Top-of-mind on AI with Mark Russinovich
summary
AI capability follows an exponential curve with ChatGPT (Nov 2022) as a visible inflection; Azure highlights concrete defenses and infrastructure: Prompt Shields and layered content safety that detect indirect prompt injection (hidden white‑text example that exfiltrates pricing), dataset hygiene and model fine‑tuning to reduce hallucinations, Llama‑2‑7b unlearning of Harry Potter and Mistral‑7B trained to avoid profanity, and infrastructure projects—Project Forge pooling global GPU capacity and Flywheel time‑slicing hundreds of sessions per GPU with guaranteed throughput.
AI-written summary · about
chapters (6)
- 0:00The evolution of AI
- 2:54Safety and enterprise AI
- 5:04Security risks and mitigation
- 9:04AI alignment research
- 12:20Infrastructure innovation
- 15:50The future of agentic AI
related talks
43mBeyond the AI Hype: What's Real, What's Next
53mHow AI is impacting cybersecurity. Does it?
48mBuilding and using AI models responsibly
41mThe AI boom: what and why?
59mBuild Next-Gen AI Apps with .NET and Azure
01h 03mIt's All About AI
47mAI and the Future of Software Development
18mDefending the unknown using AI and cybersecurity