Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Public sentiments on the fourth industrial revolution: An unsolicited public opinion poll from Twitter

Diletta Abbonato

TL;DR

This study establishes a pre-ChatGPT baseline of European public sentiment toward Fourth Industrial Revolution technologies using ~90,000 multilingual tweets and news articles from six countries (2006–2019). It combines narratives, diffusion economics, and information-environment theories within a transformer-based analytical framework (XLM-RoBERTa for sentiment, DeBERTa for themes) to measure polarization, topic salience, and echo-chamber dynamics, including misinformation exposure. The findings reveal pervasive polarization, notable cross-national variation, and emerging echo chambers, with privacy and environmental topics showing the strongest polarization and higher exposure to low-quality information, while health remains relatively moderate. The results provide a rigorous reference point for evaluating post-ChatGPT shifts in discourse and offer actionable insights for AI governance, public engagement, and domain-specific policy design in Europe.

Abstract

This paper establishes an empirical baseline of public sentiment toward Fourth Industrial Revolution (4IR) technologies across six European countries during the period 2006--2019, prior to the widespread adoption of generative AI systems. Employing transformer-based natural language processing models on a corpus of approximately 90,000 tweets and news articles, I document a European public sphere increasingly divided in its assessment of technological change: neutral sentiment declined markedly over the study period as citizens sorted into camps of enthusiasm and concern, a pattern that manifests distinctively across national contexts and technology domains. Approximately 6\% of users inhabit echo chambers characterized by sentiment-aligned networks, with privacy discourse exhibiting the highest susceptibility to such dynamics. These findings provide a methodologically rigorous reference point for evaluating how the introduction of ChatGPT and subsequent generative AI systems has transformed public discourse on automation, employment, and technological change. The results carry implications for policymakers seeking to align technological governance with societal values in an era of rapid AI advancement.

Public sentiments on the fourth industrial revolution: An unsolicited public opinion poll from Twitter

TL;DR

This study establishes a pre-ChatGPT baseline of European public sentiment toward Fourth Industrial Revolution technologies using ~90,000 multilingual tweets and news articles from six countries (2006–2019). It combines narratives, diffusion economics, and information-environment theories within a transformer-based analytical framework (XLM-RoBERTa for sentiment, DeBERTa for themes) to measure polarization, topic salience, and echo-chamber dynamics, including misinformation exposure. The findings reveal pervasive polarization, notable cross-national variation, and emerging echo chambers, with privacy and environmental topics showing the strongest polarization and higher exposure to low-quality information, while health remains relatively moderate. The results provide a rigorous reference point for evaluating post-ChatGPT shifts in discourse and offer actionable insights for AI governance, public engagement, and domain-specific policy design in Europe.

Abstract

This paper establishes an empirical baseline of public sentiment toward Fourth Industrial Revolution (4IR) technologies across six European countries during the period 2006--2019, prior to the widespread adoption of generative AI systems. Employing transformer-based natural language processing models on a corpus of approximately 90,000 tweets and news articles, I document a European public sphere increasingly divided in its assessment of technological change: neutral sentiment declined markedly over the study period as citizens sorted into camps of enthusiasm and concern, a pattern that manifests distinctively across national contexts and technology domains. Approximately 6\% of users inhabit echo chambers characterized by sentiment-aligned networks, with privacy discourse exhibiting the highest susceptibility to such dynamics. These findings provide a methodologically rigorous reference point for evaluating how the introduction of ChatGPT and subsequent generative AI systems has transformed public discourse on automation, employment, and technological change. The results carry implications for policymakers seeking to align technological governance with societal values in an era of rapid AI advancement.

Paper Structure

This paper contains 34 sections, 8 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Theoretical Framework
  3. Technology Narratives: Hopes and Fears
  4. Economic Perspectives on Technology Adoption
  5. Information Environments and Opinion Dynamics
  6. Integrating Perspectives: A Framework for Analysis
  7. Methodological Approach
  8. Transformer Models for Text Analysis
  9. Sentiment Classification with XLM-RoBERTa
  10. Thematic Classification with Zero-Shot Learning
  11. Echo Chamber Identification
  12. Misinformation Exposure Assessment
  13. Methodological Considerations
  14. Data and Empirical Strategy
  15. Sampling Strategy and Data Sources
  16. ...and 19 more sections

Figures (8)

  • Figure 1: Different aspects of public opinion dynamics
  • Figure 2: Conceptual framework for analyzing public sentiment toward 4IR technologies
  • Figure 3: Frequency of 4IR technology mentions in corpus
  • Figure 4: Temporal evolution of sentiment in 4IR discourse (2014--2019). Both newspaper coverage and user responses exhibit declining neutrality and increasing polarization.
  • Figure 5: Sentiment dynamics by 4IR technology. Green: neutral; Blue: positive; Red: negative. All technologies exhibit declining neutrality.
  • ...and 3 more figures