Week - 2.pdf

DAB106 Introduction to Artificial intelligence The Artificial Intelligence Revolution – How AI Will Change Our Lives https://www.youtube.com/watch?v=UXxyCBimRyM  What is artificial intelligence It is the science and engineering of making intelligent machines, espec...

DAB106 Introduction to Artificial intelligence The Artificial Intelligence Revolution – How AI Will Change Our Lives https://www.youtube.com/watch?v=UXxyCBimRyM  What is artificial intelligence It is the science and engineering of making intelligent machines, especially intelligent computer programs. It is related to the similar task of using computers to understand human intelligence, but AI does not have to confine itself to methods that are biologically observable. Artificial intelligence (AI), in its simplest sense, refers to the ability of a computer to perform tasks that are similar to that of human learning and decision making. The term can also refer to the study, science, and engineering of such intelligent machines, systems, and programs. A Machine with human like intelligence  Intelligence: The ability to learn and solve problems - Webster’s Dictionary Artificial intelligence (AI) is the intelligence exhibited by Artificial machines or software - Wikipedia intelligence The science and engineering of making intelligent machines -McCarthy. Definition The study and design of intelligent agents, where an intelligent agent is a system that perceives its environment and takes actions that maximize its chances of success. - Russel and Norvig AI book. What is According to the father of Artificial Intelligence, John McCarthy- who is Artificial considered the father of Artificial Intelligence, it is The science and Intelligence? engineering of making intelligent machines, especially intelligent computer programs. Artificial Intelligence is a way of making a computer, a computer-controlled robot, or a software think intelligently, in the similar manner the intelligent humans think. How do you define artificial intelligence? Link Understanding AI Defining AI: A computer system capable of performing tasks that typically require human intelligence. Implementing AI: AI works by using different technologies, such as: Machine Learning: Where computers learn from examples, like how Netflix recommends movies based on what you’ve watched. Natural Language Processing (NLP): This helps computers understand human language, like when you talk to customer support chatbots online. Robotics: AI-powered robots can perform tasks like vacuuming or even manufacturing products in a factory. Goals of AI: AI is designed to: Help humans by making their work easier, such as assisting doctors in diagnosing diseases. Automate repetitive tasks, like data entry or responding to emails. Solve difficult problems, such as predicting weather patterns or finding the best route for delivery trucks.  Understanding AI and Key Functions AI is not just automation, it’s about enabling machines to sense, comprehend, act, and learn: AI goes beyond basic automation (like a washing machine following a timer). It involves teaching machines how to: Sense the world (for example, through sensors or cameras), Understand what they sense (like recognizing a face in a photo), Act based on that understanding (such as stopping a self-driving car to avoid an accident), Learn from their experiences to get better over time (like improving speech recognition with more use). Understand - What is AI  AI: Humanizing Technology in the Fast-Paced Modern World The Evolution of Technology: Rapid Communication: Instant global connectivity through mobile devices. E-Commerce: The convenience of online shopping, bringing the marketplace to our fingertips. Interactive Tech: Conversing with our phones This Photo by Unknown Author is licensed under CC BY-SA-NC using natural language. AI's Role in Daily Life: The Vision and Mission of Modern Technology: Self-Driving Cars: Redefining transportation Creating machines that replicate human with autonomy. capabilities. Entertainment: AI-curated playlists and AI's central tenets: Reasoning, Learning, and intelligent gaming experiences. Problem Solving. Smart Interaction: Devices that understand Technology's trajectory towards seamless and respond to us. human-machine interaction. Unraveling the Essence of Artificial Intelligence (AI) Breaking Down the Terms: Artificial: Synonymous with 'man-made' or 'created by humans'. Reflects the engineered aspect of AI technology. Intelligence: The ability to acquire knowledge and apply it. Indicates the cognitive dimension of AI systems. Artificial Intelligence: Man-made systems endowed with the capability to learn, adapt, and apply knowledge. AI represents the fusion of computational structures with human-like cognitive abilities. Core Concept: AI is essentially a knowledge system engineered by humans, equipped to perform tasks that would typically require human intelligence. What Is Artificial Intelligence? https://youtu.be/kbAbh-jvpJg What is AI Intelligence The ability to acquire and apply knowledge and skills. (Oxford dictionary) Artificial Intelligence The ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. (Britannica.com) The science and engineering of making intelligent machines. (John McCarthy)  What is Intelligence? Oxford Dictionary Definition: The ability to acquire and apply knowledge and skills. Expanding on Intelligence: Intelligence encapsulates: Rational thought Purposeful action Effective environmental interaction Implications for AI: In the context of AI, this translates to: Machine learning from data and experience Autonomous decision-making Adaptive responses to situational changes  The Multifaceted Nature of Intelligence Defining Intelligence: People define intelligence in many different ways. However, you can say that intelligence involves certain mental activities composed of the following activities: Learning: This is about acquiring and processing new information. Reasoning: This means manipulating information in different ways to solve problems or make decisions Understanding: After gathering information, this is about assessing what you’ve learned and figuring out how things work. Grasping Truths: This is about evaluating whether something is true or not. Seeing Relationships: Intelligence also involves understanding how things relate to each other. Considering Meanings: This means applying truths in the right context. Distinguishing Facts from Beliefs: intelligence means being able to separate what’s true from what’s just an opinion.  Type of Intelligence According to researchers, intelligence is the ‘ability to perceive or infer information, and to retain it as knowledge to be applied towards adaptive behaviors within an environment or context.’ https://blog.adioma.com/9-types-of-intelligence-infographic/  The Spectrum of Human Intelligence Intelligence is more than just book smarts; it's a tapestry of abilities that interact with our environment and each other. Types of Intelligence: 1.Linguistic Intelligence: Mastery in the use of language for expression and comprehension. 2.Mathematical/Logical Reasoning: Proficiency in numerical calculation, logical reasoning, and scientific analysis. 3.Spatial/Visual Intelligence: The ability to visualize concepts and manipulate spatial dimensions. 4.Musical Intelligence: Sensitivity to sounds, rhythms, tones, and music creation or performance. 5.Kinesthetic Intelligence: Coordination and skill in physical movement and manipulation. 6.Intrapersonal Intelligence: Deep self-awareness including understanding one's own emotions and motivations. 7.Existential Intelligence: Capacity to ponder deep questions about existence and spirituality. 8.Naturalist Intelligence: Ability to identify and categorize patterns in nature. 9.Interpersonal Intelligence: Skill in understanding and interacting effectively with others.  Cognitive Strategies for Problem-Solving and Learning Problem-Solving Through Past Experiences: Recall and apply solutions from similar past situations to current problems. Anticipatory Thinking: Predict potential outcomes before taking action to better prepare for consequences. Reflective Learning: Post-failure, reflect on alternative approaches that could have led to success. Causal Analysis: Observe events and deduce the possible causes to understand the sequence of actions. Ownership and Value Assessment: When encountering an object, consider its ownership and value implications. Understanding Intentions: Analyze the motivations behind others' actions to gain insight into their goals.  Example of Intelligence Human Intelligence: Refers to the cognitive abilities of humans to learn, understand, and apply Human knowledge. Intelligence Example: Solving a complex math problem or making strategic decisions. Humans use language, abstract thinking, and creativity to solve problems. Animal Intelligence: Involves instinctual and learned behaviors that animals use to survive and Animal interact. Intelligence Example: A rat navigating a maze to find cheese. Animals adapt to their environment, often showing problem-solving skills. Collective Intelligence: Emerges from the collaboration and competition of many individuals. Collective Example: Ant colonies finding the shortest path to food or humans using the Intelligence internet to crowdsource solutions. It's not about individual knowledge but the synergy of many agents working together. Example of Intelligence Agent Environment Goal Perception Action Human Exam Pass Read Solve Mouse Maze Cheese See Navigate Ants Colony Protect Smell Attack Class discussion: AI in YouTube and Netflix Search recommendations The Quest of Artificial Intelligence: Emulating Human Cognition Understanding Human Thought: Going beyond observing human thought processes. Aiming to replicate the complexities of human intelligence. Creating Intelligent Entities: Constructing systems that can reason, learn, and solve problems. Building entities that can engage with their environment intelligently. Universal Applicability: AI's potential extends to every field imaginable. From healthcare to education, finance to space exploration, AI has a role. Conclusion: AI represents humanity's endeavor to create technology that not only mimics but enhances our innate cognitive abilities. The Imperative for Artificial Intelligence Creating Expert Systems: AI develops systems that emulate the decision-making abilities of human experts. Utilizes domain-specific knowledge to provide reliable advice and interpretation. Solving Complex Problems: Harnesses computational power to address intricate challenges. Delivers solutions across various domains from medical diagnostics to climate modeling. AI Schools of Four schools of thoughts (Russel & Norvig) thoughts Thinking humanly Thinking rationally The exciting new effort to make computers The study of mental faculties through the think... machines with minds, in the full use of computational models. (Charniak and literal sense. (Haugeland, 1985) and McDermott, 1985) Acting humanly Acting rationally The study of how to make computers do Computational Intelligence is the study of things which, at the moment, people are the design of intelligent agents. better. (Poole et al., 1998) (Rich and Knight, 1991)  AI Introduction to AI Schools of Thought Humanly Rationally 1.Thinking Humanly 1. The cognitive modeling approach. 2. Emphasizes processes that mimic human thought. Thinking Thinks like Thinks rationally 2.Thinking Rationally human 1. The laws of thought approach. 2. Focuses on logic and deduction. 3.Acting Humanly 1. The Turing Test approach. Acting Acts Like Acts rationally 2. AI's ability to emulate human behavior. human 4.Acting Rationally 1. The rational agent approach. 2. AI performs actions to achieve the best outcome. Thinking Humanly: The Cognitive Approach to AI Definition: Thinking humanly focuses on emulating human thought processes. Key Concepts: Cognitive Simulation Psychological Experimentation Human-like Reasoning Goals: To create AI that mirrors the nuances of human intelligence. To understand and model how humans think and make decisions. Thinking Humanly: The Cognitive Approach to AI A real-world example of Thinking Humanly: The Cognitive Approach to AI is the development of chatbots designed for therapy, such as Woebot. Woebot uses principles from cognitive behavioral therapy (CBT) to interact with users, helping them manage their mental health by simulating human-like conversations. The AI chatbot engages users by asking about their day, responding to emotional cues, and guiding them through exercises like journaling or mindfulness. It learns from interactions and adapts its responses to create a more personalized experience. Cognitive Simulation: Woebot mimics a human therapist by guiding users through Cognitive Behavioral Therapy (CBT) techniques, simulating human thought processes with psychological principles. Psychological Experimentation: Woebot mirrors human therapists by learning from user input and adjusting its responses to provide personalized mental health support, similar to real psychological methods. Human-like Reasoning: Woebot applies CBT principles to suggest activities like journaling or mindfulness based on the user’s emotions, mimicking the reasoning a therapist would use. Acting Humanly: The Turing Test Approach to AI The Turing Test Legacy: A computer is said to act humanly when it passes the Turing Test, becoming indistinguishable from a human in its actions. Biomimicry in AI: AI models that replicate human or animal behaviors. Incorporation of biological principles into machine functioning. The Turing Test Explained: An evaluation where if a human interacts with a computer and cannot reliably tell it apart from a human, the computer is considered to be 'acting humanly'. Turing's Contribution to AI: Proposed foundational components of AI: knowledge, reasoning, language understanding, and learning. Predicted challenges and skepticism regarding the potential of AI Acting Humanly: The Turing Test Approach to AI A real-world example of Acting Humanly: The Turing Test Approach to AI is Sophia, the humanoid robot developed by Hanson Robotics. Turing Test Relevance: Sophia is designed to interact with humans using natural language, facial expressions, and human-like gestures. The goal is for her to behave so similarly to a human that people can’t tell if they are interacting with a robot or a person. Human Interaction: Sophia can hold conversations, recognize human emotions, and respond appropriately, which helps her pass for human in many social contexts, simulating human-like interaction as Alan Turing envisioned. Sophia exemplifies the Turing Test approach, where AI mimics human behavior closely enough that it becomes difficult to distinguish between the actions of a human and a machine What is the Turing Test https://youtu.be/sXx-PpEBR7k Turing Test https://youtu.be/D5VN56jQMWM  The Turing Test is passed if we are not able to decipher whether the response to our questions is from a human or a computer. Meaning the computer program has evolved so much that it has fooled you. We need a lot of capabilities this is not an easy problem: Natural language processing: Natural language understanding and natural language generation Turing Test Knowledge representation: ontology (how to represent knowledge because you know asking questions which need knowledge) Automated reasoning: you have to understand the question, search for proper answers from databases question, and answers Machine learning: pattern detection and prediction for the Unseen  Thinking Rationally - The Logical Approach to AI Laws of Thought: AI that thinks rationally adheres to established laws of logic. Emphasizes the application of mathematical rigor to reasoning. Historical Foundations: Rooted in classical Greek philosophy and formalized through various logics, like propositional and first-order logic. Codifying Knowledge: Aims to represent all knowledge logically to infer new knowledge. Challenges Identified: Not all knowledge can be encapsulated in logical formulas. The issue of computational blow-up - the exponential increase in computational resources needed with increasing complexity. Thinking Rationally - The Logical Approach to AI A real-world example of Thinking Rationally in AI is IBM’s Watson, specifically its use in healthcare for diagnosing diseases. Watson processes vast amounts of medical literature and patient data to make logical, data-driven diagnoses and treatment recommendations. For example, when a doctor inputs a patient's symptoms and medical history into Watson, the AI uses logical reasoning to analyze this information, cross-referencing it with medical research and known treatment methods. Watson then provides the most likely diagnosis and suggests a treatment plan based on logical deductions and patterns found in the data. Why It’s Thinking Rationally: Laws of Thought: Watson uses logical rules to cross-reference symptoms, medical histories, and research papers. Mathematical Rigor: It applies mathematical models to analyze data and suggest the most probable outcomes. Codifying Knowledge: Watson codifies medical knowledge to help doctors make better decisions.  Acting Rationally: The Rational Agent Approach Understanding Rational Agents: Agents perform actions based on perceptual input to maximize goal achievement. Key Attributes of a Rational Agent: Perception: Interprets current environmental states. Autonomy: Operates without human intervention. Adaptability: Modifies behavior in response to environmental changes. Goal-Oriented: Actions are aimed at achieving specific objectives. Persistence: Functions over extended periods with sustained purpose. Design Philosophy: Crafting intelligent systems that act optimally within their scope of goals. Optimal Action and Goals: Rational agents take actions that they expect will lead to the best outcome according to their knowledge and the information available about the environment.  Acting Rationally: The Rational Agent Approach A real-world example of Acting Rationally: The Rational Agent Approach is a self-driving car, such as those developed by Tesla or Waymo. In a self-driving car, the AI acts as a rational agent by continuously perceiving the environment through sensors and cameras, such as identifying pedestrians, road signs, other vehicles, and obstacles. Based on this input, it makes decisions like when to stop, accelerate, or change lanes—all with the goal of safely reaching its destination while optimizing efficiency and minimizing risks. Perception: The car interprets its surroundings in real time (traffic lights, road signs, obstacles). Autonomy: It operates without human intervention. Adaptability: It adapts to new situations, like adjusting for unexpected roadblocks or changing traffic conditions. Goal-Oriented: Its primary goal is to transport passengers safely from point A to point B. Persistence: The system continually monitors and makes decisions throughout the trip. This example demonstrates Acting Rationally because the self-driving car takes actions to achieve the best possible outcomes (safety, efficiency) based on the information it perceives, without trying to mimic human thinking or behavior. It simply aims for the most rational decisions in every scenario. The Multidisciplinary Pillars of Artificial Intelligence of AI The Interconnected Disciplines: AI is built upon contributions from diverse fields, each providing unique insights and tools. Philosophy: Investigates the fundamental nature of knowledge and reality, informing AI's approach to cognition and reasoning. Mathematics: Provides the formal frameworks and algorithms essential for AI development. Economics: Offers models of decision-making and resource optimization that AI applies to problem-solving. Neuroscience: Unravels the workings of the human brain, inspiring AI neural network design. The Multidisciplinary Pillars of Artificial Intelligence of AI Psychology: Explores human thought and behavior, guiding the development of AI that thinks and learns like humans. Computer Engineering: Delivers the hardware and software infrastructure enabling AI functionalities. Control Theory and Cybernetics: Governs the systems and processes control, contributing to AI's ability to self-regulate and adapt. Linguistics: Unlocks the complexities of language, vital for natural language processing in AI.  Types/Levels of AI These are classifications based on how AI functions, the complexity of its reasoning, and how it processes information. It's more about the technical functionality of AI systems and the level of sophistication in terms of learning, reasoning, and awareness. Here are the types/levels of AI: a) Reactive Machines: Simple AI systems that only react to current inputs and cannot store memory or learn from past experiences. b) Limited Memory Machines: AI systems that can use historical data or past experiences to improve their decision-making over time (e.g., self-driving cars). c) Theory of Mind AI:AI that could understand human emotions, beliefs, and intentions and interact socially (still in development). d) Self-aware AI:The most advanced hypothetical form of AI that would have its own consciousness and self-awareness. Types or Levels of AI Reactive Machines Definition: These are the most basic forms of AI that can only react to specific inputs. They do not store memory or past experiences and can't use that information to make future decisions. Characteristics: No learning from past experiences, only immediate reaction to current situations. Examples: IBM’s Deep Blue chess computer, which could analyze the current state of the chessboard and make moves based solely on that information, without any memory of past moves. Types or Levels of AI Limited Memory Machines Definition: These AI systems can store and use past experiences or data to make decisions and improve over time. Most AI applications in use today fall under this category. Characteristics: These machines learn from historical data to make better predictions or decisions but do not "think" like humans. Examples: Self-driving cars, which use historical sensor data (speed, direction, distance to objects) to navigate roads and avoid accidents. Types or Levels of AI Theory of Mind AI Definition: This stage of AI, which is still hypothetical, refers to machines that can understand human emotions, beliefs, intentions, and thoughts. The AI would be able to interact more socially and understand that other entities (like humans) have thoughts and feelings. Characteristics: AI would be able to mimic and respond to human emotions and mental states. Examples: Although not fully developed, early precursors could be seen in advanced chatbots or social robots designed for emotional interaction, like Sophia or Emo AI. (Robot soccer, http://www.cs.cmu.edu/~robosoccer/main/ and https://www.robocup.org/, is another example of this kind of understanding, but at a simple level.) Types or Levels of AI Self-awareness AI Definition: The final and most advanced form of AI, self-aware AI refers to machines that not only understand human emotions and thoughts but also have a sense of their own existence, consciousness, and self-awareness. This level remains purely theoretical. Characteristics: These machines would have their own thoughts, desires, and possibly even their own "personalities." Examples: None currently exist. This remains in the realm of science fiction for now  Stages of Artificial Intelligence These refer to the developmental stages or the progression of AI from task-specific systems to superintelligent machines. It's more about the capabilities AI achieves over time and how it evolves to perform more complex tasks across various domains. Here are the stages of AI: a) Artificial Narrow Intelligence (ANI) or Weak AI AI that specializes in performing one specific task (e.g., voice assistants, chess-playing AI). It cannot generalize beyond its predefined function. b) Artificial General Intelligence (AGI) or Strong AI AI that has the ability to understand, learn, and apply knowledge across a wide range of tasks, similar to human intelligence. This stage has not yet been fully achieved. c) Artificial Superintelligence (ASI) AI that surpasses human intelligence in all areas, including creativity, decision-making, and complex problem-solving. This stage remains theoretical.  Narrow or Weak AI Artificial Narrow Intelligence (ANI) Examples of ANI: ANI, or weak AI, refers to AI systems designed Voice assistants like Siri and Alexa. to handle singular or limited tasks. AI in games, such as AlphaGo. They operate without genuine cognition, Social robots like Sophia. executing predefined functions. Autonomous vehicles. Characteristics of ANI: Scope of ANI: Task-specific: Optimized for particular The predominant form of AI in use today. operations. ANI systems are pervasive across various Non-sentient: Lacks self-awareness and industries, enhancing efficiency and consciousness. capabilities. Programmatically limited: Functions within a set scope without improvisation.  General or Strong AI Artificial General Intelligence (AGI) AGI's Implications: AGI or strong AI represents the While some view AGI as a potential threat, others future stage of AI where machines will have cognitive capabilities see it as an evolutionary leap in problem-solving comparable to humans. and task management. The Human Brain Analogy: Potential and Capabilities: AGI aims to emulate the versatility and adaptability Theoretical machines capable of of the human brain. understanding, learning, and Future Predictions: applying intelligence broadly across Estimates suggest AGI might emerge by mid-21st diverse tasks. century, expanding AI's ability beyond narrow tasks Current Status: to general intelligence. No current examples exist; AGI remains a theoretical concept within the realm of AI research. Superintelligent AI Artificial Super Intelligence (ASI) This Photo by Unknown Author is licensed under CC BY-SA-NC A hypothetical realm of AI where machines surpass all human capabilities, including Current Status: ASI remains a concept within cognitive functions like learning, reasoning, and the realm of science fiction and speculative decision-making. future studies. It is not yet a reality and exists Beyond Human Intelligence: At this stage, primarily in theoretical discussions and ASI systems would be able to perform any entertainment media. intellectual task that a human can, but with Cinematic Representation: Popular films often greater efficiency, accuracy, and speed. portray ASI as machines dominating the world, highlighting the potential extremes of AI development. Applications of AI https://youtu.be/NJarxpYyoFI  Deep Blue - The Chess-Playing Legacy Deep Blue, named in homage to the fictional computer from The Hitchhiker's Guide to the Galaxy, was engineered to answer the grand question of chess mastery. Grandmaster-Level Design: A parallel, RS/6000 SP-based supercomputer specifically created to compete with chess grandmasters. Processing Power: 30 High-Performance Processors: Dedicated to software search algorithms. 480 Custom VLSI Chess Processors: Specialized in move generation and hardware search. Unmatched Search Capabilities: Average Search: 126 million nodes per second. Peak Performance: Up to 330 million nodes per second. Chess Analysis Proficiency: Position Generation: 30 billion positions per move. Average Depth: Routinely reaches 14 moves deep.  Deep Blue - The Chess-Playing Legacy Sophisticated Algorithms: Iterative Deepening Alpha-Beta Minimax: A powerful search algorithm used for decision- making in game theory. Transposition Tables and Singular Extensions: Advanced techniques to enhance move evaluation. Chess-Specific Heuristics: Tailored strategies to understand and predict game play. Historical Database: A vast collection of opening and endgame strategies. Impact on AI and Gaming: Deep Blue's victory over World Chess Champion Garry Kasparov marked a significant milestone in the development of AI. It demonstrated the potential of combining raw computational power with finely tuned algorithms and vast knowledge bases. Applications of AI Jeopardy (2011): Humans vs. IBM Watson Natural Language Understanding and information extraction https://youtu.be/P18EdAKuC1U  Jeopardy (2011): Humans vs. IBM Watson IBM Watson's Landmark Victory: In 2011, IBM Watson, a powerful AI system, famously won the game show Jeopardy against human champions, showcasing the advances in Natural Language Understanding (NLU). Core Technologies Behind Watson: Automatic Speech Recognition (ASR): Watson's ability to interpret the spoken clues. Natural Language Processing (NLP): Processing and understanding complex language patterns in Jeopardy clues. Machine Learning: Utilizing vast databases to generate hypotheses, evidence scoring, and confidence ratings. NLU in Watson: Information Extraction: Utilizing techniques like Named Entity Recognition (NER) and Relation Extraction to draw out structured information from clues. Understanding Context: Employing Semantic Parsing and Coreference Resolution to understand the context of clues. Inference: Making use of Paraphrasing and Natural Language Inference to handle the nuanced language in questions. Result Interpretation and Response Generation: Question Answering (QA): Watson's integration of various data sources to formulate correct responses. Dialogue Agents: Engaging in the game show's conversational format. Summarization: Condensing information to form precise answers. Impact of Watson's Victory: Highlighted the potential for AI in various fields like healthcare, finance, and customer service. Demonstrated the effectiveness of NLU and information extraction in real-world, high-pressure situations.