BRK275 Breakout Wed 18 Nov 09:00 PST
What should you build when you can build anything?
Scott Hanselman, Katie Savage, Anthony Shaw
Summary
From Microsoft's AI summary
Scott Hanselman opened a session on what to build with AI-assisted coding, arguing that human taste and expertise lift results above average output. Anthony Shaw demonstrated OpenRCT2 experiments, using local small language models for guest inner monologues and fine-tuned decision models. Katie Savage presented Tone Mirror, a web app for practising Mandarin tones that compares recordings with native references. Hanselman showed GPU Task Man, a text interface for monitoring local AI model and GPU use. Speakers encouraged using Copilot for learning and experimentation.
Session description
Microsoft's catalogue description
What do you build when the answer is “anything”? We'll take you on a fast, fun tour of Windows, GitHub, OSS and all the fun new shiny things. Expect big ideas, surprises, and a parade of demos showing what happens when powerful tools meet relentless curiosity. Come for the awesome. Leave with new possibilities and an urge to build something.