Future of AI
summary
Combines static analysis (using tree-sitter) with generative models to auto-document codebases, noting evaluation and productionization as the hardest step; a practical pipeline reduces hallucinations by mixing keyword classification, AST inspection and LLM summarization. Brazil builds local models like Sabya (claimed 75% as good at 10% cost) and payment rails like PX that onboard 80% of the population to banking. Energy is highlighted: current compute uses ~1.4% of global energy, can hit ~5% soon, while targeted AI tooling (e.g., tire checks) can cut energy use by ~40%.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:55AI in technical docs
- 7:15AI innovation in Brazil
- 15:28AI energy consumption
- 21:09Open vs private AI
- 25:40AI future in 5 years
- 31:51Q&A
related talks
47mAI and the Future of Software Development
44mFuture-Proof Your...
27mSoftware engineering with GenAI
39mBack to Basics: Cutting Through the AI-Generated Code Noise
27mBuilding Generative AI features
26mAl and Software: The Good, The Bad, The Scary
49mMagic of Automation and Everyday Chores
29mImpact of #GenerativeAI in Development and #SoftwareTesting
