Invited talks

The Invited Talks at IDEAL 2025 bring together experts to share their latest research and practical insights straightforwardly and engagingly. These sessions are an excellent opportunity to explore new ideas, tackle current challenges, and see how cutting-edge techniques are applied in the real world. This year, we are very proud to host the following invited talks:

1. The Next Frontier in Explainable AI: Human-AI Synergy.

As AI systems become more complex, the need for effective explanations grows, especially in contexts where human decision-making is involved. This plenary talk explores the evolving role of explainable AI (XAI) in fostering human-AI collaboration. While traditional approaches to XAI focus on mathematical approaches looking for transparency but dificult for society  understanding,  the next frontier lies in creating explanations that empower users  and end-users better understand AI’s reasoning and AI’s advises and enhance their decision-making capabilities.

We will examine how well-designed explanations can bridge the gap between AI outputs and human comprehension, enabling more informed decisions in critical areas such as healthcare, finance, and autonomous systems. The focus will be on the utility of explanations in making AI more actionable and trustworthy, fostering a symbiotic relationship where humans and AI co-create decisions. By focusing on human-AI synergy, we aim to move beyond simply explaining AI’s behavior to facilitating collaborative decision-making, ensuring that explanations truly serve the needs of users and support effective, ethical decision-making.

2. Human-Centered NLP for Behavioral Health: From Conversation to Coaching

Conversations shape how we connect, learn, and heal. In behavioral and mental health, the words exchanged between patients and clinicians can profoundly influence outcomes — yet providing clinicians with feedback on how they communicate remains a slow, resource-intensive process. This keynote explores how advances in natural language processing (NLP) can help bridge that gap through AI systems that analyze, evaluate, and coach human conversations in this domain. I will present computational models that automatically assess and coach the acquisition of key counseling behaviors, such as empathy and reflective listening, offering real-time actionable feedback to clinicians and trainees. These systems are based on large language models fine-tuned for psychotherapy contexts and developed in close partnership with clinicians and practitioners. Beyond automating evaluation, they open new possibilities for personalized coaching and scalable communication training. Throughout, I will discuss the broader challenges and opportunities of applying NLP to sensitive human domains — emphasizing transparency, ethics, and the importance of keeping humans at the center. Together, these efforts point toward a future where language technologies do more than process what we say — they help us say it better.