

Microsoft is all-in on AI: Part 2 (Changelog Interviews #594)
Jun 5, 2024
Mark Russinovich, Eric Boyd, and Neha Batra discuss AI at Microsoft Build 2024, AI algorithms' challenges, using red teams to test AI systems, cybersecurity thrillers, data impact on AI training, and One Password benefits. They talk about GPT-4 development, prompt shields, hallucination detection, and choosing AI models. The podcast explores serverless Postgres, connecting with others, personal adventures at GitHub, AI-human content creation, and organizing AI events at Microsoft.
Chapters
Transcript
Episode notes
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Intro
00:00 • 3min
Discussion on Microsoft Azure and AI at Build Event
02:36 • 2min
AI Algorithms and Challenges in Minimizing Misinformation
04:10 • 6min
The Role of Red Teams in Testing AI Systems
09:56 • 14min
Exploring Cyber Security Thrillers and Silicon Valley TV Show
24:10 • 4min
Reflections on Past Work and Discussing Favorite Authors
28:10 • 3min
Challenges and Improvements in AI Models
31:35 • 8min
Exploring the Impact of Data in Training AI Models and the Evolution of AI Technology
39:13 • 14min
Exploring the Changelog Podcast
53:35 • 2min
Benefits of Using One Password for Developers
55:11 • 5min
Microsoft and OpenAI Collaboration for GPT-4 Development and Data Center Deployments
01:00:22 • 18min
The Role of Prompt Shields in Managing AI Models
01:18:20 • 6min
AI Hallucination Detection and Prevention
01:24:19 • 14min
Navigating the World of AI Models
01:38:44 • 14min
Understanding Developer Needs and Introducing Serverless Postgres with Branch Support
01:52:52 • 6min
Exploring the Value of Connecting with Others and Embracing Adventures
01:58:42 • 2min
Navigating Personal and Work Adventures at GitHub
02:00:14 • 15min
Exploring the Intersection of AI and Human Input in Content Creation
02:14:44 • 21min
Organizing an AI Summit Event at Microsoft
02:36:12 • 8min
Concluding Microsoft Build discussion and upcoming Hound Define game show
02:44:07 • 3min