NSA and IETF: Fairness
Summary
This article discusses the rise of small AI models (edge AI, TinyML) that run on devices with limited connectivity and power. It covers techniques such as pruning, distillation, and quantization to enable on-device AI, with examples like RxScanner and Arduino-based deployments, and notes open-weight models like Gemma 4 and Qwen 3.5. The piece argues that millions of small, edge-focused models could complement larger centralized models, while highlighting infrastructural and policy considerations.