Keynote: The past, present, and future of AI for application developers
summary
Starts with a live Eliza demo showing 230 lines of string replacements, then a Markov generator trained on NDC session descriptions to produce plausible titles, and a tiny C GPT‑2 implementation (lm.c) GPT‑2 trained on NDC talks (500 iterations on a laptop) that begins to form grammar. Covers modern patterns: function calling + stop tokens to force answers, Azure AI Foundry and Microsoft.Extensions.AI examples for structured extraction, a 284 MB DTA v3x small model running zero‑shot in the browser, vision traffic‑cam alerts, and a realtime voice assistant that edits JSON-backed UI.
AI-written summary · about
chapters (16)
- 0:00AI for developers
- 1:34Developer perspective overview
- 2:17Early AI beginnings
- 8:14Markov models and language
- 13:09Language model limits and Transformers
- 15:32Training GPT-2 on NDC talks
- 21:32Token predictors to Q&A
- 27:31Modern AI APIs and extraction
- 32:28AI for business automation
- 37:25Small language models
- 41:38Vision AI for traffic
- 44:50Agentic systems and trends
- 46:35Smart forms and autocomplete
- 48:20Intelligent assistants integrated into applications
- 51:53Voice assistant demo
- 57:07Summary
related talks
59mKeynote: The past, present, and future of AI for application developers
45mKeynote: Django Reimagined For The Age of AI with Marlene Mhangami
55mEveryday AI: Practical Applications Beyond the Chatbot
43mKeynote: Applied NLP in the age of Generative AI
34m"The Past, Present, and Future of GenAI"
01h 01mKeynote: Transformers: The Rise of ChatGPT
45mBringing the Power of AI to Your Application
38mKeynote: Thriving in the Age of AI