Scott and Mark Learn to Code
summary
Builds a multiplayer Asteroids prototype end-to-end using GitHub Copilot chat, ChatGPT and DALL·E for assets, SignalR and C# for the server, and JavaScript Canvas on the client, then deploys to Azure — it actually runs. Shows Copilot generating class bodies, tests and render logic but producing inconsistent variants (asks for prompt engineering), explains tokens, attention, GPT‑3 at 175B vs Megatron‑Turing 530B, temperature and context window limits, and notes LoRA fine‑tuning for efficient customization.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:05Multiplayer Asteroids project
- 3:36Game graphics with DALL-E
- 7:15Foundational models and neural networks
- 16:05Game classes with GitHub Copilot
- 26:41Transformer models and tokenization
- 36:26Temperature in AI responses
- 38:11Code fixes with Copilot
- 44:55Frontend and backend with SignalR
- 52:07Debugging Asteroids game
- 54:13Fine-tuning models with LoRA
- 57:25Deploying to Azure
- 59:10Key takeaways and AI future
related talks
46mCOPilotING My Autism in Tech
45mSupercharge your coding: Integrating ChatGPT and Copilot in long-term projects
42mPrompt. Build. Repeat. – The Vibe Coding Mindset
01h 01mUnlocking the Potential of AI
46mSupercharging game dev with Microsoft's cloud native and AI technology
57mUnlocking Success: Navigating the Code of Life with GitHub CoPilot
28mBoost your development workflow with AI
25mChatGPT Unleashed: From Code Wizards To Silly Shenanigans!