As the TWIN4DEM project progresses, Work Package 6 ensures that the project’s scientific advances extend beyond research and deliver practical value for policymakers, civil society organisations and the wider democracy community.

WP6 develops methodological toolboxes, educational resources and evidence-based policy recommendations that make computational social science methods more accessible, transparent and ethically grounded. By combining open tools, training activities and policy guidance, the work package supports the responsible use of digital technologies for democracy research and governance.

In this final interview of our Work Package video series, learn how WP6 is helping transform research results into practical tools, build the next generation of democracy researchers through the TWIN4DEM Training Camp, and provide recommendations that contribute to stronger, fairer and more resilient democratic institutions.

Work Package 5 focuses on designing, implementing, and validating the TWIN4DEM Digital Twin prototypes. Building on the requirements defined in previous work packages and the integrated datasets developed in WP4, WP5 transforms the project’s concepts into fully operational Digital Twin models capable of analysing democratic processes and simulating future policy scenarios.

The main objectives of WP5 are to:

• Design the overall Digital Twin architecture and its interconnected modules
• Develop and implement the individual Digital Twin components
• Integrate all modules into a unified and functional Digital Twin prototype
• Connect the Digital Twin with the data pipelines developed in WP3 and WP4
• Calibrate and validate the models using real-world data
• Simulate policy scenarios co-designed with stakeholders to assess their potential impact

By combining advanced modelling techniques with high-quality data and stakeholder-driven scenarios, WP5 delivers the core technological foundation of TWIN4DEM, enabling the Digital Twin to support evidence-based policy analysis, democratic resilience, and informed decision-making.

As part of the Horizon Europe project TWIN4DEM, which examines the decision-making behaviour of parliamentary representatives, we compiled a database of all MPs serving in the Hungarian Parliament between 2010 and 2026. Using these data, we can explore how “sticky” parliamentary mandates are—that is, to what extent MPs remain in office from one electoral term to the next.

Between 2010 and 2026, there were 983 parliamentary seats across the four electoral cycles. The actual number of individuals serving was somewhat higher because some MPs resigned, died, or, in the most recent cycle, lost their mandates and were replaced. Counting each individual separately in every cycle in which they served, Hungary had a total of 1,046 parliamentary mandates during this period.

In reality, however, these 983 seats (or 1,046 mandates) were occupied by only 546 different individuals. Of these, 102 (19%) served in all four cycles, 56 (10%) served in three cycles, 74 (14%) served in two cycles, and 314 (58%) served in only one of the four cycles. Naturally, this does not mean that these individuals could not have served before 2010 as well. Moreover, many MPs left Parliament after the 2010–2014 cycle, when the number of parliamentary seats was reduced by almost half.

2014 marks a significant change, since the new Fundamental law decreased the number of MPs from 386 to 199. Of the 402 MPs serving between 2010 and 2014, 172 remained in the subsequent Parliament of 199 members. Of the 211 individuals who served during the 2014–2018 cycle, 161 also served in the next cycle. Similarly, of the 215 individuals who served between 2018 and 2022, 147 continued their parliamentary careers after the 2022 election.

This means that 43% of the MPs serving in 2010 remained in Parliament after the 2014 election despite the significant reduction in seats. At the same time, these returning MPs constituted 86% of the 2014 Parliament. Taking into account mid-term replacements, they represented 82% of all individuals who served during the 2014–2018 cycle.

The corresponding figures for later cycles were as follows:

The data therefore point in two directions simultaneously. On the one hand, parliamentary turnover in Hungary is relatively limited: nearly half of all MPs served in all four electoral cycles. On the other hand, turnover has gradually increased over time. The 2014 Parliament was composed overwhelmingly of individuals who had already served in the previous Parliament, but this pattern became less pronounced in subsequent cycles. This may partly reflect the retirement of politicians who had already accumulated substantial parliamentary careers before 2010.

The 2026 election clearly represents a major break in this pattern. None of the 141 TISZA MPs had previously served in Parliament. Among the 52 members of the Fidesz–KDNP parliamentary group, now in opposition, 15 are newcomers, while one of the six members of the Mi Hazánk parliamentary group is entering Parliament for the first time.

Based on previous experience, however, one may assume that parliamentary office remains a relatively “sticky” position because obtaining a parliamentary mandate requires substantial resources. Whether the new term-limit regulation will substantially change this remains uncertain. Even under the new rules, the current cohort of 157 first-time MPs could potentially serve for up to 12 years, either continuously or with interruptions. Moreover, the historical retention rates do not suggest that significantly greater parliamentary renewal is likely from one cycle to the next. It should also be noted that there is no clear consensus regarding what constitutes an optimal level of turnover in parliamentary mandates. There may be advantages to representatives serving multiple terms, particularly when their continued re-election reflects successful representation of their constituents’ interests.

Article prepared by:
Csilla Zsigmond and Márton Gerő
ELTE Centre for Social Sciences

]]> https://twin4dem.eu/keeping-humans-in-the-loop-twin4dem-workshop-explores-ai-auditing-transparency-and-accountability/ Mon, 06 Jul 2026 08:13:58 +0000 https://twin4dem.eu/?p=1271

On 24 June 2026, TWIN4DEM organised an online stakeholder workshop titled “Keeping Humans in the Loop: AI Auditing, Transparency and Accountability in the Age of Hybrid Intelligence.” delivered by project partner Eticas. The workshop was organised in collaboration with Fondazione Bruno Kessler (FBK) and was addressed to the doctoral candidates of the HYBRIDS Marie Skłodowska-Curie Doctoral Network. The session explored how AI auditing can help ensure that AI systems remain transparent, accountable and aligned with democratic values while keeping humans at the centre of AI-enabled decision-making.

The workshop was presented by Alex Magaard, Head of Policy at Eticas, who introduced participants to the principles and practical applications of AI auditing. Drawing on Eticas’ expertise in evaluating AI systems, the session demonstrated how structured auditing methodologies can improve the reliability, fairness and transparency of AI tools throughout their lifecycle.

A key focus of the workshop was understanding where and when AI systems should be audited. Participants explored the different stages of AI development, from data collection and model training to deployment and post-deployment monitoring, highlighting the responsibilities of both AI developers and researchers using AI systems. The discussion emphasised that meaningful auditing begins with understanding how an AI system is built and how humans interact with it.

To support this process, Eticas introduced a six-component socio-technical system mapping framework, covering the data, model, outputs, human-AI interaction, societal impact, and monitoring mechanisms of AI systems. This approach enables researchers to identify potential risks before they become operational issues while ensuring that human oversight remains embedded throughout the system.

The workshop also demonstrated how system mapping can be translated into practical risk identification. Participants examined a comprehensive AI risk taxonomy addressing governance, bias and fairness, privacy, reliability, security, misuse and environmental sustainability. For each category, the session presented strategies for detecting and mitigating risks during the pre-processing, in-processing and post-processing phases of an AI system’s lifecycle.

To illustrate these concepts in practice, Eticas presented two case studies. The first examined a hate speech classifier, showing how technical and socio-technical system mapping can reveal risks related to transparency, appeals mechanisms and bias. The second focused on the TWIN4DEM Digital Twin, demonstrating how auditing methodologies are being adapted to evaluate an agent-based simulation designed to support research on executive aggrandisement and democratic resilience.

The Digital Twin example highlighted several important considerations, including documenting system limitations through model cards, improving transparency for end users, supporting consistent scenario construction, and identifying structural assumptions that may influence simulation outcomes. These examples illustrated that auditing is not only about identifying failures, but also about strengthening confidence in AI systems by clearly communicating both their capabilities and limitations.

The workshop concluded with an interactive discussion on how researchers, policymakers and technology developers can work together to ensure AI systems remain accountable, transparent and centred on human oversight. By bringing together technical expertise and stakeholder perspectives, the event contributed to TWIN4DEM’s broader mission of developing trustworthy AI tools that support democratic governance.

As the project progresses, TWIN4DEM will continue organising stakeholder engagement activities to foster dialogue, share project developments and promote responsible AI practices across research and policymaking communities.

As part of TWIN4DEM’s commitment to openness and transparency, the project’s GitHub repository is now publicly available. The repository hosts the project’s technical developments and will serve as a central location for sharing code, tools, and technical outputs as they become available throughout the project.

Researchers, policymakers, developers, and other interested stakeholders are invited to follow the repository to stay informed about future updates and developments related to the TWIN4DEM Digital Twin.

The repository currently contains the first publicly released code and technical outputs, with additional content expected to be added as research activities progress. All code is released under the EUPL v1.2 licence, supporting reuse and further development in line with open-source principles.

Follow the repository for updates and learn more about the technical work supporting TWIN4DEM’s research on democratic resilience and executive aggrandisement.

Work Package 4 focuses on creating the integrated datasets that will power the TWIN4DEM Digital Twin. Bringing together socio-economic data from surveys, elections, censuses, and other public sources, alongside textual data generated in previous project activities, WP4 establishes the data backbone needed for advanced democratic analysis.

The main objectives of WP4 are to:

• Identify the most relevant and reliable data sources for the Digital Twin
• Clean, harmonise, and standardise data from diverse sources
• Aggregate and connect datasets into coherent, interoperable structures
• Ensure compliance with legal, ethical, and data protection requirements
• Deliver integrated datasets ready to support modelling, simulation, and analysis activities in WP5

By transforming complex and fragmented information into high-quality, interconnected datasets, WP4 lays the groundwork for the development of a robust Digital Twin.

This question guided two focus group discussions held in spring 2026, bringing together researchers, civil society representatives and policy experts interested in the development of innovative tools for analysing democratic systems.

The sessions built on previous discussions and provided participants with a closer look at the latest version of a digital simulation platform designed to explore how political institutions and actors interact under different conditions.

Testing political scenarios

During the discussions, participants explored a range of simulated political scenarios, including majority and minority governments, coalition dynamics and the influence of individual political actors on decision-making processes. Through a combination of demonstrations and interactive discussions, participants examined how changes in political circumstances could affect the behaviour of institutions and broader political outcomes. The objective was not to predict the future, but to better understand how different factors may influence democratic systems and governance dynamics.

Transparency and Accessibility as Key Themes

A recurring topic throughout the discussions was the importance of making complex political analysis understandable and accessible. Participants highlighted the value of clear explanations regarding how simulations operate, how results are generated and how users can interpret the outcomes. The conversations also explored broader questions related to transparency, trust and the communication of complex information through digital platforms. The discussions demonstrated that technological innovation alone is not sufficient; users also need confidence in the information presented and a clear understanding of the underlying processes.

Looking ahead

The focus groups formed part of an ongoing effort to engage stakeholders in discussions about the opportunities and challenges of using digital technologies to explore political and institutional developments. The conversations highlighted strong interest in innovative approaches that can make complex democratic processes more visible and understandable, while fostering informed dialogue among diverse communities and experts.

What happens when democratic institutions begin to weaken, and how can we detect it early?

In this TWIN4DEM interview, the Work Package 3 leader discusses how computational text analysis and AI-driven methodologies can help identify patterns of executive aggrandisement, democratic backsliding, and societal responses across Europe.

WP3 explores large-scale political and public discourse through innovative computational approaches, contributing to the development of evidence-based tools for safeguarding democracy, transparency, and the rule of law.

The interview covers:
• The objectives and scope of WP3
• Computational approaches to democracy research
• Detecting executive aggrandisement through text analysis
• Monitoring political and societal reactions
• The role of AI and digital methods in social sciences

In this interview, the leader of Work Package 2 explains the frameworks and analytical approaches used to conceptualize the various pathways to executive aggrandizement. The discussion provides insight into the methodological foundations underpinning the TWIN4DEM project’s research on democratic resilience, examining how executive power expansion threatens democratic institutions and the analytical tools developed to study these phenomena.

In March, the TWIN4DEM project continued its participatory work through a new round of online focus groups in Hungary, Czechia, France and the Netherlands. The sessions brought together a diverse group of stakeholders, including policymakers, civil society representatives, journalists, and researchers, ensuring a wide range of perspectives.

These discussions marked an important milestone in the project’s development. Participants were invited to explore the first version of the TWIN4DEM “toy model”, an early-stage digital twin designed to simulate pathways of executive aggrandizement. This initial exposure allowed stakeholders to engage directly with the concept and provide feedback on its structure, functionality, and potential applications.

The discussions generated valuable insights into how the model can be made both meaningful and usable for its intended audiences. A recurring point raised by participants was the importance of transparency: for the model to gain trust and be widely adopted, it must clearly communicate not only its outputs and scenario variations, but also the underlying mechanisms that generate them.

In parallel, participants highlighted key ethical and governance considerations. Ensuring full compliance with data protection standards, including GDPR, was identified as a fundamental requirement. Particular attention was given to decisions around data storage and infrastructure, as well as the need to provide clear, accessible explanations of how user data is collected, processed, and managed—especially in cases where external services are integrated.

To further strengthen this co-creation process, two additional online focus group discussions are scheduled for 29 April and 11 May. These sessions will aim to gather further input, with a specific focus on increasing participation from Hungary and Czechia, in order to ensure stronger contextual representation and more locally grounded insights.

The feedback collected throughout these activities will directly inform the next phase of the project. The TWIN4DEM team will continue refining the model, expanding its data foundations, and preparing it for further testing and validation. Ongoing stakeholder engagement remains central to this process, ensuring that the digital twin evolves as a robust, transparent, and policy-relevant tool aligned with real-world democratic needs.