Negativity bias and the use of paid generative AI: A survey experiment

Abstract

The prevalence of generative artificial intelligence (GAI) in the modern age has given rise to new variants of these innovative technologies. The emergence of the premium version of these advanced tools has not only reshaped the definition of innovation but captivated the attention of the masses, particularly for learning. Despite the potential, little effort has been devoted to understanding the underlying motivations and reluctance that drive individuals to embrace or resist these versatile, emerging paid versions. This between-subjects experimental study, using a dataset of 1,703 participants, examines the impact of positive and negative information valence conditions on users' perceptions of paid GAI tools and their behaviour. Through the lens of negativity bias, the research investigates how exposure to contrasting narratives about technology impacts perceptions and ultimately the intention to incorporate such tools for learning. The results of this study confirm the dominance of negative superiority compared to positive valence regarding their impacts on the shifting of participants’ perceptions and responses. Furthermore, “social influence”, “price value” and “trust” have been identified as the prominent aspects that influence participants’ behavioural intention towards using paid GAI tools to serve for their academic goal, amongst them, “trust” is the most influential. These findings offer invaluable contributions to the current literature and future policies toward GAI technologies, highlighting that AI developers and policymakers should focus on minimising criticism rather than showcasing favourable attributes