Michael Haman
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Michael Haman has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Milan Školník
Department of Humanities, Czech University of Life Sciences Prague, Czech Republic
19 shared publications Jan Čopík
Czech University of Life Sciences Prague Praha Czech Republic
4 shared publicationsMilan Školník
Department of Humanities Faculty of Economics and Management Czech University of Life Sciences Prague Czech Republic
3 shared publications- Michal Lošťák
Department of Humanities, Czech University of Life Sciences Prague https://dx.doi.org/48371 Prague Czech Republic
2 shared publications - Aida Khakimova
Russian New University
1 shared publication - Anna Marie Nešpůrková
Student research assistant at the Centre for Ibero-American Studies, Faculty of Arts , Charles University (Voršilská 1, 116 43 Praha 1)
1 shared publication
Latest Publications
- Open Access
Can Large Language Models Replace Human Coders? Introducing ContentBench
Can low-cost large language models (LLMs) take over the interpretive coding work that still anchors much of empirical content analysis? This paper introduces ContentBench, a public benchmark suite that helps answer this replacement question by tracking how much agreement low-cost LLMs achieve and what they cost on the same interpretive coding tasks. The suite uses versioned tracks that invite researchers to contribute new benchmark datasets. I report results from the first track, ContentBench-ResearchTalk v1.0: 1,000 synthetic, social-media-style posts about academic research labeled into five categories spanning praise, critique, sarcasm, questions, and procedural remarks. Reference labels are assigned only when three state-of-the-art reasoning models (GPT-5, Gemini 2.5 Pro, and Claude Opus 4.1) agree unanimously, and all final labels are checked by the author as a quality-control audit. Among the 59 evaluated models, the best low-cost LLMs reach roughly 97-99% agreement with these jury labels, far above GPT-3.5 Turbo, the model behind early ChatGPT and the initial wave of LLM-based text annotation. Several top models can code 50,000 posts for only a few dollars, pushing large-scale interpretive coding from a labor bottleneck toward questions of validation, reporting, and governance. At the same time, small open-weight models that run locally still struggle on sarcasm-heavy items (for example, Llama 3.2 3B reaches only 4% agreement on hard-sarcasm). ContentBench is released with data, documentation, and an interactive quiz at contentbench.github.io to support comparable evaluations over time and to invite community extensions.
Open AccessCan Large Language Models Replace Human Coders? Introducing ContentBench
Can low-cost large language models (LLMs) take over the interpretive coding work that still anchors much of empirical content analysis? This paper introduces ContentBench, a public benchmark suite that helps answer this replacement question by tracking how much agreement low-cost LLMs achieve and what they cost on the same interpretive coding tasks. The suite uses versioned tracks that invite researchers to contribute new benchmark datasets. I report results from the first track, ContentBench-ResearchTalk v1.0: 1,000 synthetic, social-media-style posts about academic research labeled into five categories spanning praise, critique, sarcasm, questions, and procedural remarks. Reference labels are assigned only when three state-of-the-art reasoning models (GPT-5, Gemini 2.5 Pro, and Claude Opus 4.1) agree unanimously, and all final labels are checked by the author as a quality-control audit. Among the 59 evaluated models, the best low-cost LLMs reach roughly 97-99% agreement with these jury labels, far above GPT-3.5 Turbo, the model behind early ChatGPT and the initial wave of LLM-based text annotation. Several top models can code 50,000 posts for only a few dollars, pushing large-scale interpretive coding from a labor bottleneck toward questions of validation, reporting, and governance. At the same time, small open-weight models that run locally still struggle on sarcasm-heavy items (for example, Llama 3.2 3B reaches only 4% agreement on hard-sarcasm). ContentBench is released with data, documentation, and an interactive quiz at contentbench.github.io to support comparable evaluations over time and to invite community extensions.
- Open Access
- 23 September 2025
- Published by Cogitatio in Media and Communication
- Vol. 13
Digital Divides in Local Democracy: Size, Resources, and Facebook Adoption in Czech Municipalities
This study examines how organizational capacity and environmental pressures shape social media adoption in local government. Using a mixed‐methods approach, we analyze Facebook adoption patterns across 6,254 Czech municipalities by scanning their official websites for Facebook links, a questionnaire survey (𝑁 = 330), and thematic analysis of posts from municipalities with extended powers. Results reveal a strong relationship between municipal size and Facebook adoption, with rates ranging from 14.1% in the smallest municipalities (fewer than 200 inhabitants) to 88.6% in the largest (10,000+). Primary adoption motivations include communication speed, ensuring citizen awareness, and meeting public expectations for a modern online presence, while resource constraints and fear of negativity constitute major barriers. Thematic analysis reveals dominance of practical information and community event promotion over dialogic engagement. The findings demonstrate how resource limitations create a digital divide in local government communication, with smaller municipalities facing structural barriers to social media adoption. This research contributes to understanding technology diffusion dynamics in fragmented administrative systems and the gap between social media’s dialogic potential and actual government communication practices.
- 23 September 2025
- Published by Cogitatio in Media and Communication
- Vol. 13
Open AccessDigital Divides in Local Democracy: Size, Resources, and Facebook Adoption in Czech Municipalities
This study examines how organizational capacity and environmental pressures shape social media adoption in local government. Using a mixed‐methods approach, we analyze Facebook adoption patterns across 6,254 Czech municipalities by scanning their official websites for Facebook links, a questionnaire survey (𝑁 = 330), and thematic analysis of posts from municipalities with extended powers. Results reveal a strong relationship between municipal size and Facebook adoption, with rates ranging from 14.1% in the smallest municipalities (fewer than 200 inhabitants) to 88.6% in the largest (10,000+). Primary adoption motivations include communication speed, ensuring citizen awareness, and meeting public expectations for a modern online presence, while resource constraints and fear of negativity constitute major barriers. Thematic analysis reveals dominance of practical information and community event promotion over dialogic engagement. The findings demonstrate how resource limitations create a digital divide in local government communication, with smaller municipalities facing structural barriers to social media adoption. This research contributes to understanding technology diffusion dynamics in fragmented administrative systems and the gap between social media’s dialogic potential and actual government communication practices.
- 16 February 2025
- Published by Taylor & Francis in Accountability in Research
- Vol. 33 (1) , 1-3
Fake no more: The redemption of ChatGPT in literature reviews
- 16 February 2025
- Published by Taylor & Francis in Accountability in Research
- Vol. 33 (1) , 1-3
Fake no more: The redemption of ChatGPT in literature reviews
- 1 October 2024
- Published by Oxford University Press (OUP) in Journal of Public Health
- Vol. 46 (4) , e715-e716
The rise of talking machines: balancing the potential and pitfalls of voice chatbots for mental wellbeing
Dear Editor, Recently, the debate about the impact of artificial intelligence (AI) on public health has been reignited in the Journal of Public Health. 1, 2 The editorial ‘What’s artificial intelligence (AI) got to do with it-inequality and public health?’ 3 brought this debate to life, highlighting both the transformative potential of AI and the looming risks of exacerbating existing inequalities. In the context of this discussion, particularly in light of the recent launch of Google’s voice interface in Gemini Live 4 and the planned expansion of Advanced Voice Mode 5 on ChatGPT to all paid subscribers in the coming months, we would like to contribute a new perspective that has not yet been fully explored: the potential of voice chatbots. While ChatGPT’s text-based interface has already proven popular, voice interaction has the potential to deepen user engagement, provide a more natural form of communication and overcome some of the barriers associated with text-based interaction. When exploring the potential of voice chatbots in the mental health field, it is also essential to critically assess the potential drawbacks and risks associated with their use. Two key areas of concern are anthropomorphization and language barriers.
- 1 October 2024
- Published by Oxford University Press (OUP) in Journal of Public Health
- Vol. 46 (4) , e715-e716
The rise of talking machines: balancing the potential and pitfalls of voice chatbots for mental wellbeing
Dear Editor, Recently, the debate about the impact of artificial intelligence (AI) on public health has been reignited in the Journal of Public Health. 1, 2 The editorial ‘What’s artificial intelligence (AI) got to do with it-inequality and public health?’ 3 brought this debate to life, highlighting both the transformative potential of AI and the looming risks of exacerbating existing inequalities. In the context of this discussion, particularly in light of the recent launch of Google’s voice interface in Gemini Live 4 and the planned expansion of Advanced Voice Mode 5 on ChatGPT to all paid subscribers in the coming months, we would like to contribute a new perspective that has not yet been fully explored: the potential of voice chatbots. While ChatGPT’s text-based interface has already proven popular, voice interaction has the potential to deepen user engagement, provide a more natural form of communication and overcome some of the barriers associated with text-based interaction. When exploring the potential of voice chatbots in the mental health field, it is also essential to critically assess the potential drawbacks and risks associated with their use. Two key areas of concern are anthropomorphization and language barriers.
- Open Access
Who Would Chatbots Vote For? Political Preferences of ChatGPT and Gemini in the 2024 European Union Elections
This study examines the political bias of chatbots powered by large language models, namely ChatGPT and Gemini, in the context of the 2024 European Parliament elections. The research focused on the evaluation of political parties represented in the European Parliament across 27 EU Member States by these generative artificial intelligence (AI) systems. The methodology involved daily data collection through standardized prompts on both platforms. The results revealed a stark contrast: while Gemini mostly refused to answer political questions, ChatGPT provided consistent ratings. The analysis showed a significant bias in ChatGPT in favor of left-wing and centrist parties, with the highest ratings for the Greens/European Free Alliance. In contrast, right-wing parties, particularly the Identity and Democracy group, received the lowest ratings. The study identified key factors influencing the ratings, including attitudes toward European integration and perceptions of democratic values. The findings highlight the need for a critical approach to information provided by generative AI systems in a political context and call for more transparency and regulation in this area.
Open AccessWho Would Chatbots Vote For? Political Preferences of ChatGPT and Gemini in the 2024 European Union Elections
This study examines the political bias of chatbots powered by large language models, namely ChatGPT and Gemini, in the context of the 2024 European Parliament elections. The research focused on the evaluation of political parties represented in the European Parliament across 27 EU Member States by these generative artificial intelligence (AI) systems. The methodology involved daily data collection through standardized prompts on both platforms. The results revealed a stark contrast: while Gemini mostly refused to answer political questions, ChatGPT provided consistent ratings. The analysis showed a significant bias in ChatGPT in favor of left-wing and centrist parties, with the highest ratings for the Greens/European Free Alliance. In contrast, right-wing parties, particularly the Identity and Democracy group, received the lowest ratings. The study identified key factors influencing the ratings, including attitudes toward European integration and perceptions of democratic values. The findings highlight the need for a critical approach to information provided by generative AI systems in a political context and call for more transparency and regulation in this area.
- 19 August 2024
- Published by Elsevier in Resuscitation
- Vol. 202, 110367
Beyond English: ChatGPT’s instructions across EU languages
- 19 August 2024
- Published by Elsevier in Resuscitation
- Vol. 202, 110367
Beyond English: ChatGPT’s instructions across EU languages
- Open Access
- 16 April 2024
- Published by Springer Nature in Security Journal
- Vol. 37 (4) , 1447-1462
Drawing the line: public attitudes towards warranted and unwarranted government surveillance in European countries
This article investigates public attitudes towards government surveillance for law enforcement and national security in different European countries, considering the broader context of citizens' security concerns. These concerns encompass crime within individual countries, potential terrorist attacks, activities of foreign agents, and even inter-state conflicts, such as the Russian Federation's military aggression against Ukraine. Using data from the International Social Survey Programme, we examine opinions about various types of government surveillance. Regression analysis is employed to determine which variables influence these opinions. Notably, there is a tangible hesitation among citizens regarding such surveillance, stemming from fears of privacy invasion and potential misuse of collected data. Our findings indicate greater support for government surveillance in Western and Nordic countries compared to post-Communist countries. Respondents with a comprehensive understanding of political issues and those advocating for government spending on security generally held a more positive view of government surveillance rights. On the other hand, perceptions of higher corruption were linked to negative views on surveillance.
- 16 April 2024
- Published by Springer Nature in Security Journal
- Vol. 37 (4) , 1447-1462
Open AccessDrawing the line: public attitudes towards warranted and unwarranted government surveillance in European countries
This article investigates public attitudes towards government surveillance for law enforcement and national security in different European countries, considering the broader context of citizens' security concerns. These concerns encompass crime within individual countries, potential terrorist attacks, activities of foreign agents, and even inter-state conflicts, such as the Russian Federation's military aggression against Ukraine. Using data from the International Social Survey Programme, we examine opinions about various types of government surveillance. Regression analysis is employed to determine which variables influence these opinions. Notably, there is a tangible hesitation among citizens regarding such surveillance, stemming from fears of privacy invasion and potential misuse of collected data. Our findings indicate greater support for government surveillance in Western and Nordic countries compared to post-Communist countries. Respondents with a comprehensive understanding of political issues and those advocating for government spending on security generally held a more positive view of government surveillance rights. On the other hand, perceptions of higher corruption were linked to negative views on surveillance.
- Open Access
- 1 April 2024
- Published by Walter de Gruyter GmbH in ACTA POLITOLOGICA
- Vol. 16 (1) , 1-17
Why Did I Lose My Seat in the United States Congressional Elections? You Didn‘t Advertise Online Enough!
In this paper, we examine campaign spending on Facebook during the 2020 U.S. congressional elections. We draw conclusions based on data from the Federal Election Commission and the Facebook Ad Library. This is one of the first articles on this topic. We point out and show how this data differs and how other researchers should treat it. Specifically, we focus on U.S. Democratic congresspeople who lost elections. Indeed, during the Democratic Party debate, voices were raised that its progressive wing was to blame for the loss of seats; in contrast, Rep. Alexandria Ocasio-Cortez, a rising star of the wing, argued that the lack of Facebook advertising was a factor in why seats were lost. The article concludes that the really well-known Democratic figures who lost the elections spent very little on Facebook ads compared to Ocasio-Cortez. We also compare Republican opponents with Democrats. In addition, we highlight how the Ocasio-Cortez campaign and the others differed with respect to the timing of campaign spending and the targeting of Facebook ads to specific states. Further, we compare data regarding campaigns available through the Federal Election Commission and Facebook and underscore the importance of the transparency of the Facebook Ad Library.
- 1 April 2024
- Published by Walter de Gruyter GmbH in ACTA POLITOLOGICA
- Vol. 16 (1) , 1-17
Open AccessWhy Did I Lose My Seat in the United States Congressional Elections? You Didn‘t Advertise Online Enough!
In this paper, we examine campaign spending on Facebook during the 2020 U.S. congressional elections. We draw conclusions based on data from the Federal Election Commission and the Facebook Ad Library. This is one of the first articles on this topic. We point out and show how this data differs and how other researchers should treat it. Specifically, we focus on U.S. Democratic congresspeople who lost elections. Indeed, during the Democratic Party debate, voices were raised that its progressive wing was to blame for the loss of seats; in contrast, Rep. Alexandria Ocasio-Cortez, a rising star of the wing, argued that the lack of Facebook advertising was a factor in why seats were lost. The article concludes that the really well-known Democratic figures who lost the elections spent very little on Facebook ads compared to Ocasio-Cortez. We also compare Republican opponents with Democrats. In addition, we highlight how the Ocasio-Cortez campaign and the others differed with respect to the timing of campaign spending and the targeting of Facebook ads to specific states. Further, we compare data regarding campaigns available through the Federal Election Commission and Facebook and underscore the importance of the transparency of the Facebook Ad Library.
- Open Access
- 20 December 2023
- Published by Taylor & Francis in Journal of Information Technology & Politics
- Vol. 22 (2) , 167-171
The unverified era: politicians’ Twitter verification post-Musk acquisition
- 20 December 2023
- Published by Taylor & Francis in Journal of Information Technology & Politics
- Vol. 22 (2) , 167-171
Open AccessThe unverified era: politicians’ Twitter verification post-Musk acquisition