The second part of the course Introduction to AI TDDC65 Artificial intelligence and Lisp Peter Dalenius petda@ida.liu.se Department of Computer and Information Science Linköping University What is intelligence? Intelligence is a very general mental capability that, among other things, involves the ability to – – – – – – reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. Using the functional Lisp programming language, we will continue to study some basic concepts and techniques of artificial intelligence. A series of lectures and laboratory exercises. Russell & Norvig (2003) Artificial Intelligence: A Modern Approach Final written exam December 20, 2008. What is intelligence? It is only a word that people use to name those unknown processes with which our brains solve problems we call hard. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings – “catching on”, “making sense” of things, or “figuring out” what to do. Gottfredson, Linda (1997) Mainstream Science on Intelligence. Intelligence 24:1 Quote attributed to Marvin Minsky, one of the first researchers in AI. What is artificial intelligence? Different approaches to AI Artifical Which concept is most important? intelligence tries to understand how we think in order to build intelligent entities (e.g. machines or computer programs). AI is a new science, only 50 years old, but is heavily influenced by other fields. What do we measure success against? Humans Rationality Systems that Systems that Thought think like humans think rationally processes Behaviour Systems that act like humans Systems that act rationally 1 1. Systems that think like humans 2. Systems that act like humans But how do humans think?! Computer models from artificial intelligens Can machines think? Experimental methods from psychology Cognitive science Can machines pass a behavioral intelligence test? The Turing test Trying to build models and theories about how humans perceive and think. Turing, Alan (1950) Computing Machinery and Intelligence The Turing test (one variety) Part 1: Two persons. A B ? Part 2: A is replaced by a computer program. A B ? Skills needed to pass the test Natural language processing Knowledge representation Automated reasoning Machine learning Goal: Try to guess which one is male/female. A helps the interrogator to make the wrong decision, B helps making the right decision. 3. Systems that think rationally Aristotle’s laws of thought: Syllogisms Formalized patterns of reasoning: Logic Any solveable problem expressed in logic can be solved! Problems with this approach: – Difficult to express informal knowledge using logic – Computational explosion when trying to draw conclusions from a large knowledge base We will look at some of these in the course… 4. Systems that act rationally Rational agents – Something or someone that acts (a machine or a computer program) – Autonomous, perceiving the environment, persisting over time, adapting to change, pursuing goals Advantages of this approach: – More general than logic approach. There is more to rationality than just correct reasoning. – A completely rational agent has all the skills needed to pass a Turing test. – Easier to develop systems when you can define a degree of rationality for your specific project. 2 Influence from other disciplines An agent perceives its environment through sensors and execute actions through its actuators. The input at any given time is called percept, and the complete history of all percepts is called a percept sequence. The agent is controlled by the agent program, which implements a function from input to output. Philosophy Mathematics Economics Neuroscience Psychology Computer engineering Control theory and cybernetics Linguistics Intelligent agents Agent Intelligent agents What is rational? Environment Performance Sensors measure – Predefined way of telling how good I am. Percepts Prior knowledge of the environment – What do I know about the world? Agent program Set of possible actions – What can I do? Actuators Percept Actions sequence – What has happened so far? Specifying the task environment Properties of task environments Agent Performance measure Environment Actuators Sensors Fully Taxi driver Safe, fast, legal, comfortable Roads, other traffic, customers Steering, accelerator, brake, signal Camera, speedometer, GPS Part-picking robot Percentage of parts in correct bin Conveyor belt with parts, bins Jointed arm and hand Camera, joint angle sensors Interactive tutor Maximize Set of student’s score students, on test testing agency Display exercises, suggestions Keyboard entry or partially observable Deterministic or stochastic Episodic or sequential Static or dynamic Discrete or continuous Single agent or multiagent PEAS = Performance, Environment, Actuators, Sensors 3 Agent architectures Simple reflex agent Agent Simple Sensors What the world is like now Environment reflex agent Model-based reflex agent Goal-based agent Utility-based agent Learning agents What action I should do now Condition-action rules Actuators Model-based reflex agent Agent Goal-based agent Agent Sensors Sensors State State What my actions do Condition-action rules What the world is like now How the world evolves What my actions do What action I should do now What it will be like if i do action A Goals Actuators Learning agent Agent Sensors Sensors What it will be like if i do action A How happy I will be in such a state What action I should do now Actuators feedback changes Learning element Performance element knowledge learning goals Environment Utility What the world is like now Environment What my actions do Performance standard Critic State How the world evolves What action I should do now Actuators Utility-based agent Agent Environment What the world is like now Environment How the world evolves Problem generator Actuators 4 WITAS UAV Project Architecture 5