Exam 1 Notes.docx - Lesson 1 Intro to KBAI u25cf u25cf...

This preview shows page 1 - 4 out of 52 pages.

Lesson 1: Intro to KBAI Conundrums in and characteristics of AI KBAI and the 4 schools of AI Cognitive Systems Topics in AI Conundrums in AI Intelligent agents have limited computational resources. How can we get performance? Computation is local, but most AI problems have global constraints Logic is fundamentally deductive, but many problems are not. The world is dynamic, but knowledge is limited. How can an AI agent address a new problem? Problem solving, reasoning, and learning are complex, but explanation and justification are even more complex. Characteristics of AI Problems Knowledge often arrives incrementally, not all at once. Problems exhibit recurring patterns. Problems have multiple levels of granularity. Many problems are computationally intractable. The world is dynamic, but knowledge of the world is static. World is open-ended but knowledge is limited. Characteristics of AI Agents Agents have limited computing power. Agents have limited sensors. Agents have limited attention. Can’t focus on everything at once. Computational logic is fundamentally deductive. AI agents’ knowledge is incomplete relative to the world. How does Watson play Jeopardy? Read the clue Search through knowledge base Decide answer Phrase answer in form of a question What is KBAI?
Reasoning: understand natural language, generate national language, make decisions Learning: gets the right answer sometimes, stores it, makes note of mistakes Memory: If you’re going to learn, resulting knowledge has to be stored somewhere. If you’re going to reason using knowledge, it has to be accessed from somewhere. As a group, called “Deliberation Process”. One part of the overall architecture of an AI agent. Four Schools of AI Agents that think optimally (machine learning, Google Maps) Agents that act optimally (airplane autopilot, Roomba) Agents that think like humans (semantic web, Siri) Agents that act like humans (improvisational robots, C-3PO) Course materials: AI: A modern Approach Ch. 1 Section 1. What are Cognitive Systems? Cognitive: dealing with human-like intelligents Systems: multiple interactive components like learning, reasoning, memory Cognitive systems: systems that exhibit human-like intelligence through processes like learning, reasoning, and memory.
Cognitive System Architecture World -> Input -> Cognitive System -> Output -> World So far we have talked about a single cognitive system, but there can be multiple systems that interact with one another. Inside of a cognitive system, takes input information about the world, returns output as actions on the world. Reaction (no planning, direct mapping of input to action, e.g. slamming on brakes once lights of car in front of you turn red) Deliberation (come up with a plan to determine action to take, using reasoning, learning, and memory, e.g. determining whether to change lanes) Metacognition (reasoning about the internal mental world, reasoning about reaction or deliberation, e.g. changing lanes at the wrong time causing cars to honk - may reconsider

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture