PhD students

Lilla Vicsek has been researching the social and ethical dimensions of AI since 2018. Part of her work examines how visions of the future of work and automation are constructed across multiple arenas—media narratives, expert debates, and lay interpretations of one’s own prospects. She will present two studies within this thematic focus. The first investigates expert debates, showing that these discussions frequently frame technological change as an autonomous force, posing questions such as whether robots will replace human labour. Such technologically deterministic framings obscure the political, economic, and institutional choices through which technologies are shaped, diverting attention from issues of power, inequality, working conditions, and environmental sustainability. Forecasts often rest on problematic assumptions, extrapolate from transient hype cycles, and overlook uncertainty, societal dynamics, and the interests embedded in particular predictions. The second study explores individuals’ expectations about their personal futures in an automated labor market and the cognitive biases that inform these expectations. The findings demonstrate that the psychological patterns are stratified by education and income, suggesting unequal capacities to interpret and prepare for anticipated change. Taken together, the two lines of inquiry show that imagined futures of work are not neutral projections but socially produced visions shaped by institutions, interests, cognitive tendencies, and wider cultural narratives about technology and its role in society.

AI has great potential to revolutionize products, processes, and strategies across individuals, organizations, social entities, and the broader economy. The primary obstacle to this AI revolution is the lack of trustworthiness of AI solutions, mainly due to the absence of Precise, Explainable, and Provable (PEP) AI outcomes. Therefore, our goal is to introduce the PEP AI concept via three groundbreaking methodologies to achieve PEP principles in AI solutions. First, we would like to introduce a new AI computational paradigm, the Neural Mesh replacing current Neural Networks used in AI engines. This leads to greater power and precision in AI. Second, we would like to develop a fundamental methodology to provide mathematically provable evidence for the stability and performance of AI-driven control solutions, replacing unreliable experience-based evaluations that are unacceptable in control systems (e.g., flight control, medical applications). Third, we would like to propose a new research direction aimed at extracting linguistically interpretable and explainable AI decisions to better understand the reasoning behind them. In addition to theoretical research, we  deliver proof-of-concept implementations integrated into healthcare applications while addressing future technology acceptance and EU AI Act aspects to foster post-project innovation and exploitation. We strongly collaborate with the University of Washington (UW), and The Chinese University of Hong Kong (CUHK), ranked 12th and 36th, respectively, by QS in AI and Informatics subjects. The key scientists are Y. Nesterov (CUB), recipient of the World Laureates Association Prize, along with four globally renowned scientists (from CUB,

CUHK, UW) contributing to PEP AI with their inventions, published in numerous top 2% journal papers.

Impact of AI Tools on Learning Outcomes:

Decreasing Knowledge and Over-RelianceStudents at all levels of education are increasingly relying on generative artificial intelligence (AI) tools to complete assignments and achieve higher exam scores. However, it remains unclear how this reliance affects their motivation, their genuine understanding of the material, and the extent to which it substitutes for the process of knowledge acquisition. To investigate the impact of generative AI on learning outcomes, an experiment was conducted at Corvinus University of Budapest. In an operations research class, students were randomly assigned into two groups: one was permitted to use AI tools during classes and examinations, while the other was not. To ensure fairness, a compensation mechanism was introduced: students in the lower-performing group received point adjustments until the average performance of the two groups was equalized. Despite the organizers’ best efforts to explain the design and to create equal opportunities for all participants, many students perceived the experiment as a major disruption. Although the experiment was approved by every relevant university authority — including the Ethics Board, the Head of Department, the Program Director, and the Student Council — students escalated their concerns to the media and eventually to the State Secretary for Higher Education of Hungary. As a result, the experiment had to be substantially revised before completion: on the final exam the test group was merged with the control group. Still, the data allowed us to draw decisive conclusions regarding the students’ learning habits. Uncontrolled use of AI tools leads to disengaged students and low understanding of material. The extreme reactions of the students proved even more revealing than the data collected: generative AI tools have already become indispensable for students, raising fundamental questions about the validity of their learning process.

The article is available: https://arxiv.org/abs/2510.16019

This research examines the relationship between artificial intelligence (AI) and disinformation and seeks to answer whether AI is opening a new era in disinformation. Róbert Pintér have conducted a systematic investigation, introducing disinformation chain as a new concept. Using this theoretical framework, he describes the motivation, creation, dissemination, detection, reception and wider social impact of disinformation. He also describes the application of AI in these steps. The results show the role of AI is particularly significant in the dissemination and social impact of disinformation, while the emergence of generative AI has not proved to be decisive in the creation of disinformation, contrary to prior expectations. If we can talk about a new era, it is more related to social media algorithms and participatory propaganda with the active involvement of people, which has led to the emergence of a post-truth era since 2016. The research also proposes a set of actions for effective measures against disinformation along the disinformation chain. Finally, the researcher concludes by expressing his belief that the statement ‘we cannot believe our eyes and ears’ crystallises not as a fatalistic creed but as a healthy scepticism, since no society can function without human trust.

The article is available in Hungarian: https://m2.mtmt.hu/api/publication/36395201

Enhanced content analysis with the use of artificial intelligence

This research investigates why SMEs often struggle with Industry 4.0 (I4.0) adoption and develops an SME-specific readiness evaluation model. The study uses two AI techniques to analyse expert interview data:

•             Topic modeling using BerTopic

•             LLM-based Knowledge Graph

Together, these AI-enabled steps offer a complex analytical pipeline. This approach enhances the robustness, transparency, and theoretical grounding of the findings, surpassing traditional qualitative content analysis.

The resulting readiness model is therefore both conceptually validated and practically usable; however, the research also acknowledges its limits: such AI-enhanced methodologies depend on existing conceptual frameworks and may not be directly transferable to all domains.

Overall, the study demonstrates how AI can enhance qualitative research, facilitating the extraction of deeper insights, improved validation, and more systematic theory building in the context of I4.0 readiness.

The Library offers both printed and electronic resources, which can be searched using the following three search interfaces:

The Open Public Access Catalogue (OPAC): contains documents available in the library’s physical collection and is suitable for finding specific printed books. You might find some electronic books and links to electronic versions associated with the printed record.

Journals and Ebooks: a search engine that helps you to find full-text electronic articles and e-books available in the library. Start the search with the title of the journal/e-book, even if you are searching for an article or a book chapter. In addition, bibliographic data for printed journals is also available with a link to the OPAC. The search engine includes both subscribed and free sources (e.g. DOAB, DOAJ, JSTOR OA books, etc.). Electronic resources can be accessed remotely using VPN.

Super Search: a comprehensive one-step search engine that allows you to search in the content of our databases containing scholarly literature, as well as printed books from the Library’s physical collection. In addition, two Corvinus repositories (Corvinus Research and PhD Dissertations) and other free, relevant sources are also integrated. Databases containing data (financial, statistical) and legal materials cannot be searched in Super Search; these must be used separately.

Although Super Search is suitable for finding specific works, it is mainly helpful for topic search. The search should start with keywords and be refined with filters and field restrictions. Most of the sources are in English. Supersearch also lists documents that are not available in full text but may be helpful for a given topic. These can be requested through interlibrary loan by clicking the link below the search results and copying the data.

Google Scholar: searches for scientific literature, but also harvests predatory publishers’ journals, which do not have a peer review process before publication. We have compiled the characteristics of predatory publishers and ways to check publications under Predatory publishers.

You can also set Google Scholar to mirror the Corvinus collection (remotely, via VPN) by following the instructions here.

Grey literature: documents such as research reports, working papers, and dissertations belong to this category. Non-publishing units, such as organisations, research institutes, and universities, usually publish them.

Finding them is difficult because there is no platform where all of these can be searched; they lack a uniform description, their formats vary, they are not permanently accessible, etc. Platforms where grey literature can be found include: BASE, OpenAIRE, CORE and various institutional and other repositories. General search engines (Super Search, GS) also search (to some extent) for this content independently or through multiple indexes.

Datasets: To search for datasets, try Google with e.g., “sociological dataset” as the search term. The results list the significant collections (ICPSR, Roper, US Census, CNTS, etc.) and provide university Libguide pages where various datasets have been collected.

Another option is Google Dataset Search, which lets you search for specific datasets. Free datasets are available by clicking on the Free filter in the next step.

SAGE Research Methods: SRM is a research methodology database that provides theoretical background information for various research methods (e.g. social network analysis). Books, encyclopaedias, journals, some podcasts and videos are available in the database. The Little Blue series focuses on qualitative methods, while the Little Green series focuses on quantitative methods.

We do not have access to Cases or Datasets (content with a grey background), but individual researchers can request a 30-day free trial for them.

It is recommended to enable the “Content available to me” filter when searching to return only documents with full text.

In addition to documents, the database offers several research support tools, such as Project Planner and Which Stats Test, which help select statistical methods.

JSTOR is an archive database containing peer-reviewed journals, academic e-books, and primary sources. Our subscription covers the entire JSTOR collection, ensuring representation across all academic fields. Journals are typically available from early volumes (in most cases from the very beginning), but due to the nature of the database, the last 1-5 years are not. The built-in AI tool can be activated with an account created within the database.

WoS and Scopus are bibliographic databases that contain references to major scientific journals across various scientific fields. They support scientometric analysis, map individual career paths, and serve as a starting point for systematic literature search. The full text of the documents is only available if the library has a subscription to the journal.

The overlap between the two databases in the field of social sciences in the 1990-2025 period is approximately 80-90%.

There are some differences between the two databases. WoS collects documents more broadly (conference proceedings and papers, in addition to books and articles), and documents are available from 1900, whereas in Scopus they date back to 1970.

Leading journals in national languages may be omitted from both databases if they do not meet the rigorous evaluation and selection criteria.

The following Corvinus journals are indexed in Scopus and/or WoS: Corvinus Journal of Sociology and Social Policy, Public Finance Quarterly, and Society and Economy.

Supersearch and WoS/Scopus – Which one to use?

Supersearch:

• supports general topic searches
• provides full-text access; if this is not available, it offers interlibrary loan
• not only includes leading journals, but also draws from a broader range of sources

WoS and Scopus:

• helps in mapping references
• supports systematic literature search
• provides abstracts and links to the full text
• contains outstanding, internationally recognised publications in various scientific fields

The differences between general and systematic literature review are summarised in the table below.

A specific example of the difference:

We recommend viewing the following series in English: Systematic & scoping reviews for social sciences & education: Part 1 – Part 5

Each part presents the basics of systematic literature search, search strategy, interpretation of the PRISMA flowchart, etc. It also introduces the subscription-based Covidence software, which facilitates systematic literature review.