People increasingly rely on Artificial Intelligence (AI) based systems to aid decision-making and perform daily tasks. We explored three stages of human-AI interactions: choice, usage, and outcomes. In the first project, we examined the effects of users’ perception of AI systems’ warmth (perceived intent) and competence (perceived ability) on their choices, showing that, similar to the judgments of other people, there is often primacy for warmth over competence. The second and third projects examine the effects of AI-generated text suggestions on writers’ usage, their texts, and readers’ perception of the texts. We found a discrepancy between the effects of text suggestion on writers, readers, and texts. While text suggestions influenced the texts, they had a limited impact on writers’ and readers’ perceptions. Our results suggest that it may be vital for AI systems designers to consider and communicate the system’s warmth characteristics to its potential users. Furthermore, our initial results might suggest potential risks of using AI systems to assist writers in text composition.
In the current study, we aim to examine this approach in the financial markets field, exploring the prediction of human behavior and stock market price movement. The current thesis is divided into two main parts. The first part investigates the behavioral biases of individual retail investors in stock market trading. Following the behavioral biases discovered in the first part, the second part tests the additive value of the approach combining ML and behavioral economics compared to pure ML in predicting stock market returns in the financial markets. The analysis will cover a recent dataset we constructed, as well as two state-of-the-art benchmark datasets which utilize deep learning models for stock market movement prediction.