RESEARCH ARTICLE

 

Generative AI: Concerns, usage, challenges, opportunities and sentiments

 

 

Hossana Twinomurinzi^I; Sibukele Gumbo^II

^ICentre for Applied Data Science, University of Johannesburg, South Africa. Email: hossanat@uj.ac.za (corresponding)
^IIWalter Sisulu University, East London, South Africa. Email: sgumbo@wsu.ac.za

 

 

---

ABSTRACT

Despite widespread adoption and enthusiasm, Generative AI faces persistent technical challenges, prompting the need to understand its broader impact. This study employed bibliometric analysis to examine scholarly discourse and key drivers of Generative AI adoption. The findings reveal the key factor driving adoption as transdisciplinary collaboration regardless of region; the concerns relate to inconsistent accuracy, demanding infrastructure, and ethical and privacy issues. Generative Al's rapid proliferation, spurred by ChatGPT, marked a post-digital era marked by transformative innovations across multiple fields. The scholarly discourse has grown by 3,164%, reflecting its significant influence on research and disciplinary progress. Education, medical, and biomedical sectors lead in adoption, to enhance teaching experiences, diagnostics, personalised medicine, and pandemic responses. However, the digital divide persists, leaving certain demographics and regions unable to access its benefits equally. The absence of robust ethical, legal, and regulatory frameworks further exacerbates these challenges. This study reveals Generative AI's transformative potential and its ability to bridge disciplines despite technical hurdles. The study makes recommendations for education reform to embrace new curricula, for practitioners and policymakers to create flexible workforce options, to proactively address important ethical, legal, and regulatory considerations; and to be intentional about bridging the widening digital divide.
Categories ● Applied computing ~ Computers in other domains ● Computing methodologies ~ Artiflcial Intelligence

Keywords: artificial intelligence, generative AI, ChatGPT, bibliometric, ethics, scoping review, sentiment analysis, transdisciplinarity, post digital theory

---

 

 

1 INTRODUCTION

Machine Learning (ML), Artificial Intelligence (AI), and Natural Language Processing (NLP) have advanced rapidly in recent years, enabling the creation of powerful platforms like ChatGPT and other Generative AI (GenAI) models, which are collectively referred to as Large Language Models (LLMs). While AI embraces a broad spectrum of technologies aimed at simulating human intelligence, GenAI specifically focuses on generating new content such as text, images, or code by learning patterns from existing data (Hagos et al., 2024). GenAI technologies were initially discovered and developed in the mid-2010s but their widespread recognition surged from November 2022, following the launch of ChatGPT by OpenAI. ChatGPT reached over a million users within just five days of its launch, marking a pivotal moment that demonstrated the practical applications of GenAI and accelerated its widespread acceptance. This paper builds on preliminary findings presented at the South African Institute of Computer Scientists and Information Technologists (SAICSIT) 2023 Conference, extending the analysis to provide a more comprehensive exploration of GenAI's transformative potential, challenges, and societal implications (Twinomurinzi & Gumbo, 2023).

The core components of ChatGPT, and by extension other GenAI models, are constructed on three foundational building blocks; the transformer architecture, pre-training process, and fine-tuning process (Iskender, 2023). The transformer architecture utilises a neural network that is able to analyse large datasets at scale to allow for parallel processing of input data. Previously, the transformer model relied on self-attention mechanisms, which enhanced its ability to process sequential data efficiently. This foundational concept was later significantly advanced and expanded upon with the development of the Generative Pre-trained Transformer (GPT) series which used generative pre-training to learn language patterns from extensive text data. GPT-2 advanced the technology with multi-task training, improving the model's ability to generalise across various tasks. GPT-3 increased the model size, improving its fluency and comprehension capabilities (Iskender, 2023). With InstructGPT, the focus shifted towards integrating human feedback into the training process, resulting in more accurate and context-aware responses. The development of ChatGPT and GPT-4, meant GenAI was not only scaling up in size, but also integrating human feedback mechanisms (prompts) and the capacity to process image inputs (Iskender, 2023; OpenAI, 2025; Temsah et al., 2024; Zhong et al., 2024). Building upon these advancements, OpenAI also recently introduced the o1 model, which includes enhanced reasoning capabilities to reduce hallucinations and instances where AI generates incorrect or fabricated information (OpenAI, 2025). This improvement is particularly crucial in fields like healthcare, where accuracy is critical. For example, o1 has demonstrated superior performance in generating coherent and accurate radiology reports, outperforming other evaluated models (Temsah et al., 2024; Zhong et al., 2024).

GenAIs are versatile and, therefore, extend to a wide range of use cases, with about 75% of their value projected across the four key areas of customer operations, marketing and sales, software engineering, and R&D (McKinsey & Company, 2023). In customer operations, GenAI enhances customer experience and agent productivity. In marketing and sales, GenAI revolutionises content creation and personalisation. It facilitates the generation of personalised messages and marketing materials suited to individual customer preferences, behaviours, and interests. GenAI's capability extends to creating advertisements, social media posts, and product descriptions, substantially enhancing the productivity and creativity of marketing functions. In software engineering, GenAI is redefining the coding process by treating computer languages as natural languages. This transformation allows software developers to complete tasks 56% faster (McKinsey & Company, 2023). GenAI's influence in R&D is significant, particularly in

accelerating the development of new drugs and materials. For example, in the life sciences and chemical industries, GenAI foundation models are being used to generate candidate molecules, significantly speeding up the R&D process (McKinsey & Company, 2023). These advancements across different sectors illustrate the extensive impact of GenAI, not just in enhancing current practices but also in paving the way for novel approaches and efficiencies in various sectors. The technology's ability to understand and generate natural language plays a pivotal role in this transformation, marking a significant leap in automation and innovation potential. These examples highlight the versatility and potential of GenAIs to enhance productivity, streamline content creation, and improve the efficiency of text-related tasks in various domains. However, to ensure their responsible and ethical deployment in real-world applications, it is important to carefully consider the ethical implications and technological challenges associated with the use of GenAI. There are, therefore, divergent discourses surrounding the use of GenAI tools such as ChatGPT and similar GenAI. For example, Italy initially banned ChatGPT citing privacy and data infringement concern (McCallum, 2023).

The main objective of this study was to provide a holistic view of the usage, concerns, challenges, and optimisms of GenAI. A bibliometric analysis was done using the Scopus and Web of Science (WoS) databases, and the identified publications were analysed and visualised using Bibliometrix® and VOSViewer® tools (Aria & Cuccurullo, 2017; VOSviewer, 2025). Specifically, the study sought to answer the following research question:

How are prevalent patterns of GenAI usage represented in scholarly discourse, and what key factors are identified as driving the adoption of these technologies?

The study contributes to our understanding of GenAI integration and its societal impact. Firstly, it provides insights into the transdisciplinary dynamics shaping GenAI adoption across diverse sectors, highlighting the complex relationships between technological advancements, ethical considerations, regulatory frameworks, and societal values. Secondly, it offers an analysis of the ethical challenges inherent in GenAI development and deployment, shedding light on issues of bias, fairness, transparency, and accountability, and proposing strategies to navigate these ethical dilemmas responsibly. Thirdly, the study contributes to the discourse on regulatory governance and policy interventions necessary to ensure the responsible and equitable integration of GenAI, emphasising the importance of adaptive regulatory approaches that balance innovation incentives with societal values and individual rights. Lastly, the research underscores the need for inclusive and participatory approaches to GenAI development and deployment, emphasising the importance of engaging diverse stakeholders in transparent and informed discussions about the opportunities, risks, and ethical implications of GenAI.

This study distinguishes itself from other similar bibliometric reviews of GenAI by adopting a broad, cross-disciplinary lens to explore the impacts, going beyond domain-specific focuses like ChatGPT's role in healthcare (Lopes, 2024) or education (Liu et al., 2024). Unlike Khan et al. (2024), which emphasises general trends and geographic analyses, or R. Dwivedi and El-luri (2024), which limits its scope to technical fields and topic modelling, our study employs inductive thematic analysis to uncover richer insights across disciplines. Also, by separating pre- and post-ChatGPT trends and addressing GenAI's role in stimulating new fields and educational reform, this study offers a unique and forward-looking perspective on GenAI's transformative potential.

The remainder of the study is structured as follows: The next section provides an overview of the methodology employed to conduct the study followed by the results and discussion. Finally, the contributions, limitations, conclusions of the study, and proposed research agenda are presented.

 

2 RESEARCH METHODOLOGY

Bibliometric analyses are increasingly popular as a means to identify trends in data, such as the influence of authors, the most cited articles, and top contributing countries (Aria & Cuccurullo, 2017). In this study, we employed the method to explore the scholarly discourse on GenAI, focusing on its usage, challenges, opportunities, concerns and sentiments. The process involved collecting, analysing and visualising data, and then reporting the findings and inferences.

For the data, we conducted a search of Scopus and WoS on 9 November 2023 using the search string: ('Generative AI' OR 'generative artificial intelligence' OR 'AI generation' OR 'GenAI' OR 'ChatGPT' OR 'large language models'). We used WoS and Scopus because they are compatible with bibliometric tools like Bibliometrix®, ensure high-quality data, and currently represent the most practical options. Grey literature was not included to maintain a focus on peer-reviewed, high-quality, and standardised academic sources, which are critical for ensuring rigour and replicability.

We retrieved 3320 articles from Scopus and 1869 articles from WoS. Since we preferred peer-reviewed publications, we narrowed down the list to include only refereed journal articles and conference publications. We also included journal articles in press. Additionally, we limited the search to only those in the English language resulting in 1987 articles on Scopus and 1211 on WoS. After merging the two datasets, removing duplicate articles, and a peer review of each abstract and title, we found 1314 documents with missing abstracts or unrelated to GenAI leaving a total of 1884 documents. Biblioshiny, an online interface for Bibliometrix®, was used to analyse and visually represent the bibliometric data. We also conducted the thematic analysis manually, and sentiment and usage analyses using OpenAI's interface guided by the results of the manual analysis. Figure 1gives a visual representation of the method.

 

3 RESULTS AND DISCUSSION

3.1 Main Information

In the scoping review shown in Table 1 we manually delineated the document analysis into two distinct periods: 2012-2022, specifically 30 November 2022 when ChatGPT was released, and 01 December 2022-2024, to capture the evolving dynamics of GenAI, particularly highlighting the pivotal role of ChatGPT's emergence as a significant turning point (Twinomurinzi & Gumbo, 2023). The first period, 2012-2022, had 56 documents from 34 sources, with a steady growth rate of 37.97%, an average number of 4.39 co-authors per document, and 26.79% in international collaboration.

 

 

The second period, 2022-2024, revealed a sharp increase in research activity, with 1828 documents from 941 sources. The interest and research output grew exponentially by 3,164%, and with the number of authors surging to 6283. The average co-authors per document was 4.01, and a slightly lower rate of international collaboration at 22.98%. This period also saw a notable increase in journal articles, including early access articles, and a substantial rise in conference papers to 516. ChatGPT was definitively an inflection or tipping point for GenAI and AI usage (Twinomurinzi & Gumbo, 2023).

3.2 Annual Scientific Production

The publication trends between 2012-2022, Figure 2(a), indicates a slow and steady increase in publications peaking at 25 publications until November 2022, suggesting a nominal interest in GenAI. In contrast, the period December 2022-2024, in Figure 2(b) reveals an explosive growth from 4 publications in December 2022 to 1808 in 2023. The sharp reduction in 2024 reflects the 16 early access publications in the forthcoming 2024.

3.3 Top 10 Country Specific Production

Between 2012-2022, the scholarly discourse on GenAI was predominantly led by the United States of America (USA), which contributed 137 publications, nearly twice the total output of the nine other leading countries in GenAI research combined, 75 publications (Table 2(a)). China followed with 20, marking a substantial gap from the top article producer, USA. The presence of the Netherlands (9), Germany (8), Greece (7), Ireland (7), and Switzerland (5) reveals Europe's contribution in the first period. Japan (8), Canada (6), and Argentina (5) represented Asia, North America, and South America, respectively, depicting a nearly global diversity in scholarly production. Notably, African countries were absent. In the subsequent period (2022-2024), the USA maintained its lead with 1877 occurrences (Table 2(b)). China also saw a substantial increase to 639, reinforcing its position as a global research producer. In Europe, Germany (348), the United Kingdom (UK) (315), and Italy (299) were in the lead. Australia (286) emerged in the top 10, suggesting its growing role. Asia's expanded role had India (252), South Korea (143), and Singapore (126). Canada (138), showed a slower growth rate compared to other leading countries. Both periods had no representation from Africa in the top 10.

 

 

3.4 Most relevant sources

Between 2012-2022 (Table 3(a)), the sources were centered around Computer Science, particularly in specialist AI conference series. Between 2022-2024 (Table 3(b)), there was a significant shift in the scholarly discourse on GenAI, with the medical and biomedical fields emerging as the new leaders. This transition marks an increased integration of GenAI technologies in healthcare, particularly in diagnostics, personalised medicine, and pandemic response, reflecting the sector's growing reliance on GenAI.

3.5 Most relevant authors

From 2022-2024 (Table 4(b)), there was a notable shift in the leading scholars on GenAI.

While all top 10 contributors in the period 2012-2022 (Table 4(b)) came from the USA, 20222024 saw a significant pivot towards Asia. China particularly emerged as the leading scholarly voice on GenAI. This shift indicates the rising influence of Asian researchers in GenAI.

3.6 Top Affiliations

Between 2012-2024, the affiliations of leading contributors also underwent a notable shift. Initially (Table 5(a)), the USA dominated, reflecting its strong research infrastructure. However, from 2022-2024 (Table 5(b)), a significant diversification occurred, with a broader international representation emerging. This period saw the rise of institutions from the United Arab Emirates (UAE), Singapore, Australia, Poland, and the UK, among others.

 

 

3.7 Country collaboration

There was hardly any international collaboration on GenAI during the first 11-year period (Figure 3(a)), with only one discernible partnership between the USA and Argentina. This contrasts sharply with the surge in global collaboration from 2022 onward (Figure 3(b)), spanning nearly every continent.

Within the past year, GenAI has markedly enhanced global collaboration, establishing itself as a pivotal platform and catalyst across various disciplines. Its broad appeal, as seen in its application across diverse research fields in Section 3.10, shows GenAI's role in fostering transdisciplinary integration. The integration has enabled scholars from diverse fields to pool their expertise and perspectives, thereby enriching the collaborative efforts.

GenAI platforms such as ChatGPT, Google's Bard, and Microsoft's Copilot have been designed with user-friendliness in mind, reducing the need for specialist technical expertise that was once necessary for engaging with advanced AI tools as seen in the period 2012-2022. This approachability has significantly contributed to their widespread adoption, allowing a broader range of users from various backgrounds to utilise these platforms effectively. By lowering the barrier to entry, these GenAI platforms have democratised access to advanced AI capabilities, facilitating a more inclusive environment for users worldwide including those from marginalised communities who often face barriers due to the digital divide. GenAI therefore offers opportunities for greater participation and representation in AI-driven advancements.

Their accessibility also ensures that individuals across different technical proficiencies can participate, further promoting an inclusive environment conducive to collaboration. This ease of access is a key factor driving GenAI's popularity as a collaborative tool.

Moreover, the rise of GenAI mirrors a broader shift in the research and academic culture towards openness and collective problem-solving. This paradigm shift acknowledges the complexity of contemporary challenges, which often demand efforts from more than one discipline, a process seamlessly facilitated by GenAI. The network effects triggered by GenAI's rapidly expanding user base further increase its value. As more scholars, individuals and institutions engage with GenAI platforms, the resultant growth not only enhances GenAI's usefulness but also fosters a virtuous cycle of increased scholarly adoption and collaboration.

3.8 Top 10 cited documents

The top ten papers from both periods (Tables 6 and 7) illustrate a shift from the ethical concerns and technical proficiency in the use of GenAI to a large scale usage of GenAI particularly in the medical and education fields.

3.9 GenAI Usage and Opportunities

A thematic analysis using an open coding approach was conducted for the usage and opportunities, as well as the concerns, between the two periods (Braun & Clarke, 2006). Initially, all the abstracts in the period 2012-2024 were manually analysed to derive codes, and from the codes, overarching themes, and then finally constructing a narrative to describe the themes that emerged. Atlas.ti was used for this process. ChatGPT was then used to compare its thematic analysis with the manual analysis which proved sufficiently similar and revealed a few more codes. ChatGPT GPT-4 version was particularly useful for the second period 2022-2024 with 1828 abstracts.

Before the emergence of ChatGPT (Table 8, GenAI was prominently engaged in the arts and design sector due to its creative capabilities. In education, GenAI was leveraged to enrich learning experiences, frequently through robotic intermediaries. GenAI had already begun integrating into a few other disciplines, illustrating its potential even before ChatGPT.

Post-ChatGPT (Table 9, there has been a substantial surge in the usage and integration of GenAI, marking a new era of technological adoption. The scholarly discourse has evolved past the previous ethical concerns surrounding GenAI, embracing its integration across various disciplines. Education has particularly been at the forefront of using GenAI, significantly enhancing the learning experiences. Similarly, healthcare has expanded its use of GenAI, enriching medical science curricula and improving patient outcomes. GenAI has also facilitated the development of autonomous robots equipped with communicative capabilities.

3.10 GenAI Challenges and Concerns

Prior to ChatGPT (table 10, the concerns about GenAI included inaccuracies and ethical issues stemming from biased training data. These inaccuracies, ethical concerns and societal impacts remain the dominant challenge with GenAI post-ChatGPT (Table 11 despite the widespread usage. This shows that scholars in the GenAI community recognised these challenges early on and raised the alarm.

3.11 Sentiments before and after ChatGPT

From 2012-2022 (Table 12, the sentiment towards GenAI was predominantly positive, reflecting optimism in sectors like healthcare, where GenAI's accuracy in disease identification promised to revolutionise diagnostics. However, concerns about privacy and ethical implications of GenAI in surveillance were noted, emphasising the need for thoughtful regulation. Years 2022-2024 witnessed a huge surge in positive sentiment towards GenAI, particularly for its applications in various fields, signaling its potential to address pressing global challenges. Yet, the rising optimism has not been without reservations, particularly around the impact of GenAI on job automation, highlighting the need for a balanced approach in embracing GenAI advancements.

 

4 DISCUSSION OF FINDINGS

In order to offer a balanced discussion of the key findings, we adopted a post-digital lens mainly because it allows for a critical examination of the seamless integration of digital technologies into daily life, highlighting the intertwined nature of digital and non-digital realms and the cultural, social, and economic implications of this ubiquity (Cramer, 2015).

4.1 More Rapid Technological Evolutions and their Impacts

The rapid integration of ChatGPT and similar GenAI technologies marks a significant evolution in AI capabilities. This evolution is characterised by the sophisticated development of transformer architectures, pre-training processes, and fine-tuning methodologies. These technical advancements have revolutionised sectors like healthcare, education, and software engineering, demonstrating the potential for significant innovations and efficiency enhancements (Y. K. Dwivedi et al., 2023). This technological leap signifies a transformative change in how problem-solving, knowledge creation, and service delivery are approached across different sectors. The leap introduces a new era of enhanced productivity and innovation, where

GenAI's capabilities can be harnessed to tackle complex challenges more effectively. For example, a 2024 McKinsey survey shows that GenAI adoption has more than doubled in the previous five years, and there is an increase in GenAI investment (McKinsey & Company, 2024). For practitioners and policymakers, this new era points to the urgency of adapting to rapidly changing technologies. For educators and healthcare professionals, GenAI offers tools for unprecedented advancements in personalised education and patient care. Ultimately, GenAI impacts society at large, promising to elevate the quality of life through improved services and innovations.

4.2 Transdisciplinary Integration

The post-ChatGPT era has facilitated greater disciplinary collaboration regardless of region, enabling the integration of GenAI into diverse fields. This unprecedented collaboration is driven by the data-rich insights that GenAI platforms provide, which offer a deeper, evidence-based understanding of complex phenomena across domains. These data-driven capabilities empower researchers to not only test existing theories but also extend them into areas previously unexplored, thereby facilitating unique innovations and the creation of new theories and frameworks (Klein-Avraham et al., 2024). For example, researchers in Information Systems (IS) and Computer Science can use GenAI to provide unique opportunities to refine and adapt their core theories to address challenges in other domains, such as healthcare, education, and the arts (Twinomurinzi et al., 2022). This dynamic interaction allows for the tailoring of theoretical constructs to suit the specific needs and contexts of other disciplines (Y. K. Dwivedi et al., 2023).

Moreover, the insights generated through GenAI encourage the emergence of entirely new disciplines that reflect the convergence of traditional fields. For example, the integration of computational methodologies with social sciences or the arts leads to the development of hybrid fields such as computational sociology or AI-driven creative design (Davidson, 2024; Tan & Luhrs, 2024). This transdisciplinary potential not only bridges existing gaps between disciplines but also creates a fertile ground for the development of more holistic approaches to solving complex, global challenges.

4.3 Ethics Reflections

The rapid integration of GenAI has brought ethical considerations to the forefront, particularly concerning privacy, bias, and the equitable distribution of its benefits. These dilemmas challenge existing social and technical frameworks, necessitating careful navigation to ensure fairness and accountability (Xivuri & Twinomurinzi, 2022). For regulators and ethicists, this signals an urgent need to develop guidelines that promote the responsible and transparent use of GenAI (Cherner et al., 2024). For the general public, GenAI represents a dual-edged sword: it offers significant advancements while also posing risks to privacy, equity, and access to opportunity. Balancing these concerns is critical to ensuring that GenAI's transformative potential benefits society as a whole and does not increase the digital divide.

4.4 Legal Implications and Societal Adjustments

The integration of GenAI into everyday life presents new and intricate legal, societal, and regulatory complexities. GenAI technologies challenge the boundaries of existing societal norms and legal frameworks, exposing significant gaps in their ability to address the rapidly evolving nature of emerging technologies. This means that legal experts and policymakers need to learn to innovate within legal structures to safeguard individual and collective rights without limiting technological progress.

The transformative potential of GenAI extends to the creation of entirely new disciplines and the jobs required within them. As these emerging fields take shape, educational institutions will need to rise to the challenge of designing entirely new curricula that equip learners with the skills necessary for the future world of work. This includes new programmes that merge technical expertise with ethical, social, and creative insights, ensuring future-fit skills.

For society, embracing a new world of work means rethinking traditional employment structures and adopting more flexible, adaptive approaches to workforce development. Policymakers should consider more experimental and dynamic methods in crafting policies that anticipate and respond to the fluid realities of a GenAI- and AI-driven economy. This could include policies that support continuous learning, universal access to reskilling opportunities, and equitable participation in the benefits of GenAI. The same applies to the legal domain which needs to become more reflexive and adaptive, developing mechanisms that not only regulate GenAI but also encourage its responsible and equitable deployment without stifling innovation.

The deeper societal issue is what it means to be human in an era shaped by these technologies that supercede human intelligence in many ways, and the role of human values such as creativity, spirituality, empathy, and connection.

4.5 Limitations

The study is limited in its usage of only two academic databases without taking into account the vast non-academic literature, nor other GenAI platforms that have emerged since ChatGPT. The limited scope is partially a result of the tools available to conduct bibliometric analyses.

 

5 CONCLUSIONS

The study investigated the scholarly discourse on GenAI since before ChatGPT and after ChatGPT using a bibliometric review and thematic analysis. We sought to interrogate how prevalent patterns of GenAI usage have manifested in scholarly discourse and the primary drivers behind the adoption of these technologies. The reality of the transition from one million "early adopters" in five days to an "early majority" of 100 million users within two months indicates that ChatGPT ushered an inflection point in society (Altman, 2022).

The key factor that drives the scholarly discourse of GenAI is transdisciplinary collaboration across regions, disciplines and sectors. Transdisciplinarity is about building bridges between disciplines that have different notions of reality, which bridges themselves become a reality of their own with a new set of language to make sense of the bridge (Klein, 2004). GenAI has managed to navigate the challenges of transdisciplinarity by addressing key limitations inherent in traditional, reductionist approaches. The traditional approaches often prioritise rational logic over relational or qualitative understanding, overlooking the relatedness of disciplines and the role of intuition in knowledge creation (Max-Neef, 2005). In contrast, GenAI fosters a more integrated perspective, bridging the gaps caused by the uni-disciplinary focus prevalent in most educational institutions, particularly at the undergraduate level, and the compartmentalisation of funding into narrowly defined tasks and sub-disciplines (Max-Neef, 2005). While current universities guard disciplinary silos and primarily engage in interdisciplinary collaboration, GenAI opens ways for truly transdisciplinary innovation.

While GenAI has democratised access to advanced AI capabilities, the digital divide continues to pose a significant challenge. Certain demographics and regions, especially in Africa, have unequal access to technology, infrastructure, and digital literacy. These disparities risk exacerbating existing inequalities, leaving marginalised communities further behind.

The rapid adoption of GenAI has also brought ethical dilemmas to the forefront, particularly around issues such as privacy, bias, and equitable access. Ethical lapses in GenAI development and deployment could deepen societal mistrust and perpetuate existing inequalities. A proactive approach is needed to embed ethical considerations into the design, training, and deployment of GenAI models. This includes promoting transparency, accountability, and inclusivity at every stage of development.

The absence of robust legal and regulatory frameworks for GenAI presents significant risks. Without clear guidelines, the deployment of GenAI could lead to misuse, data breaches, and violations of individual rights. Legal experts and policymakers must act swiftly to develop adaptive and reflexive regulations that safeguard privacy, ensure accountability, and promote equitable use. Such frameworks should encourage innovation while protecting societal interests, ensuring that GenAI evolves in a manner that benefits all.

This paper makes a contribution to knowledge and offers some implications for academia, researchers, practice and policy. In IS, GenAI provides a basis for extending existing IS theories using data-driven insights. GenAI allows for the creation of hybrid theoretical frameworks that incorporate both technical efficiency and social implications. In Computer Science, GenAI demands a rethinking of traditional frameworks to include human feedback, contextual accuracy, and ethical AI design. The advancements in transformer models and reasoning capabilities prompt theoretical explorations into designing adaptive systems that balance technical innovation with accountability.

Educators face the challenge of preparing learners for a world reshaped by GenAI. This requires the design of new curricula that integrate technical, ethical, and creative dimensions of GenAI, equipping students with the skills to navigate hybrid roles across emerging fields and disciplines. Educational institutions must prioritise disciplinary collaboration, offering programmes that blend technical expertise with insights from the humanities and social sciences. Additionally, there is a pressing need to foster lifelong learning by embedding AI literacy and digital skills training across all levels of education. Such efforts will ensure that students are not only proficient in GenAI and AI technologies but also prepared to address their societal implications.

Practitioners will similarly need to adapt to the evolving demands of a GenAI-driven world by embracing continuous learning and innovation. Workforce development initiatives should focus on reskilling and upskilling to align with the hybrid roles created by GenAI.

Policymakers will be required to navigate the complexities of integrating GenAI into society by developing adaptive and reflexive regulatory frameworks. These regulations should address critical issues such as privacy, bias, accountability, and equitable use, while encouraging an environment that supports innovation and humanity. Closing the digital divide is another pressing priority, requiring investments in infrastructure, digital literacy, and development policies. Policymakers should also adopt experimental approaches, such as sandbox regulations, to allow for dynamic and iterative responses to the challenges posed by GenAI.

 

References

Ali, S., DiPaola, D., Lee, I., Hong, J., & Breazeal, C. (2021). Exploring generative models with middle school students. Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems, 1-13. https://doi.org/10.1145/3411764.3445226

Alkaissi, H., & McFarlane, S. I. (2023). Artificial hallucinations in ChatGPT: Implications in scientific writing. Cureus. https://doi.org/10.7759/cureus.35179

Altman, S. (2022). ChatGPT launched on Wednesday. Today it crossed 1 million users! [Retrieved 15 June 2025]. https://x.com/sama/status/1

Aria, M., & Cuccurullo, C. (2017). bibliometrix: An R-tool for comprehensive science mapping analysis. Journal of Informetrics, 11(4), 959-975. https://doi.org/10.1016/J.JOI.2017.08.007        [ Links ]

Ayers, J. W., Poliak, A., Dredze, M., Leas, E. C., Zhu, Z., Kelley, J. B., Faix, D. J., Goodman, A. M., Longhurst, C. A., Hogarth, M., & Smith, D. M. (2023). Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum. JAMA Internal Medicine, 183(6), 589. https://doi.org/10.1001/jamainternmed.2023.1838        [ Links ]

Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77-101. https://doi.org/10.1191/1478088706QP063OA        [ Links ]

Cascella, M., Montomoli, J., Bellini, V., & Bignami, E. (2023). Evaluating the feasibility of ChatGPT in healthcare: An analysis of multiple clinical and research scenarios. Journal of Medical Systems, 47(1). https://doi.org/10.1007/s10916-023-01925-4        [ Links ]

Cherner, T., Foulger, T. S., & Donnelly, M. (2024). Introducing a generative AI decision tree for higher education: A synthesis of ethical considerations from published frameworks & guidelines. TechTrends, 69(1), 84-99. https://doi.org/10.1007/S11528-024-01023-3        [ Links ]

Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2023). Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innovations in Education and Teaching International, 61(2), 228-239. https://doi.org/10.1080/14703297.2023.2190148        [ Links ]

Cramer, F. (2015). What is 'Post-digital'? In D. M. Berry & M. Dieter (Eds.), Postdigital aesthetics: Art, computation and design (pp. 12-26). London: Palgrave Macmillan UK. https://doi.org/10.1057/9781137437204_2        [ Links ]

Davidson, T. (2024). Start generating: Harnessing generative artificial intelligence for sociological research. Socius, 10. https://doi.org/https://doi.org/10.1177/23780231241259651        [ Links ]

Dwivedi, R., & Elluri, L. (2024). Exploring generative artificial intelligence research: A bibliometric analysis approach. IEEE Access, 12, 119884-119902. https://doi.org/10.1109/ACCESS.2024.3450629        [ Links ]

Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Ahuja, M., Albanna, H., Albashrawi, M. A., Al-Busaidi, A. S., Balakrishnan, J., Barlette, Y., Basu, S., Bose, I., Brooks, L., Buhalis, D., ... Wright, R. (2023). Opinion paper: "So what if ChatGPT wrote it?" Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policy. International Journal of Information Management, 71, 102642. https://doi.org/10.1016/J.IJINFOMGT.2023.102642        [ Links ]

Gilson, A., Safranek, C. W., Huang, T., Socrates, V., Chi, L., Taylor, R. A., & Chartash, D. (2023). How does ChatGPT perform on the United States Medical Licensing Examination (USMLE)? The implications of large language models for medical education and knowledge assessment. JMIR Medical Education, 9, e45312. https://doi.org/10.2196/45312        [ Links ]

Gupta, A., Müller, A. T., Huisman, B. J., Fuchs, J. A., Schneider, P., & Schneider, G. (2017). Generative recurrent networks for de novo drug design. Molecular Informatics, 37(1-2). https://doi.org/10.1002/minf.201700111        [ Links ]

Hagos, D. H., Battle, R., & Rawat, D. B. (2024). Recent advances in generative AI and large language models: Current status, challenges, and perspectives. IEEE Transactions on Artificial Intelligence, 5, 5873-5893. https://doi.org/10.1109/TAI.2024.3444742        [ Links ]

Iskender, A. (2023). Holy or unholy? Interview with Open AI's ChatGPT. European Journal of Tourism Research, 34, 3414. https://doi.org/10.54055/EJTR.V34I.3169        [ Links ]

Khan, N., Khan, Z., Koubaa, A., Khan, M. K., & bin Salleh, R. (2024). Global insights and the impact of generative AI-ChatGPT on multidisciplinary: A systematic review and bibliometric analysis. Connection Science, 36(1), 2353630. https://doi.org/10.1080/09540091.2024.2353630        [ Links ]

Klein, J. T. (2004). Prospects for transdisciplinarity. Futures, 36(4), 515-526. https://doi.org/10.1016/J.FUTURES.2003.10.007        [ Links ]

Klein-Avraham, I., Greussing, E., Taddicken, M., Dabran-Zivan, S., Jonas, E., & Baram-Tsabari, A. (2024). How to make sense of generative AI as a science communication researcher? A conceptual framework in the context of critical engagement with scientific information. Journal of Science Communication, 23(6), A05. https://doi.org/10.22323/2.23060205        [ Links ]

Liu, J., Wang, C., Liu, Z., Gao, M., Xu, Y., Chen, J., & and, Y. C. (2024). A bibliometric analysis of generative AI in education: Current status and development. Asia Pacific Journal of Education, 44(1), 156-175. https://doi.org/10.1080/02188791.2024.2305170        [ Links ]

Lopes, A. A. (2024). Generative artificial intelligence in health professions: A bibliometric descriptive analysis. 18^tth International Technology, Education and Development Conference, INTED2024 Proceedings. https://doi.org/10.21125/INTED.2024.1936

Max-Neef, M. A. (2005). Foundations of transdisciplinarity. Ecological Economics, 53(1), 5-16. https://doi.org/10.1016/J.ECOLECON.2005.01.014        [ Links ]

McCallum, S. (2023). ChatGPT banned in Italy over privacy concerns [Retrieved 12 May 2025]. https://www.bbc.com/news/technology-65139406

McKinsey & Company. (2023). The economic potential of generative AI: The next productivity frontier [Retrieved 12 May 2025]. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier#/

McKinsey & Company. (2024). What is generative AI? [Retrieved 12 May 2025]. https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai

Mirsky, Y., & Lee, W. (2021). The creation and detection of deepfakes: A survey. ACM Computing Surveys, 54(1), 1-41. https://doi.org/10.1145/3425780        [ Links ]

OpenAI. (2025). OpenAI [Retrieved 12 May 2025]. openai.com/about/

Pavlik, J. V. (2023). Collaborating with chatgpt: Considering the implications of generative artificial intelligence for journalism and media education. Journalism & Mass Communication Educator, 78(1), 84-93. https://doi.org/10.1177/10776958221149577        [ Links ]

Rives, A., Meier, J., Sercu, T., Goyal, S., Lin, Z., Liu, J., Guo, D., Ott, M., Zitnick, C. L., Ma, J., & Fergus, R. (2021). Biological structure and function emerge from scaling unsuper-vised learning to 250 million protein sequences. Proceedings ofthe National Academyof Sciences, 118(15), e2016239118. https://doi.org/10.1073/pnas.2016239118        [ Links ]

Rudolph, J., Tan, S., & Tan, S. (2023). ChatGPT: Bullshit spewer or the end of traditional assessments in higher education? Journal of Applied Learning and Teaching, 6(1), 342363. https://doi.org/10.37074/JALT.2023.6.1.9        [ Links ]

Salvagno, M., Taccone, F. S., & Gerli, A. G. (2023). Can artificial intelligence help for scientific writing? Critical Care, 27(1). https://doi.org/10.1186/s13054-023-04380-2        [ Links ]

Suh, M. (, Youngblom, E., Terry, M., & Cai, C. J. (2021). AI as social glue: Uncovering the roles of deep generative AI during social music composition. Proceedings ofthe 2021 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3411764.3445219

Sun, J., Liao, Q. V., Muller, M., Agarwal, M., Houde, S., Talamadupula, K., & Weisz, J. D. (2022). Investigating explainability of generative AI for code through scenario-based design. 27^th International Conference on Intelligent User Interfaces, 212-228. https://doi.org/10.1145/3490099.3511119

Tan, L., & Luhrs, M. (2024). Using generative AI Midjourney to enhance divergent and convergent thinking in an architect's creative design process. The Design Journal, 27(4), 677699. https://doi.org/10.1080/14606925.2024.2353479        [ Links ]

Temsah, M., Jamal, A., Alhasan, K., Temsah, A. A., & Malki, K. H. (2024). OpenAI ol-Preview vs. ChatGPT in healthcare: A new frontier in medical AI reasoning. Cureus, 16(10), e70640. https://doi.org/10.7759/CUREUS.70640        [ Links ]

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). https://doi.org/10.1186/s40561-023-00237-x        [ Links ]

Twinomurinzi, H., & Gumbo, S. (2023). ChatGPT in scholarly discourse: Sentiments and an inflection point. In A. Gerber & M. Coetzee (Eds.), SAICSIT2023 (pp. 258-272, Vol. 1878). Springer. https://doi.org/10.1007/978-3-031-39652-6_17

Twinomurinzi, H., Mhlongo, S., Bwalya, K., Bokaba, T., & Mbeya, S. (2022). Multidisciplinarity in data science curricula. African Conference on Information Systems and Technology, 21. https://digitalcommons.kennesaw.edu/acist/2022/presentations/21        [ Links ]

VOSviewer. (2025). VOSviewer, visualizing sientific landscapes [Retrieved 12 May 2025]. https://www.vosviewer.com/

Weisz, J. D., Muller, M., Houde, S., Richards, J., Ross, S. I., Martinez, F., Agarwal, M., & Tala-madupula, K. (2021). Perfection not required? Human-AI partnerships in code translation. 26^th International Conference on Intelligent User Interfaces, 402-412. https://doi.org/10.1145/3397481.3450656

Williams, R., Park, H. W., & Breazeal, C. (2019). A is for artificial intelligence: The impact of artificial intelligence activities on young children's perceptions of robots. Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, 1-11. https://doi.org/10.1145/3290605.3300677

Wu, Y., Mou, Y., Li, Z., & Xu, K. (2020). Investigating American and Chinese subjects' explicit and implicit perceptions of AI-generated artistic work. Computers in Human Behavior, 104, 106186. https://doi.org/10.1016/j.chb.2019.106186        [ Links ]

Xivuri, K., & Twinomurinzi, H. (2022). A Habermasian approach to fair processes in AI algorithms. Communications in Computer and Information Science, 1551 CCIS, 335-343. https://doi.org/10.1007/978-3-030-95070-5_22

Zhang, Z., Xu, Y., Wang, Y., Yao, B., Ritchie, D., Wu, T., Yu, M., Wang, D., & Li, T. J. (2022). StoryBuddy: A human-AI collaborative chatbot for parent-child interactive storytelling with flexible parental involvement. CHI Conference on Human Factors in Computing Systems, 1-21. https://doi.org/10.1145/3491102.3517479

Zhong, T., Liu, Z., Pan, Y., Zhang, Y., Zhou, Y., Liang, S., Wu, Z., Lyu, Y., Shu, P., Yu, X., Cao, C., Jiang, H., Chen, H., Li, Y., Chen, J., Hu, H., Liu, Y., Zhao, H., Xu, S., ... Liu, T. (2024, September). Evaluation of OpenAI o1: Opportunities and challenges of AGI [Retrieved 12 May 2025]. https://arxiv.org/abs/2409.18486v1

 

 

Received: 27 June 2024
Accepted: 21 January 2025
Online: 2 July 2025