AI Salon Zurich x Gian Klain x CAIF - AI and Trust
AI Salon Zurich x Gian Klain x CAIF
AI and Trust — How Can We Trust the System?
As AI systems increasingly operate in high-stakes environments – shaping decisions, generating knowledge, and acting with growing autonomy – the question is no longer what AI can do, but whether and how it can be trusted.
This second AI Salon Zurich invites a focused, interdisciplinary conversation on AI and Trust, moving beyond surface-level discussions toward the deeper technical, societal, and epistemic challenges of trusting intelligent systems.
Trust in AI is not a given. It is constructed and often contested across multiple layers:
the reliability and robustness of models under real-world conditions
the security of systems against manipulation, adversarial attacks, and misuse
the interpretability of outputs and the limits of explainability
the alignment between system behavior and human values
the governance structures that define accountability and responsibility
Rather than treating trust as a vague aspiration, this salon asks more precise questions:
Han we meaningfully trust AI systems, or only calibrate risk?
What does “trustworthy AI” actually mean in adversarial or safety-critical contexts?
Where do technical approaches (e.g. alignment, evals, red-teaming, interpretability) succeed – and where do they fail?
How do we distinguish between perceived trust and justified trust?
What are the attack surfaces, e.g. technical, social, and institutional, that undermine trust in AI systems?
Who is accountable when AI systems fail, mislead, or are exploited?
We are particularly interested in perspectives from those working in:
AI safety, security engineering, cybersecurity, ML research, governance, and related fields alongside voices from design, philosophy, and the social sciences.
This salon is not about abstract ethics alone. It is about the practical and philosophical limits of trusting complex, probabilistic systems embedded in real-world power structures.
Format
Rather than panels or presentations, the AI Salon is a moderated, closed-door dialogue designed for depth, candor, and exchange:
8–16 carefully curated participants
Moderated group conversation
No presentations, no pitching
Emphasis on listening, challenge, and multi-perspective thinking
Informal networking afterward
Participation & Registration
This salon is by application and invitation only to maintain a high-trust environment and diversity of expertise.
If you’re interested in joining, please register and briefly share:
your background and relationship to AI
why this question of trust in AI matters to your work or thinking
We will follow up with confirmations.
About AI Salon Zurich
AI Salon is a global, community-driven format bringing together thinkers and practitioners across disciplines to explore the cultural, technical, and societal implications of artificial intelligence.
The Zurich chapter is led by Pascale Speck and, for this session, hosted in collaboration with Gian Klain and the Creative AI Foundation (CAIF), connecting Switzerland’s AI ecosystem with critical, interdisciplinary perspectives.