AI Lecture 1 PDF

Introduction to Artificial Intelligence Lecture 1 Presented By Dr. Mohamed Awni  Outline ❑What is Artificial Intelligence ❑Agents and environments ❑Rationality ❑PEAS (Performance measure, Environment, Actuators, Sensors) ❑Environme...

Introduction to Artificial Intelligence Lecture 1 Presented By Dr. Mohamed Awni  Outline ❑What is Artificial Intelligence ❑Agents and environments ❑Rationality ❑PEAS (Performance measure, Environment, Actuators, Sensors) ❑Environment types  Human Intelligence (Golden Standard) Intelligence is one of the defining features of being human and it comes in various forms. Verbal-Linguistic Intelligence Ability to generate and comprehend language in the forms of reading writing, spoken Spatial Intelligence Ability to observe the world with the mind’s eye. Logical – Mathematical Intelligence Ability to solve mathematical problems. Emotional intelligence Ability to identify and manage your own emotions and the emotions of others  Animal Intelligence Combination of skills and abilities that allow animals to live in and adapt to their specific environments. Octopuses can carry out complex tasks: ❑Opening a jar to get to its contents. ❑Have a good short- long term memory ❑A remarkable ability to learn new skills from the moment they’re born. Dolphins: ❑ Extremely sociable creatures. ❑ Highly developed ability to adapt to their habitat. ❑ Help one another when injured ❑ Able to pass on their knowledge to others. Elephants ❑ Largest brain of all land animals. ❑ Can express a wide range of emotions, including happiness ❑ Greater memory than even us humans, an elephant truly never forgets!  What is Intelligence? “Judgment, otherwise called “good sense,” “practical sense,” “initiative,” the faculty of adapting one's self to circumstances.. auto-critique “ Alfred Binet (July 8, 1857 – October 18, 1911) was a French psychologist who invented the first practical intelligence test (An intelligence quotient (IQ); a total score derived from one of several standardized tests designed to assess human intelligence) ❑Judgment: Ability to make good decisions. It’s about evaluating situations carefully and coming to sensible conclusions based on the available information. ❑Good sense or practical sense: Ability to apply common sense and practical knowledge to everyday problems. It's not just about knowing facts, but also knowing how to use that knowledge effectively in real-world situations. ❑Initiative: Ability to take action on your own without waiting for instructions. Someone with initiative can recognize what needs to be done and take steps toward it, even in unfamiliar situations. ❑Adapting to circumstances: Being flexible and able to adjust your behavior or approach when situations change.  Artificial Intelligence AI is a machine that able to: Think Understand Languages Solve Problems Play chess Paint images Drive cars AI is as a computer system with the ability to perform tasks commonly associated with often defined human intelligent beings. Some foundations of artificial intelligence  What is Artificial Intelligence ❑ There are hundreds of definitions of artificial intelligence. ❑ Most contain a bias as to whether the writer of the definition sees AI as: ▪ Dealing with thinking versus acting. ▪ Trying to model humans or capturing intelligence (rationality) A system is rational if it does the "right thing" given what it knows ❑ The study of agents that receive percepts from the environment and perform actions. (Russell and Norvig)  Thinking Vs. Acting Thinking Acting Process of generating internal Acting in AI refers to the process of representations, reasoning, planning, and executing actions or behaviors in decision-making based on available response to the environment or a given information. situation. It involves mental processes such as It involves translating the results of problem-solving, logical inference, cognitive processes (thinking) into abstraction, and learning from past concrete actions that interact with the experiences. external world. Thinking in AI often involves symbolic Acting in AI may include tasks such as manipulation, probabilistic reasoning, moving a robot, making knowledge representation, and search recommendations to users, controlling a algorithms to arrive at solutions or make game character, or responding to natural predictions. language queries.  Humanly Vs. Rationally Humanly Rationally AI system mimics or replicates human Rationality, on the other hand, refers to cognitive abilities, behaviors, and thought the ability of an AI system to make processes. decisions and take actions that are optimal or "rational" according to a specified Evaluating AI systems based on humanly utility function or goal. criteria involves comparing their Evaluating AI systems based on performance to that of humans on tasks rationality involves assessing their ability such as perception, language to achieve desired outcomes or objectives understanding, problem-solving, given their available knowledge, creativity, and social interaction. resources, and constraints.  Definitions may be organized into four categories. Systems that think like humans. Systems that act like humans. Systems that think rationally. Systems that act rationally. What is Artificial Intelligence  Think Humanly Cognitive modeling approach ❑ Means thinking in a way that is similar to how humans think. This includes using human-like reasoning, problem-solving, and learning strategies. ❑ Concerned with the internal processes of the mind We can try to gain insights about the mechanisms and patterns in the human mind by: Psychological experiments. (Observe a person on the action.) Introspection (Catch our thoughts and see how it flows.) Brain imaging. (MRI or fMRI scanning, Observe a person’s brain in action.) If we understand how the human brain works, we can simulate or rebuild it.  Think Humanly Cognitive modeling approach(cont.) ❑ If we are able to catch the human brain’s actions and give it as a theory, then we can convert that theory into a computer program. ❑ If the input/output of the computer program matches with human behavior, then it may be possible that a part of the program may be behaving like a human brain. Artificial neurons and perceptron were inspired by the biological process's scientists were able to observe in the brain back in the 50s What is a Thought? How the Brain Creates New Ideas | Henning Beck | TEDxHHL https://www.youtube.com/watch?v=oJfFMoAgbv8&t=856s  Act Humanly(Turing Test approach) Turing test (1950): Can a human interrogator tell whether (written) responses to her (written) questions come from a human or a machine? Turing Test Approach To pass Turing test, computer needs: Natural language processing Knowledge representation Automated reasoning Machine learning  Act Humanly (The Chinese Room Argument) If person inside does a great job of answering questions, can we say s/he understands? Even if (s)he is only blindly following rules? (Obviously, the ‘person inside’ is acting like an AI program) ❑ The question Searle wants to answer is this: does the machine literally "understand" Chinese? Or is it merely simulating the ability to understand Chinese? ❑ Searle calls the first position "strong AI" and the latter "weak AI".  Act Humanly (The Chinese Room Argument) Strong vs. Weak AI Hypotheses? WEAK AI hypothesis; We can accurately simulate animal / human intelligence in a computer. STRONG AI Hypothesis; We can create algorithms that are intelligent ( Consciousness ?..Self-Awareness ?.. Free-will ? ) I Robot  Thinking humanly Vs. Acting humanly: Thinking humanly: Acting humanly: ❑Understanding and responding to natural ❑Exhibit (show) human-like emotions, such as language happiness, sadness, anger, and fear ❑Reasoning about the world and making ❑Engage in human-like social behavior, such as decisions based on that understanding conversation, cooperation, and competition ❑Learning and adapting to new information and ❑Performing human-like physical movements, such as experiences walking, talking, and manipulating objects ❑Planning and problem solving ❑Being creative and imaginative Thinking humanly and acting humanly are not mutually exclusive  Think Rationally (Laws Of Though approach) ❑ Means using logic and evidence to form conclusions. ❑ It involves being able to identify and avoid biases, and to consider all of the relevant information before making a decision Socrates is a man ❑ Uses logic to reach a decision given some facts via logical inferences. All men are mortal ------------------------ ❑ Greek philosopher Aristotle was the first to try to codify the way of thinking. Therefore, Socrates is mortal ❑ His deductive reasoning always gave correct conclusions when given correct premises. Limitations: Not all knowledge can be expressed with logical notation, Could be computationally expensive if there are too many premises.  Act Rationally ❑ Means making decisions that are consistent with your goals and beliefs. ❑ It involves weighting the pros and cons of each option and choosing the course of action that is most likely to achieve your desired outcome. ❑ Rational Agent Approach Perceive the environment, and act so as to achieve one’s goal. Not necessary to do the best action: There is not always an absolutely best action. There is not always time to find the best action. An action that’s good enough can be acceptable. Example: Game playing. Sample approach: Tree searching strategies. Problem: Choosing what to do given the constraints. Ch.(2): Intelligent Agents  Agents and environments AI definition :The study of agents that receive percepts from the environment and perform actions. Agent Environment Sensors Percepts ? Actuators Actions An agent perceives its environment through sensors and acts upon it through actuators (or effectors, depending on whom you ask)  Agents and environments Agent Environment Sensors Percepts ? Actuators Actions ▪ Are humans agents? ▪ Yes! ▪ Sensors = vision, audio, touch, smell, taste, proprioception ▪ Actuators = muscles, secretions, changing brain state  Agents and environments Agent Environment Sensors Percepts ? Actuators Actions ▪ Are pocket calculator agent? ▪ Yes! ▪ Sensors = key state sensors ▪ Actuators = digit display  Agents and environments Agent Environment Sensors Percepts ? Actuators Actions ▪ AI is more interested in agents with large computational resources and environments that require nontrivial decision making  Agent functions ▪ The agent function maps from percept histories to actions: ▪ Defines the relationship between the agent's inputs (percepts) and outputs (actions) without specifying the internal mechanisms or implementation details. ▪ f : P* → A ▪ I.e., the agent’s actual response to any sequence of percepts NEXT NEXT NEXT NEXT Percept Action LEFT LEFT DROP RIGHT  Agent programs ❑ The implementation or the architecture of the agent, refers to the concrete algorithm or code that implements the agent function. It represents the internal mechanisms and decision-making processes of the agent, including data structures, control logic, and algorithms. The agent program translates the abstract agent function into executable instructions that the agent can follow to perceive its environment and select appropriate actions. ▪ The agent program l runs on some machine M to implement f : ▪ f = Agent( l, M) ▪ Real machines have limited speed and memory, introducing delay, so agent function f depends on M as well as l  Example: Vacuum world The robot has three sensors: ❑ A location sensor that tells the robot whether it is on a clean or dirty square ❑ A left sensor that tells the robot whether there is a wall to the left A B ❑ A right sensor that tells the robot whether there is a wall to the right. The robot can also perform three actions: move left, move right, and suck. The goal of the robot is to find a sequence of actions that will clean all of the dirty squares and bring it back to its starting position. ▪ Percepts: [location, status], e.g., [A, Dirty] ▪ Actions: Left, Right, Suck  Solve the vacuum world problem using rules Agent function Agent program Percept sequence Action function Reflex-Vacuum-Agent([location, status]) [A, Clean] Right returns an action if status = Dirty then return Suck [A, Dirty] Suck else if location = A then return Right [B, Clean] Left else if location = B then return Left [B, Dirty] Suck [A, Clean],[B, Clean] Left [A, Clean],[B, Dirty] Suck What is the right agent function? etc etc Can it be implemented by a small agent program? (Can we ask, “What is the right agent program?”)  Solve the vacuum world problem using a search algorithm 1. Define the state space. The state space of the vacuum world problem consists of all of the possible configurations of the robot and the dirty squares. 2. Define the goal state. The goal state is the state in which all of the dirty squares are clean, and the robot is back at its starting position. 3. Define the successor function. The successor function takes a state as input and returns all of the possible next states that can be reached from that state. 4. Define the cost function. The cost function takes a state transition as input and returns the cost of moving from the current state to the next state. 5. Run the search algorithm. The search algorithm starts from the initial state and explores the state space until it finds the goal state or determines that the goal state cannot be reached.  A B Rational Agent ❑ In artificial intelligence, the central problem is creation of a rational agent, ❑ A rational agent: Entity that has goals and tries to perform a series of actions that yield the best/optimal expected outcome given these goals (Do Right Things). ❑ In order to know the right thing, we need a performance measure. ❑ The performance measure is usually chosen by the agent designer. ❑ For each possible percept sequence, a rational agent should select an action that is expected to maximize the performance measure given the evidence provided by the percept sequence and whatever built-in knowledge the agent has. ❑ Design of a rational agent depends on the environment it operates in.  Rationality, cont. ▪ Are rational agents omniscient ( known all things) ▪ No – they are limited by the available percepts ▪ Do rational agents explore and learn? ▪ Yes – in unknown environments these are essential ▪ Are rational agents autonomous (i.e., transcend initial program)? ▪ Yes – as they learn, their behavior depends more on their own experience An agent in Pacman  The task environment - PEAS ▪ Framework for describing the task environment of an intelligent agent. ▪ Valuable tool for designing and evaluating intelligent agents. It can help to ensure that the agent is able to perform its task effectively in the real world. ▪ Performance measure :Defines how well the agent is performing its task ▪ -1 per step; + 10 food; +500 win; -500 die; +200 hit scared ghost ▪ Environment : World in which the agent operates. It includes all of the objects and other agents that the agent can interact with. ▪ Pacman dynamics (include ghost behavior) ▪ Actuators :Physical devices that the agent uses to interact with the environment ▪ Left Right Up Down ▪ Sensors :Devices that the agent uses to perceive the environment ▪ Entire state is visible (except power pellet duration)26  PEAS: Automated taxi ▪ Performance measure ▪ Income, happy customer, vehicle costs, fines, safe, fast, legal ▪ Environment ▪ US streets, other drivers, customers, weather, police… ▪ Actuators ▪ Steering, brake, gas, display/speaker ▪ Sensors ▪ Camera, radar, accelerometer, engine Image: http://nypost.com/2014/06/21/how-google- sensors, microphone, GPS might-put-taxi-drivers-out-of-business/  Environment types ❑ The design of an agent heavily depends on the type of environment the agents acts upon. ❑ We can characterize the types of environments in the following ways. ❑ Fully observable ( vs. partially observable ) ❑ Deterministic ( vs. stochastic / strategic: ) ❑ Episodic ( vs. sequential ) ❑ Static ( vs. dynamic) ❑ Discrete ( vs. continuous ) ❑ Single agent ( vs. MultiAgent.. cooperative / competitive )  Environment types Partially observable environments: the agent does not have full information about the state and thus the agent must have an internal estimate of the state of the world. Fully observable environments: the agent has full information about their state. Stochastic environments: have uncertainty in the transition model. Deterministic environments: taking an action in a state has a single outcome that is guaranteed to happen. Multi-agent environments the agent acts in the environments along with other agents. For this reason the agent might need to randomize its actions in order to avoid being “predictable" by other agents. Static environments: If the environment does not change as the agent acts on it. Dynamic environments: Changes as the agent interacts with it. Environment types  Ref: https://www.zooportraits.com/animal-intelligence-smartest-animal-species/ https://www.gopichandrakesan.com/thinking-humanly-the-cognitive-modeling-approach-artificial- intelligence/ https://towardsdatascience.com/the-differences-between-artificial-and-biological-neural-networks- a8b46db828b7 CS 188 Fall 2022 | Introduction to Artificial Intelligence at UC Berkeley