WEBINAR "Trust What You Can Trace: Making Agentic AI Explainable"
Join this event
About this event
Pre-registration is REQUIRED. RSVP here - https://hubs.li/Q046VS--0
Speaker: Michael Novack, Solutions Architect at AIceberg
Topic title: "Trust What You Can Trace: Making Agentic AI Explainable, Secure, and Enterprise-Ready"
As enterprises adopt increasingly powerful AI systems, one critical challenge remains: understanding how and why these systems make decisions. Many modern models - especially large language models - operate as opaque “black boxes,” making transparency, compliance, and trust difficult to achieve.
This webinar introduces the AI Explainability Scorecard, a practical framework for evaluating the transparency of AI systems based on criteria such as faithfulness, consistency, accessibility, and comprehensibility. Attendees will learn how different model architectures - from K-Nearest Neighbors to neural networks and transformers - vary dramatically in explainability and risk.
The session will also explore real-world techniques for improving visibility into complex models, including surrogate monitoring approaches that map AI behavior to comparable examples. These methods help organizations build observable, auditable AI systems without sacrificing performance or scalability.
By the end of this session, attendees will understand how to move beyond black-box AI toward transparent, accountable, and secure AI deployment - a critical step for enterprises scaling agentic AI in high-stakes environments.
You will learn:
- Why Explainability Is the Foundation of Trustworthy AI
Learn why transparency isn’t just a technical preference - it’s becoming a legal, ethical, and operational requirement for enterprise AI systems.
- The AI Explainability Scorecard Framework
Understand the five key criteria - faithfulness, comprehensibility, consistency, accessibility, and optimization clarity - for evaluating how explainable an AI model truly is.
- How Different AI Models Compare in Transparency
Discover why some models are inherently interpretable while others require advanced methods to understand their behavior.
- Practical Methods for Making Black-Box AI Observable Explore modern techniques - including surrogate monitoring models - that help organizations understand and audit large language models at scale.
🔗 Stay Connected
- Learn: Free talks and trainings on the Ai+ Training platform.
- Read: The latest in AI research on the ODSC Blog.
- Chat: Join the conversation in our Slack Channel.
- Community: Review our Code of Conduct.
Want to find AI jobs?
Join thousands of AI professionals finding their next opportunity