How can explainable AI enhance Human-AI Collaboration

Autonomous agents permeate society and the workplace, transforming the way organisations operate and offering human-AI collaboration opportunities in almost every domain. However, the lack of transparency and interpretability of AI systems 1 has raised several novel ethical questions. For instance, the diminishment of the value of human input

in the decision making process due to over-reliance on AI models as 'black box' systems. Over-reliance can foster a false sense of trust in the system's correctness leading to a habitual acceptance of the model's outputs without rationalisation and verification of its process of producing them2 . This poses the risk of rendering the system less resilient and

less human-centric as it lacks quality feedback from its human counterpart. An important method to counteract this process and increase the efficacy of human-AI collaboration is the use of explainable AI(XAI) 3 . By investigating methods that improve the explainability of AI models, we can develop tools that make it easier to understand the

model's capabilities and limitations, detect unwanted behaviours and allow the human to steer the model's behaviour to the appropriate direction. Tailored explanations to the user's expertise and background can help build trust and enable effective collaboration between humans and AI systems 4 . Aim and Methods

The aim of this research proposal is to explore the emerging field of human-AI collaboration, with a specific focus on explainable AI, as a method to increase efficiency and adaptation. The objectives of the research include: 1. Literature review: Gathering knowledge and insights on human-AI collaboration,

explainable AI, human-centric design, and ethical considerations in AI systems. It will be achieved by conducting a comprehensive review of existing literature on these topics. This will provide a foundation for understanding the current state of research and identify gaps.