Learning from ChatGPT

#4488 Learning from LLM Thinking and Reasoning Reports #4488  Describe the thinking and reasoning phase of an LLM. Can the user view this information? The "thinking and reasoning" phase of a Large Language Model (LLM)—often referred to as chain-of-thought (CoT) reasoning or internal deliberation —represents a paradigm shift in how AI processes complex queries. Instead of jumping straight from a prompt to a final answer, the model generates an internal, step-by-step monologue to break down problems, check its own logic, and plan its response. Here is a look at what happens behind the scenes during this phase, and whether you, the user, can actually see it. 1. Inside the Thinking Phase: What is the LLM Doing? When a reasoning-focused LLM receives a complex prompt, it activates a structured, hidden generation cycle. This phase isn't just a static calculation; it is a dynamic text-generation process where the model talks to itself. [User Prompt] ──> [Hidden Thinking Phase...

#4361 AI Self-Learning The Man Behind AlphaGo Thinks AI Is Taking the Wrong Path https://www.wired.com/story/david-silver-ai-ineffable-intelligence-reinforcement-learning/ Evolving AI may arrive before AGI and create hard-to-control risks https://techxplore.com/news/2026-04-evolving-ai-agi-hard.html #4361  Consider https://www.wired.com/story/david-silver-ai-ineffable-intelligence-reinforcement-learning/ and the following:   """AI Self-Learning A Google DeepMind program AlphaGo taught itself to play the game of Go with a kind of mastery that went far beyond mimicry. AI internal thoughts: This is what I am trying to learn now What have I learned? What should I do with that new knowledge? What additional new knowledge would be useful?""" The Wired article explores David Silver’s departure from Google DeepMind to launch his new startup, Ineffable Intelligence , which recently raised $1.1 billion. The core of his vision is a transition from the current ...

#4338 Self-Learning Knowledge Graph Agent EXPOSED: The Dirty Little Secret of AI (On a 1979 PDP-11) https://www.youtube.com/watch?v=OUE3FSIk46g #4338  Describe a self-learning framework that creates a mind map in the form of a knowledge graph, that cycles on a task, checking its work and improving until meeting a required threshold of success, then determining the next task and continuing with that. This framework describes a Recursive Knowledge Graph Agent (RKGA) . Unlike standard linear AI workflows, this system treats learning as a closed-loop refinement process where the knowledge graph isn't just a storage unit—it’s the "brain" that evolves. 1. The Core Architecture: The " OODA " Loop for Knowledge The framework operates on a continuous cycle of observation, orientation, decision, and action , centered around a dynamic knowledge graph. Phase Action Purpose Ingestion Parse task requirements and existing graph nodes. Contextual awareness. Execution Generate a...