True utility emerges when AI recognizes its own boundaries and knows when to ask clarifying questions instead of providing confident, yet erroneous, answers. This philosophy guides my research, as I am convinced that understanding the limitations of ML models is crucial for building truly trustworthy, robust, and deployable AI systems in high-stakes environments. See scholar for publication list.