved generative AI factual accuracy. Research includes retrieval-based methods that generate responses from current, verified data sources like databases or knowledge graphs. To improve AI responses based on human judgment and reduce errors, reinforcement learning from human feedback (RLHF) has been extensively studied. These accuracy-boosting methods can prevent costly or harmful misinformation in precision-intensive fields like legal and medical advice.

Research is also looking at prompt engineering to improve output accuracy. In customer service AI, prompt engineering improves model prompts to match company policies and facts. This developing field may reduce practical hallucinations.

AI bias mitigation and ethics

Ethical AI research creates fairer models to reduce output bias. More model developers use bias detection tools and bias auditing frameworks to ensure ethical and non-discriminatory AI outputs. Transfer learning with diverse datasets is popular for generative model bias reduction. Adversarial debiasing, which reduces model output discrimination through adversarial training, is also widely studied. Industry partnerships like the Partnership on AI and IEEE's Ethically Aligned Design standards shape ethical AI research and applications. To ensure AI system fairness and accountability, companies prioritize these ethical frameworks. AI ethics and bias mitigation standards may become responsible AI deployment standards as regulatory oversight increases.

Green AI, Effective Model Design

Generational AI models require a lot of computation, driving energy-efficient model architecture research. Green AI promotes reducing the carbon footprint of modeling and deploying large AI systems. Researchers are pruning, distilling, and quantizing AI models to make them lighter and less resource-intensive. Pruning removes unnecessary model parameters, speeding it up without affecting accuracy. Also relevant is the trend toward "small data" techniques, which help AI models perform well with limited data. For companies without large AI training datasets, small data methods can make generative AI implementation more sustainable. Data-efficient transformers and synthetic data generation are being studied to increase data availability without model training's environmental impact. Green businesses like these methods because they lower operational costs and support corporate sustainability.

HiTL systems

Human-in-the-loop (HITL) systems combine human oversight with AI decision-making to produce accurate generative AI outputs. AI should complement human expertise in sensitive fields like healthcare, finance, and customer service, so this hybrid approach is essential. HITL system research is investigating human-AI collaboration workflows to improve efficiency and reduce errors. Interactive feedback loops let humans improve AI outputs in real time with expert judgment. Recent HITL advances in active learning allow models to identify cases requiring human input, improving AI output accuracy. Businesses can strategically involve humans to balance AI automation with human expertise to address job displacement and ensure high-stakes application quality. HITL research will increase as more sectors responsibly implement AI to ensure human judgment is essential to decision-making.

Model Robustness and Adversarial Testing

Model robustness, especially against adversarial attacks, is another hot topic. Adversarial inputs can mislead generative AI models. AI-driven decision-making is protected by adversarial robustness in cybersecurity, finance, and legal services. Adversarial training challenges models to strengthen model manipulation resilience. As AI systems are integrated into security-sensitive applications, vulnerabilities can have serious financial and reputational consequences, making this research crucial. Fig. 3.3 shows the challenges in implementing ChatGPT in several business sectors.

Transparency and Ethics in AI Governance

Governance frameworks for AI implementation are prioritized due to regional AI regulatory uncertainty. AI governance research sets policies, standards, and accountability to ensure ethical AI use across sectors. For instance, "algorithmic auditing" is being developed to assess AI output ethics. Impact assessments should evaluate how AI implementations affect employees, customers, and society to help businesses address ethical issues. Organizations also appoint AI stewards to ensure ethical and regulatory compliance. Governance framework research is essential for businesses facing changing regulations. AI Act regulators like the EU set precedents. As more countries consider AI legislation, governance research will be needed to align AI practices with legal, ethical, and social standards.

Domain-Specific Customization

Domain-specific generative models for healthcare, finance, and manufacturing are recent AI advances. Domain-specific models, unlike ChatGPT, can be tailored to sector-specific standards, protocols, and terminology, reducing errors and improving usability. Healthcare models are being developed with medical guidelines and research to improve consultations and diagnosis. Specialized model research is growing as businesses realize generalized AI is unsuitable for sector-specific applications. Domain adaptation techniques modify models on industry-relevant data to improve AI integration in businesses.

Table 3.1 Challenges in implementing ChatGPT and generative artificial intelligence in various business sectors.

3.4 Sector-Specific Challenges in implementing ChatGPT and generative artificial intelligence

Fig. 3.4 shows the challenges in implementing ChatGPT in business.  ChatGPT and generative AI are transforming industry operations, customer engagement, and decision-making. Each sector faces different technology integration, scaling, and optimization challenges. To adopt AI-driven solutions, businesses must overcome operational, regulatory, and cultural issues.

Complex Regulations and Data Privacy in Healthcare

Integrating ChatGPT and generative AI into clinical workflows, patient care, and administration is difficult for healthcare providers. US HIPAA and European GDPR require careful data handling, making patient data privacy a priority. Because generative AI models need large datasets to provide accurate insights, healthcare providers must balance patient data for AI training with data privacy laws. Health professionals must also ensure that AI-generated insights follow clinical best practices and do not introduce bias that could affect patient outcomes. Generative AI may favor certain patient demographics, resulting in biased treatment recommendations that harm patient trust and care quality.

Financial Services: Compliance, Bias, Data Security

Financial institutions use ChatGPT and generative AI models for fraud detection, risk assessment, and personalized banking. These applications have unique algorithmic bias mitigation and financial regulation compliance issues. AI algorithms may discriminate in lending and insurance based on historical financial data. Minority credit scoring can be hampered by AI model bias, complicating financial inclusion. The EU's Digital Operational Resilience Act (DORA) mandates AI transparency and accountability. Generative AI tools exchange a lot of data, increasing cyber-attack risk for financial institutions. Integrating AI-driven solutions while maintaining customer trust and regulatory compliance is difficult.

Personalization and Data Privacy in Retail and E-commerce

ChatGPT and generative AI are changing retail and e-commerce customer service, recommendation, and inventory management. These apps provide personalized customer experiences, but data collection and management are difficult. AI-driven personalization systems require massive amounts of customer data, which raises privacy concerns as data protection laws tighten worldwide. Companies' data collection to personalize shopping experiences may make customers uncomfortable, eroding trust. Generative AI solutions in retail require robust infrastructure, which can be costly for smaller companies competing with industry giants. Retailers must reduce product recommendation and customer interaction misinformation to improve customer satisfaction and brand reputation.

Fig. 3.3 Challenges in implementing ChatGPT in several business sectors

Manufacturing: Integration and Skill Gaps

Generative AI optimizes production, predictive maintenance, and quality control, revolutionizing manufacturing. AI integration is difficult due to legacy systems and a lack of AI-savvy workers. Many manufacturers' decades-old systems are incompatible with modern AI solutions, making generative AI integration into production lines difficult. The skills gap makes this problem worse because most manufacturing workers lack the technical knowledge to use AI. Thus, lengthy training programs take time and money. ChatGPT-based predictive maintenance or automated quality control may also cause unexpected errors and operational disruptions. Manufacturers may struggle to gain stakeholder trust and delay adoption with AI-based solutions due to transparency issues.

Legal Sector: Ethics and Accuracy

Generative AI and ChatGPT could simplify legal document analysis, research, and contract drafting. It's ethical and operationally difficult to implement these technologies. Historical data used by generative AI may be inaccurate or biased and influence legal decisions. Legal professionals worry about the ethical implications of using AI in legal contexts because AI-driven models may ignore legal principles. In the risk-averse legal industry, AI-generated content errors could damage reputation, client trust, and legal action. Artificial intelligence must be calibrated to complement human expertise in legal processes. Legal integrity and client safety depend on this balance.

Quality and adaptability in education

ChatGPT and generative AI personalize learning, grade, and generate content in education. AI-generated educational content quality is difficult to ensure. Teachers and administrators worry about generative AI models producing misinformation or culturally insensitive content. Customizing AI technologies for different learning styles and environments is also necessary. Universal AI solutions are difficult due to education system diversity. Many AI-driven tools collect student data, raising security and parental consent concerns. AI-driven tests may reinforce stereotypes and biases, affecting student performance.

Infrastructure and Sustainability in Energy

Energy uses generative AI for predictive maintenance, energy management, and climate modeling. However, infrastructure and sustainability issues hinder AI implementation in this field. AI adoption is hindered by the lack of advanced technology in energy infrastructures, especially in developing regions. Running large-scale generative AI models requires a lot of energy, which contradicts sector sustainability goals. Energy companies must balance technological advances with eco-friendly practises, so AI solutions' environmental impact is debated. Inaccurate AI predictions could harm the environment and economy by inefficient energy distribution or resource allocation. To support sustainable energy strategies, the sector needs energy-efficient AI models and accurate AI predictions.

Bias and Employee Privacy in HR

HR departments improve recruitment, engagement, and performance management with ChatGPT and generative AI. AI in HR raises concerns about bias and privacy. Training AI-driven recruitment tools on past hiring preferences can perpetuate bias. Discrimination, especially against underrepresented groups, may threaten companies' diversity and inclusion goals. HR AI systems collect sensitive employee data, raising privacy and misuse concerns. AI-driven performance evaluations may seem unfair or lack nuance, unsettling employees. HR professionals must ensure AI tools are transparent and ethical to build trust and avoid legal issues.

Creativity and Ethics in Media and Entertainment

ChatGPT and generative AI are transforming media and entertainment content, personalization, and audience engagement. The sector struggles to balance AI-driven creativity and ethics. Generative AI can personalize content but may reduce creativity or encourage plagiarism. Lack of emotional depth and cultural sensitivity may also disappoint audiences in AI-generated content. Generative AI models trained on copyrighted material may violate entertainment IP. Media companies value accuracy, so AI-generated misinformation and fake content pose ethical risks. In this industry, integrating generative AI to complement human creativity while adhering to ethics is difficult.

Public sector transparency, accountability, and ethics

ChatGPT and generative AI improve government services, citizen engagement, and administration. Inexplicable AI-driven decisions make public sector transparency and accountability difficult. Mistrust arises when government agencies struggle to explain decisions due to complex generative AI algorithms. Public sector ethics include bias-free AI to avoid unfair treatment of certain groups, but historical government data can unintentionally reinforce biases, especially in law enforcement and social welfare. Governments must follow data privacy and security laws to maintain public trust. To maintain trust and democracy, public sector AI systems must be ethical, transparent, and inclusive.

Transportation and Logistics: Real-time Data and Infrastructure Compatibility

In logistics and transportation, generative AI optimises route planning, predictive maintenance, and supply chain management. Industry issues include real-time data management and infrastructure compatibility. The transportation sector uses real-time data, but AI models require robust data processing systems that older vehicles and logistics facilities may lack. Highways, public transit, and airport logistics are difficult to integrate AI with, especially in urban areas with expensive and slow infrastructure updates. AI solutions collect and analyze passenger data, raising consent and data collection concerns. To benefit from generative AI, the sector must modernize infrastructure, address data privacy, and integrate AI with real-time operational data.

Agriculture's Environmental and Resource Constraints

Growing numbers of farmers use AI for crop monitoring, pest control, and yield prediction. Integrating ChatGPT and generative AI into farming poses environmental and operational challenges. Agriculture AI models use extensive environmental data to predict, but weather, soil, and crop health affect accuracy. AI technologies and infrastructure are expensive for small-scale farmers, so this sector has limited resources. The use of water and pesticides by AI-driven farming practices may harm the environment, so generative AI in agriculture must address sustainability issues. The agricultural sector needs affordable, scalable AI solutions that consider environmental and operational constraints to reduce resource scarcity and promote sustainable farming.

Real Estate: Market Prediction Accuracy and Privacy

ChatGPT and generative AI improved property valuation, market forecasting, and customer engagement in real estate. AI-driven market predictions are difficult because the real estate market is volatile and influenced by many unpredictable factors, such as economic changes, geopolitical events, and consumer behavior. Prediction errors can cost investors and real estate firms. Generational AI applications in real estate use a lot of customer data, including income, credit, and lifestyle preferences, raising privacy concerns. Buyers and tenants may be uneasy about AI-powered tools that collect personal data without consent. AI applications in real estate must comply with data privacy laws and provide accurate, explainable insights to build consumer trust and market credibility.

Hospitality and Tourism: AI Ethics and Personalization

ChatGPT and generative AI are changing hospitality and tourism guest services, reservation management, and marketing personalization. These apps struggle with guest privacy and personalization. AI-driven hospitality personalization requires extensive customer data to tailor services to individual preferences, which raises privacy concerns, especially in regions with strict data protection laws. AI in hospitality raises ethical concerns because some guests may be uncomfortable with AI tools handling personal requests or making data-driven recommendations. Pandemics and political changes can disrupt AI-based predictions and change tourism demand, which is sensitive to external events. A successful AI integration in hospitality and tourism requires ethical AI use, data protection, and quick AI system adaptation to market changes.

Insurance: Bias Reduction and Regulation ChatGPT and generative AI assess risks, automate claims, and personalize customer interactions. The industry struggles to reduce bias and follow rules. AI algorithms trained on historical insurance data may bias health, car, and life insurance, potentially discriminating against certain demographics. If AI-driven insurance models use past underwriting decisions, bias may reinforce stereotypes. Regulators need AI decision transparency to inform consumers about insurance premiums. Insurance companies must follow data protection laws because they handle personal data. Insurance companies must ensure that AI systems are impartial, transparent, and compliant with privacy laws while maintaining fair pricing and coverage to address these issues.

Construction and Engineering: Unstructured Data and Safety ChatGPT and generative AI aid project management, risk assessment, and design optimization. Construction projects require unstructured data like blueprints, field reports, and environmental impact studies, which AI algorithms cannot handle. Unstructured data prevents generative AI models from providing construction planning and execution insights. Safety is paramount in construction and engineering projects, so AI recommendations must be strict. AI errors could cause design or construction hazards that endanger workers and residents. Generative AI in construction requires quality control, data structuring, and safety standards to ensure project integrity and safety.

Fig. 3.4 Challenges in implementing ChatGPT in business

Telecom data privacy and network reliability

Telecoms use ChatGPT and generative AI for customer support, network optimization, and predictive maintenance. These apps struggle with network reliability and privacy. Telecommunications networks handle high data traffic, so AI-driven optimization systems must be reliable to avoid service disruptions. Analysis and usage prediction of network data AI models use a lot of computational power, straining infrastructure during peak hours. Telecommunications companies handle sensitive customer data, raising privacy and data protection concerns. To maintain customer trust and service, telecom providers must make AI systems resilient, secure, and privacy-compliant.

Pharmaceutical Compliance and IP Risk Management

Pharmaceutical companies use generative AI for drug discovery, clinical trial management, and personalized medicine. Due to strict FDA and EMA drug development and approval guidelines, AI implementation in this industry is difficult. AI models need rigorous validation to make safe, effective, and impartial recommendations. AI-driven drug discovery poses IP risks because pharmaceutical companies struggle to patent AI-discovered compounds. AI models use a lot of data, making IP attribution and protection of AI-generated molecules difficult. AI integration in pharmaceuticals balances innovation, patient safety, and legal protections by addressing regulatory compliance and IP issues.

Conclusions

ChatGPT and generative AI can transform business sectors by improving productivity, customer engagement, and efficiency. Despite the potential benefits, major obstacles prevent industry-wide integration. As organizations use these technologies more, ethical, data privacy, operational alignment, and financial issues must be addressed. Ethics are a major issue when using ChatGPT and generative AI. The widespread availability of generative AI raises concerns about misinformation, bias, and transparency. AI models often inherit biases from their training data, which can unintentionally amplify social inequalities or stereotypes, putting businesses at ethical and reputational risk. If unchecked, these biases can have serious consequences in hiring, finance, and healthcare, where impartiality is crucial. Additionally, generative AI models like ChatGPT generate human-like text that can be hard to distinguish from human-written content. Businesses may use generative AI for automated communications or decision-making, blurring human and machine-generated content and raising accountability concerns. These ethical issues require constant vigilance, open communication, and a commitment to fair and responsible AI systems.

Generative AI faces additional challenges in data privacy and security. ChatGPT and other models need lots of data, including sensitive customer and business data. Data exposure increases cybersecurity risks and data misuse concerns. Businesses using AI-driven insights struggle to comply with data protection laws in a country-specific regulatory environment. GDPR have an impact on finance and healthcare. To protect sensitive data, businesses must establish robust data governance frameworks that meet regulatory requirements. Due to the complexity of these regulations and the technical challenges of data encryption and anonymization, generative AI implementations require a secure and compliant data infrastructure. Integrating ChatGPT and generative AI across business functions requires operational alignment. Technology and culture must change to accommodate AI-driven solutions in traditional operational workflows. Customer service operations that use ChatGPT must adapt workflows to handle automated and human interactions. Due to a lack of understanding of AI's capabilities and limitations, businesses may have unrealistic expectations and inconsistent implementation. Business goals and AI outcomes often misalign, resulting in poor performance and ROI. Operational harmony requires ongoing staff training, clear AI performance metrics, and iterative improvement as technology and business evolve.

Financial constraints complicate generative AI adoption, especially for SMEs. AI model development, maintenance, and scaling are expensive. SMEs often struggle to compete with larger companies due to the high cost of high-quality datasets, computing power, and specialized talent. To stay relevant, generative AI needs ongoing infrastructure and model fine-tuning, which strains finances. Businesses must carefully weigh the potential gains against these costs and develop long-term strategies to maximize return on investment. Because generative AI technology is dynamic, businesses must deal with rapid advancements that may make existing models or infrastructures obsolete. AI technology evolves quickly, and models like ChatGPT are frequently updated to improve capabilities and fix bugs. Businesses may need to invest time and resources to upgrade and adapt their AI systems due to this relentless pace. New models are more complex and may require more computational power and data storage, straining organizational infrastructure. Companies must stay abreast of AI developments to avoid falling behind more agile competitors in adoption. To navigate the complex world of generative AI, regulators, developers, and end-users must collaborate. Organizations can maximize ChatGPT and generative AI for innovation and efficiency by proactively addressing these challenges. As this technology matures, collaboration, regulation, and education will help all sectors reap its benefits, boosting productivity and digital progress.

References

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.

Arman, M., & Lamiyar, U. R. (2023). Exploring the implication of ChatGPT AI for business: Efficiency and challenges. International Journal of Marketing and Digital Creative, 1(2), 64-84.

Aydın, Ö., & Karaarslan, E. (2023). Is ChatGPT leading generative AI? What is beyond expectations?. Academic Platform Journal of Engineering and Smart Systems, 11(3), 118-134.

Chakraborty, U., Roy, S., & Kumar, S. (2023). Rise of Generative AI and ChatGPT: Understand how Generative AI and ChatGPT are transforming and reshaping the business world (English Edition). BPB Publications.

Chuma, E. L., & De Oliveira, G. G. (2023). Generative AI for business decision-making: A case of ChatGPT. Management Science and Business Decisions, 3(1), 5-11.

Cribben, I., & Zeinali, Y. (2023). The benefits and limitations of ChatGPT in business education and research: A focus on management science, operations management and data analytics. Operations Management and Data Analytics (March 29, 2023).

Deike, M. (2024). Evaluating the performance of ChatGPT and Perplexity AI in Business Reference. Journal of Business & Finance Librarianship, 29(2), 125-154.

Diantoro, K., Munthe, E. S., Herwanto, A., Mubarak, R., & Istianingsih, N. (2024). The Role of ChatGPT in Business Information Systems to Support Strategic Decision Making in Medium-Scale Enterprises. Jurnal Minfo Polgan, 13(1), 382-389.

George, A. S., & George, A. H. (2023). A review of ChatGPT AI's impact on several business sectors. Partners universal international innovation journal, 1(1), 9-23.

Gilardi, F., Alizadeh, M., & Kubli, M. (2023). ChatGPT outperforms crowd workers for text-annotation tasks. Proceedings of the National Academy of Sciences, 120(30), e2305016120.

Haleem, A., Javaid, M., & Singh, R. P. (2022). An era of ChatGPT as a significant futuristic support tool: A study on features, abilities, and challenges. BenchCouncil transactions on benchmarks, standards and evaluations, 2(4), 100089.

Jarco, D., & Sulkowski, L. (2023, June). Is ChatGPT better at business consulting than an experienced human analyst? An experimental comparison of solutions to a strategic business problem. In Forum Scientiae Oeconomia (Vol. 11, No. 2, pp. 87-109).

Javaid, M., Haleem, A., & Singh, R. P. (2023). A study on ChatGPT for Industry 4.0: Background, potentials, challenges, and eventualities. Journal of Economy and Technology, 1, 127-143.

Jusman, I. A., Ausat, A. M. A., & Sumarna, A. (2023). Application of chatgpt in business management and strategic decision making. Jurnal Minfo Polgan, 12(2), 1688-1697.

Kalla, D., Smith, N., Samaah, F., & Kuraku, S. (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).

Kocoń, J., Cichecki, I., Kaszyca, O., Kochanek, M., Szydło, D., Baran, J., ... & Kazienko, P. (2023). ChatGPT: Jack of all trades, master of none. Information Fusion, 99, 101861.

Liu, J., Liu, C., Zhou, P., Lv, R., Zhou, K., & Zhang, Y. (2023). Is chatgpt a good recommender? a preliminary study. arXiv preprint arXiv:2304.10149.

Liu, Y., Han, T., Ma, S., Zhang, J., Yang, Y., Tian, J., ... & Ge, B. (2023). Summary of chatgpt-related research and perspective towards the future of large language models. Meta-Radiology, 100017.

Nugroho, S., Sitorus, A. T., Habibi, M., Wihardjo, E., & Iswahyudi, M. S. (2023). The role of ChatGPT in improving the efficiency of business communication in management science. Jurnal Minfo Polgan, 12(1), 1482-1491.

Opara, E., Mfon-Ette Theresa, A., & Aduke, T. C. (2023). ChatGPT for teaching, learning and research: Prospects and challenges. Opara Emmanuel Chinonso, Adalikwu Mfon-Ette Theresa, Tolorunleke Caroline Aduke (2023). ChatGPT for Teaching, Learning and Research: Prospects and Challenges. Glob Acad J Humanit Soc Sci, 5.

Patil, D., Rane, N. L., Rane, J., & Paramesha, M. (2024). Artificial intelligence and generative AI, such as ChatGPT, in transportation: Applications, technologies, challenges, and ethical considerations. In Trustworthy Artificial Intelligence in Industry and Society (pp. 185-232). Deep Science Publishing. https://doi.org/10.70593/978-81-981367-4-9_6

Rahman, M. M., & Watanobe, Y. (2023). ChatGPT for education and research: Opportunities, threats, and strategies. Applied Sciences, 13(9), 5783.

Raj, R., Singh, A., Kumar, V., & Verma, P. (2023). Analyzing the potential benefits and use cases of ChatGPT as a tool for improving the efficiency and effectiveness of business operations. BenchCouncil Transactions on Benchmarks, Standards and Evaluations, 3(3), 100140.

Rane, J., Kaya, O., Mallick, S. K., & Rane, N. L. (2024a). Enhancing customer satisfaction and loyalty in service quality through artificial intelligence, machine learning, internet of things, blockchain, big data, and ChatGPT. In Generative Artificial Intelligence in Agriculture, Education, and Business (pp. 84-141). Deep Science Publishing. https://doi.org/10.70593/978-81-981271-7-4_3

Rane, J., Kaya, O., Mallick, S. K., & Rane, N. L. (2024b). Impact of ChatGPT and similar generative artificial intelligence on several business sectors: Applications, opportunities, challenges, and future prospects. In Generative Artificial Intelligence in Agriculture, Education, and Business (pp. 27-83). Deep Science Publishing. https://doi.org/10.70593/978-81-981271-7-4_2

Rane, J., Kaya, O., Mallick, S. K., & Rane, N. L. (2024c). Influence of digitalization on business and management: A review on artificial intelligence, blockchain, big data analytics, cloud computing, and internet of things. In Generative Artificial Intelligence in Agriculture, Education, and Business (pp. 1-26). Deep Science Publishing. https://doi.org/10.70593/978-81-981271-7-4_1

Rane, J., Mallick, S. K., Kaya, O., & Rane, N. L. (2024d). Artificial intelligence, machine learning, and deep learning in cloud, edge, and quantum computing: A review of trends, challenges, and future directions. In Future Research Opportunities for Artificial Intelligence in Industry 4.0 and 5.0 (pp. 1-38). Deep Science Publishing. https://doi.org/10.70593/978-81-981271-0-5_1

Rane, J., Mallick, S. K., Kaya, O., & Rane, N. L. (2024e). Federated learning for edge artificial intelligence: Enhancing security, robustness, privacy, personalization, and blockchain integration in IoT. In Future Research Opportunities for Artificial Intelligence in Industry 4.0 and 5.0 (pp. 93-135). Deep Science Publishing. https://doi.org/10.70593/978-81-981271-0-5_3

Rane, J., Mallick, S. K., Kaya, O., & Rane, N. L., (2024f). Scalable and adaptive deep learning algorithms for large-scale machine learning systems. In Future Research Opportunities for Artificial Intelligence in Industry 4.0 and 5.0 (pp. 39-92). Deep Science Publishing. https://doi.org/10.70593/978-81-981271-0-5_2

Rane, N. (2023). Role and challenges of ChatGPT and similar generative artificial intelligence in business management. Available at SSRN 4603227.

Rane, N. L., & Paramesha, M. (2024). Explainable Artificial Intelligence (XAI) as a foundation for trustworthy artificial intelligence. In Trustworthy Artificial Intelligence in Industry and Society (pp. 1-27). Deep Science Publishing. https://doi.org/10.70593/978-81-981367-4-9_1

Rane, N. L., & Shirke S. (2024). Digital twin for healthcare, finance, agriculture, retail, manufacturing, energy, and transportation industry 4.0, 5.0, and society 5.0. In Artificial Intelligence and Industry in Society 5.0 (pp. 50-66). Deep Science Publishing. https://doi.org/10.70593/978-81-981271-1-2_3

Roumeliotis, K. I., & Tselikas, N. D. (2023). Chatgpt and open-ai models: A preliminary review. Future Internet, 15(6), 192.

Sharma, S., & Yadav, R. (2022). Chat GPT–A technological remedy or challenge for education system. Global Journal of Enterprise Information System, 14(4), 46-51.

Shen, Y., Heacock, L., Elias, J., Hentel, K. D., Reig, B., Shih, G., & Moy, L. (2023). ChatGPT and other large language models are double-edged swords. Radiology, 307(2), e230163.

Shihab, S. R., Sultana, N., & Samad, A. (2023). Revisiting the use of ChatGPT in business and educational fields: possibilities and challenges. BULLET: Jurnal Multidisiplin Ilmu, 2(3), 534-545.

Singh, S. K., Kumar, S., & Mehra, P. S. (2023, June). Chat gpt & google bard ai: A review. In 2023 International Conference on IoT, Communication and Automation Technology (ICICAT) (pp. 1-6). IEEE.

Wu, T., He, S., Liu, J., Sun, S., Liu, K., Han, Q. L., & Tang, Y. (2023). A brief overview of ChatGPT: The history, status quo and potential future development. IEEE/CAA Journal of Automatica Sinica, 10(5), 1122-1136.

Yeo, Y. H., Samaan, J. S., Ng, W. H., Ting, P. S., Trivedi, H., Vipani, A., ... & Kuo, A. (2023). Assessing the performance of ChatGPT in answering questions regarding cirrhosis and hepatocellular carcinoma. Clinical and molecular hepatology, 29(3), 721.

Yu, H. (2023). Reflection on whether Chat GPT should be banned by academia from the perspective of education and teaching. Frontiers in Psychology, 14, 1181712.

Zhong, Q., Ding, L., Liu, J., Du, B., & Tao, D. (2023). Can chatgpt understand too? a comparative study on chatgpt and fine-tuned bert. arXiv preprint arXiv:2302.10198.

Zhou, J., Ke, P., Qiu, X., Huang, M., & Zhang, J. (2023). ChatGPT: potential, prospects, and limitations. Frontiers of Information Technology & Electronic Engineering, 1-6.