Deterministic GenAI Outputs With Provenance
summary
Claims a method for deterministic GenAI outputs with provenance and shows an open source CI/CD pipeline that tags releases, builds AMIs and can run models offline (Llama/meta via OpenRouter, Bedrock). Demonstrations include a Cyber Boardroom prototype that queries multiple LLMs and applies a three-LLM consensus pattern, an editable conversation history to force reproducibility, and air‑gapped deployments. Demos work but LLMs still hallucinate; provenance comes from forcing models to only use provided source documents rather than external knowledge.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 12:38GenAI revolution misunderstood
- 23:06Live demo: deterministic GenAI
- 36:51Patterns for safe GenAI outputs
- 44:54GenAI cybersecurity future
- 51:15Call to action
related talks
19mWe Know Security but How Do We Secure GenAI End-to-End?
01h 42mBuilding Responsible AI Agents in Decentralized Open-Source Environments
50mFrom Prompt Hacking to Architectural Determinism: Engineering Reliable GenAI Systems
27mAI Demystified Unveiling the Intelligence Behind the Machine
55mAI Showdown: The Good, The Glitchy, and The Risky
33mAI sovereignty, provenance and compliance
30mDeveloper Productivity for GenAI Apps
23mTrust, Track, and Verify: Securing AI Pipelines End-to-End