From Explainable to Interactive AI: A Literature Review on Current Trends in Human-AI Interaction (2405.15051v1)

Abstract: AI systems are increasingly being adopted across various domains and application areas. With this surge, there is a growing research focus and societal concern for actively involving humans in developing, operating, and adopting these systems. Despite this concern, most existing literature on AI and Human-Computer Interaction (HCI) primarily focuses on explaining how AI systems operate and, at times, allowing users to contest AI decisions. Existing studies often overlook more impactful forms of user interaction with AI systems, such as giving users agency beyond contestability and enabling them to adapt and even co-design the AI's internal mechanics. In this survey, we aim to bridge this gap by reviewing the state-of-the-art in Human-Centered AI literature, the domain where AI and HCI studies converge, extending past Explainable and Contestable AI, delving into the Interactive AI and beyond. Our analysis contributes to shaping the trajectory of future Interactive AI design and advocates for a more user-centric approach that provides users with greater agency, fostering not only their understanding of AI's workings but also their active engagement in its development and evolution.

• The paper demonstrates that moving from passive explainability to interactive design enhances user control and system transparency.
• It uses a comprehensive literature review to analyze trends in feature customization, algorithmic transparency, and collaborative design within AI.
• The study calls for further research on integrating iterative feedback and ethical considerations to develop more adaptive, user-centric AI systems.

The paper "From Explainable to Interactive AI: A Literature Review on Current Trends in Human-AI Interaction" provides a comprehensive survey that addresses the evolving landscape of Human-AI interaction, particularly advancing from mere explainability towards deeper and more substantive forms of engagement. This paper, distinguishing itself from prior literature, underscores the necessity of granting users more significant roles and influence over AI systems.

The authors start by contextualizing the increasing adoption of AI systems across various domains and the accompanied societal concerns. They argue that despite the progress in Explainable AI (XAI) and Contestable AI, which primarily focus on enabling users to understand and challenge AI decisions, these approaches often fall short of offering users substantial agency or participation in shaping AI behavior.

The paper goes beyond conventional frameworks by exploring the concept of Interactive AI, which emphasizes the active involvement of users not merely as passive recipients or occasional contesters but as integral collaborators in the design and evolution of AI systems. This shift envisages users who can adapt AI functionalities to their needs and even co-design its internal algorithms, bestowing them with more control and empowerment.

The authors organize their review around several emergent trends and identify gaps in current research. They suggest future directions for creating more user-centric AI systems, including:

1. Feature Customization: Allowing users to tailor AI features to better suit personal or situational requirements.
2. Algorithmic Transparency: Not just explaining AI decisions, but providing users with insights into the underlying mechanics and letting them tweak or understand these processes deeply.
3. Collaborative Design: Facilitating user participation in the iterative design process of AI systems, making collaboration a core part of AI development.
4. Feedback Integration: Mechanisms for continuously integrating user feedback to refine and optimize AI performance dynamically.
5. Ethical Considerations: Addressing ethical issues related to user agency and ensuring that user involvement does not compromise fairness or introduce biases.

By advocating for these advancements, the paper highlights the transformative potential of Interactive AI. It aims to foster a paradigm where users are empowered to engage with, modify, and enhance AI systems, ultimately leading to more adaptive, responsive, and user-driven technology. The authors call for a broader perspective in Human-Centered AI research, one that pushes the boundaries of how humans interact with and influence intelligent systems.

This comprehensive review aids in setting the future trajectory for Interactive AI, emphasizing a collaborative and user-centric approach that could lead to more effective and socially responsible AI systems.