Paige Bailey, ETLS Vegas (06)
summary
Google describes a code AI suite built on over a trillion tokens and 500,000 aggregate years of engineering telemetry and 80 million code review edits powering Gemini and AI Studio features. It claims 26% of Google code is ML generated and weekly peaks near 50% checked in, automates code review edits, performs large codebase migrations (example: Flax 0.7.5 → 0.8.5 upgrade in 30 seconds), spots inefficiencies that save tens to hundreds of millions, and generates friction logs from user videos.
AI-written summary · about
related talks
11mapproach to generative AI
24mHow to create and scale AI Products, MLOPs and Generative AI
45mFrom Buzz to Benefits: Google AI in Action
32mFrom Vibes to Outages: Riding the AI Code Wave
35mBeyond the Hype and Hyperbole
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
38mUnlocking the Future: GenAI, DevOps, and Fintech Innovation
19mNow in AI - How We Got Here
