Generative AI’s potential to democratise knowledge is accompanied by risks. Over-reliance on AI-generated information and the spread of misinformation are concerns, especially for individuals with limited critical thinking skills or less exposure to technology. These issues could exacerbate existing disparities, particularly in education and healthcare, where accurate, personalised information is crucial for effective outcomes.

This research focuses on tuning the parameter settings of Generative AI models. In one condition, the AI tool will provide more deterministic, precise outputs, while in the other condition, the tool will introduce creativity and diversity at the expense of coherence and accuracy. The study will examine how different parameter settings affect comprehension, innovation, and decision-making across various groups, particularly those from disadvantaged backgrounds.

By understanding the effects of communication style in AI responses, this study aims to improve the way Generative AI is deployed in sensitive sectors. The findings will provide valuable insights for developers, educators, and policymakers, promoting responsible AI use while ensuring that its applications are equitable. Ultimately, the research seeks to contribute to advancing both the theoretical and practical understanding of human-AI interaction, particularly in how AI can support innovation while minimising risks such as misinformation.