Original notes
16 files this room was made from.
The notes this room was made from.
Chapter 1: Essential Background of AI Chapter Summary This chapter establishes the foundational context for AI in business. It frames AI as a productivity tool analogous to power tools—immensely capable but requiring skill and caution. It traces AI's origins from Alan Turing's 1…
Chapter 2: Deep Learning and How It Works Chapter Summary This chapter provides a technical but accessible explanation of deep learning mechanics. Starting from the perceptron—the basic computational unit—it builds up to full artificial neural networks with hidden layers, explai…
Chapter 3: Useful Types of AI Systems Chapter Summary An LLM is not an AI system—it is one component. This chapter surveys the major system architectures that make LLMs practically useful: the role of APIs and system prompts, Retrieval Augmented Generation (RAG) for knowledge gr…
Chapter 4: Proprietary vs. Open Source Models Chapter Summary AI models can be distributed as proprietary (closed source, accessed via API or user interface) or open source (publicly available code and/or weights). This chapter explains how software is written, compiled, and dis…
Chapter 5: LLMs in Various Sizes Chapter Summary Model size—measured in parameters and context window size—profoundly affects capability, cost, speed, and deployment options. This chapter traces the growth from GPT-1 (117M parameters) to GPT-5 (rumored 2–5T parameters), while sh…
Chapter 6: Assessing AI System Performance Chapter Summary Selecting the right LLM requires systematically assessing performance across four dimensions: accuracy, efficiency, scalability, and usability. This chapter surveys quantitative benchmarks (MMLU, Humanity's Last Exam, LM…
Chapter 7: Hosting AI System Components Chapter Summary AI system components can be hosted locally (edge device, self-owned server) or remotely (cloud provider, SaaS). The key architecture is client-server: a thin/thick client interacts with a back-end server via APIs. Cloud pro…
Chapter 8: AI Tool Development Chapter Summary This chapter surveys the full spectrum of AI tool development approaches, from zero-effort SaaS solutions to building custom LLMs from scratch. The central analogy is landscaping: some tasks are best outsourced, others are DIY. Key …
Chapter 8A: AI Tool Development Frameworks Chapter Summary This supplement digs deeper into the landscape of no-code and vibe-coding frameworks. It distinguishes prompt-based frameworks (single system prompt controls the chatbot) from block-based frameworks (multiple connected p…
Chapter 9: Prompt Engineering Chapter Summary Prompt engineering is the practice of crafting LLM inputs to guide desired outputs. Because LLMs are probabilistic (stochastic) rather than deterministic, and exhibit emergent behavior not fully understood even by their creators, the…
Chapter 9A: Prompt (Context) Caching Chapter Summary Prompt caching (also called context caching or cache-augmented generation / CAG) is an optimization that preprocesses a large context document through the LLM — tokenizing, embedding, positionally encoding, and running attenti…
Chapter 10: Multimodal AI Models Chapter Summary Multimodal AI extends beyond text to handle images, audio, and video as both inputs and outputs. The chapter traces multimodal origins from Frank Rosenblatt's Mark I Perceptron (1956) through GAN image generation (2014) to modern …
Chapter 11: AI Safety and Security Chapter Summary AI introduces both traditional and novel security risks. This chapter covers protecting intellectual property (patents vs. trade secrets), preventing IP leakage through encryption and access control, the threat of deep fakes for…
Chapter 12: Legal Issues of AI Chapter Summary AI creates legal risks across three domains: privacy, intellectual property (copyright and patents), and liability for AI-generated decisions. Copyright law protects human creative expression but does not currently extend to AI-gene…
Chapter 13: Ethical Issues of AI Chapter Summary AI ethics centers on four pillars: truthfulness, fairness, accountability, and transparency. LLM hallucinations threaten truthfulness; training data biases threaten fairness in high-stakes decisions (hiring, lending, sentencing); …
Chapter 14: Societal Impact of AI Chapter Summary AI's societal impact spans job automation (with potential for long-tail income inequality and UBI debates), personalization algorithms (echo chambers, addiction risk), AI in education (personalized learning vs. reduced human inte…