Opening Lines

As AI and machine learning become deeply embedded in software systems—from autonomous vehicles to healthcare diagnostics—ensuring their safety and reliability has never been more critical. Yet, the unique challenges posed by AI/ML, such as opaque decision-making, data bias, and evolving attack surfaces, demand new paradigms in verification, testing, and risk mitigation. How do we reconcile the rapid pace of AI innovation with the non-negotiable demands of software dependability? And what lessons can traditional software engineering teach us—or learn from—this new frontier?

Panel Scope

This panel brings together leading researchers and practitioners to discuss:

• The evolving landscape of software safety/reliability in AI-driven systems.
• Bridging the gap between classical software engineering (e.g., formal methods, static analysis) and AI/ML-specific challenges (e.g., explainability, adversarial robustness).
• Industry-ready solutions—tools, frameworks, and best practices for deploying high-assurance AI/ML systems.
• Regulatory and ethical dimensions, including compliance with emerging standards (e.g., EU AI Act, ISO/IEC 23053).

Audience Engagement Hook

We will debate contentious questions:

• Can we ever "prove" an AI system is safe?
• Who bears responsibility when learning systems fail?
• And how can academia and industry collaborate to turn theory into practice?

Moderator

Andrea Bondavalli Italy

University of Florence

---

Panelists

Professor Henrique Madeira Portuga

University of Coimbra

---

Professor Franz Wotawa Austria

Graz University of Technology

---

Professor Zhi Jin China

Peking University

---

Professor Yongwang Zhao China

Zhejiang University

---

Dr. Fei Shen China

Basic Cloud Platform Development Department, Terminal Cloud,
Huawei Technologies Co., Ltd.

---