TL;DR
- A curated set of the clearest technical explainers of foundational AI concepts - the ones that build real intuition, not just vocabulary
- Covers fundamentals (backprop, neural networks), transformers and language models, and generative models like diffusion
- Authors include Andrej Karpathy, Stephen Wolfram, and NVIDIA’s developer team - high signal, low fluff
- Read one piece slowly rather than skimming five - the value is in working through the maths, not collecting links
- Pairs well with the courses list if you want a structured path after the explainers click
A curated collection of clear, technical explanations of foundational AI concepts. These resources help build intuition about how modern AI systems actually work.
Fundamentals
- Yes you should understand backprop - Andrej Karpathy’s definitive explanation of backpropagation, the fundamental algorithm behind neural network training
- Deep Learning in a Nutshell: Core Concepts - NVIDIA’s accessible overview of deep learning architectures and their applications
Transformers & Language Models
- What is ChatGPT doing & why does it work? - Stephen Wolfram’s phenomenal breakdown of transformer architecture and the surprising effectiveness of next-token prediction
- Word2Vec Explained - Foundation for understanding how words become numerical representations that models can process
Generative Models
- How Stable Diffusion Works - Detailed technical walkthrough of diffusion models for image generation, with clear diagrams and intuitive explanations
Courses & Practical Learning
- Practical Deep Learning - Fast.ai’s top-down course that teaches you to build working deep learning systems before diving into theory