AI spring and its regulation discourse: a bibliometric study of trends in literature
Katja Debelak, Primož Pevcin, Rok Hržica · 2026
Read AI spring and its regulation discourse: a bibliometric study of trends in literature by Katja Debelak, Primož Pevcin, Rok Hržica (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13190172.txt.
Introduction
Katja Debelak, Primož Pevcin, and Rok Hržica map how research on AI rules and accountability has grown since 2018, in a bibliometric study from 2026. The transformer era sped up AI skill fast. At the same time, law and public governance moved to the center of the debate.
The team searched the Web of Science, cross-checked with Scopus, and charted trends with mapping tools. Output jumped after 2022, in step with milestones such as the EU AI Act. European authors sat at the hubs of the teamwork networks. Core themes were legal: law, transparency, governance, accountability, and data protection. Topics such as legitimacy and rights in practice stayed underworked, and the field remains split into fragments.
Catalog record
| CREATOR | Katja Debelak, Primož Pevcin, Rok Hržica |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 90 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
AI spring and its regulation discourse: a bibliometric study of trends in literature Authors: Katja Debelak, Primož Pevcin, Rok Hržica Published: 2026 ABSTRACT IntroductionThis study was prompted by the rapid acceleration of AI capabilities in the trans former era since 2018 and the concurrent regulatory shift that elevated legal accountability and public governance to central policy and research priorities. It contributes by treating 2018–2025 as a distinct governance regime in which transformer-enabled capability scaling and foundation models shifted AI gov ernance debates toward enforceable accountability architectures.MethodsThe study maps the regulation–accountability–public governance nexus as an operational problem: which accountability forums dominate, which regulatory instruments anchor in the field, and which public-administration mechanisms remain underdeveloped. Using multiple queries in the Web of Science Core Collection, validated with Scopus, and analyzed with Bibliometrix and complementary science-mapping techniques, the study examines publication trends, influen tial contributors and outlets, collaboration networks, and citation and thematic structures.ResultsPublication output increases sharply after 2022, aligning with major regulatory milestones such as the EU AI Act. Results show a strong European concentration, with European actors serving as central hubs in collaboration networks and indicate that 2018–2021 publications form a foundational intel lectual core. The field is anchored in legally oriented concepts (law, transparency, governance, accountability, data protection), while themes such as legitimacy, institutional logics, and rights operationalization remain underdeveloped.DiscussionDespite growing interdisciplinarity, thematic fragmentation persists, highlighting the need for stronger integration across legal scholarship, public administration, and tech nical AI research, and providing a focused basis for future research and policy agendas. Introduction Artificial Intelligence (AI) has experienced several ups and downs since its emergence in the 1950s (Nilsson, 2009). These fluctuations have generated several waves of AI; some authors refer to these as seasons (Haenlein and Kaplan, 2019). Currently, we are experiencing a new boom, often labeled the AI Boom or AI Spring (Cruz and Treisman, 2018), initiated by the development of the transformer architecture. The pioneering work introducing this deep learning architecture, known as the transformer, was published in December 2017 by a Google-sponsored group of authors (Vaswani et al., 2017). The advantage of transformers is that they do not have recurrent units, which reduces training time (Hochreiter and Schmidhuber, 1997). Transformers have improved previous machine translation methods (Zhao et al., 2023) and have since been applied in many areas, including natural language processing, reinforcement learning, and audio processing (Chen et al., 2021; Monastirsky et al., 2023). As noted, the introduction of the transformer model initiated the current wave of AI applications. This supports using 2018 as the starting year for new discussions and research related to AI (Singh and Mahmood, 2021; Yadav, 2024). This is further emphasized by the increasing number of results in the Web of Science database, where publications related to AI and transformers have risen steadily since 2018, especially after 2020 (Web of Science, 2025). The AI boom gained international prominence after 2020, corresponding to the age of integration and ethics, when AI’s integration into business, public services, and everyday technology became more profound (Kashefi et al., 2024; Microsoft, 2026). This has led to growing concerns about bias, privacy, and the societal impacts of AI, moving beyond the technological aspects of AI utilization, and resulting in the current hype surrounding AI (Mora-López et al., 2025). Moreover, selecting 2018 as the starting point is important for the uniqueness and value added by the research, as it marks the transition from experimental to widespread deployment of AI, when organizations began integrating AI into core processes and global attention to ethical, regulatory, and societal implications intensified following developments such as GDPR and growing concerns about algorithmic bias. This year also represents the inflection point preceding the post-2020 AI boom, allowing the review to capture both the foundations and acceleration of AI’s modern integration and societal impact. By grounding the timeframe in this socio-technological shift, the study offers a unique perspective that goes beyond technical evolution and situates AI development within the broader context of ethics, governance, and real-world adoption. AI systems increasingly influence public sector decision-making, administrative processes, and regulatory environments. Scholars and policymakers have begun to examine how automated decision systems interact with legal principles, public accountability, and governance mechanisms. This has expanded research across law, public administration, political science, computer science, and ethics, creating a rapidly growing but fragmented body of literature. Issues such as algorithmic accountability, regulatory compliance, liability, transparency obligations, and risk governance have become central to discussions of AI in democratic societies, particularly within the European context, where regulatory initiatives such as the EU AI Act and OECD guidelines are shaping international debates on responsible AI. Despite this growth, the field lacks an evidence-based, reproducible map of who drives the discourse, where it is published, how its intellectual foundations are structured, and which themes are consolidating versus underdeveloped, limiting cumulative theory building and actionable guidance for public governance. Given the rapid growth and interdisciplinary nature of this field, a systematic mapping of the scientific landscape is necessary to understand its evolution, structure, and research gaps. Bibliometric analysis offers a quantitative and reproducible method for examining publication trends, identifying influential authors and institutions, analyzing collaboration networks, and revealing the conceptual and intellectual structures of a research domain (Cobo et al., 2011; Zupic and Čater, 2015; Aria and Cuccurullo, 2017; Donthu et al., 2021). This mapping is particularly timely because the post-2018 transformer era and the regulatory turn have reshaped both the volume and content of scholarship, yet the resulting knowledge structure has not been systematically consolidated. Applying this approach to literature at the intersection of AI, regulation, legal accountability, and public governance enables identification of how scholarly attention has shifted alongside technological and regulatory developments. The objective of this study is to provide a reproducible bibliometric mapping of research on artificial intelligence regulation, legal accountability, and public governance from 2018 to 2025. Specifically, the study (i) quantifies publication growth, (ii) identifies influential authors, institutions, countries, and outlets, (iii) maps collaboration and citation structures, and (iv) diagnoses thematic clusters and gaps. The core contribution is an evidence-based account of how the field is structured in the transformer era, enabling clearer integration across legal scholarship, public administration, and technical AI research, and providing a focused basis for future research and policy agendas. The research questions follow a standard bibliometric approach, moving from mapping the size and dynamics of the field to explaining its structure and implications. RQ1 establishes the growth trajectory of scientific production in the 2018–2025 transformer-era window. RQ2 and RQ3 identify where the discourse is produced and curated by locating influential contributors and primary publication outlets. RQ4 reveals the field’s intellectual foundations and their evolution through citation and co-citation structures. RQ5 assesses diffusion and cross-jurisdictional knowledge accumulation by mapping collaboration patterns. RQ6 identifies thematic consolidation around accountability, transparency, ethical principles, and regulatory frameworks. RQ7 synthesizes these empirical mappings to specify persistent research gaps and translate them into priorities for future scholarship and policy design. Accordingly, this study is guided by the following research questions: RQ1: What are the global trends in scientific production on artificial intelligence regulation and public governance between 2018 and 2025 (van Raan, 2014; Zupic and Čater, 2015; Donthu et al., 2021; Passas, 2024)? This establishes the field’s production trajectory