What the World Really Thinks About AI – From Hopes to Fears

Understanding public hopes and fears about artificial intelligence (AI) is crucial, as these perceptions shape public trust, influence policy decisions, and guide the trajectory of responsible AI development. However, such sentiments vary widely across cultural and national contexts, underscoring the need for predictive models that can generalize across diverse populations.

To date, most research has focused on descriptive analyses or geographically limited case studies. Specifically, there has been little effort to develop predictive models capable of inferring individual hopes or fears from free-text responses, particularly in a cross-cultural or multilingual setting. This leaves a significant gap in creating scalable tools grounded in rich, diverse datasets.

To partially address this gap, the proposed work will develop predictive models that classify whether a person expresses hope or fear about AI based on textual input. These models will be trained and evaluated using:

1. The cross-national dataset collected by Dong et al., covering 20 countries (downloadable here).
2. The U.S.-focused dataset from Moreira et al., collected via an interactive platform.
3. A new dataset to be collected through the deployment of the "Hopes and Fears" platform in the English-speaking countries from Dong et al.'s sample.

References

• Dong et al. – Fears About Artificial Intelligence Across 20 Countries and Six Domains of Application
• Moreira et al. – The Hall of AI Fears and Hopes: Comparing the Views of AI Influencers and those of Members of the U.S. Public Through an Interactive Platform –