Artificial Intelligence (AI) - Addis Ababa University

Chapter Three: Artificial Intelligence (AI) EmTe 1012 Prepared by : Surafiel H. Department of Computer Science, Addis Ababa University March, 2022  Introduction Artificial Intelligence is composed of two words, ea...

Chapter Three: Artificial Intelligence (AI) EmTe 1012 Prepared by : Surafiel H. Department of Computer Science, Addis Ababa University March, 2022  Introduction Artificial Intelligence is composed of two words, each with its own meaning. ✓ Artificial defined as “man-made” ✓ Intelligence refers to thinking power, or the ability to learn, solve problems, and acquire and apply knowledge. AI: A man-made thinking power. ✓ The ability to learn and solve problems; ✓ The ability to acquire and apply knowledge. Knowledge is the information acquired through experience. Experience is the knowledge gained through exposure (training). 2  Intelligence is composed of: 1. Learning: the acquisition of knowledge or skills through study, experience, or being taught. 2. Reasoning: the act of thinking about something in a logical sensible way. 3. Problem-Solving: the process of finding solution to difficult or complex issues. 4. Perception: the ability to detect or become aware of something through the senses. 5. Linguistic Intelligence: the ability to understand and use spoken and written language. 3 4  Artificial Intelligence (AI) Artificial Intelligence (AI) is a wide-ranging branch of Computer Science concerned with building smart machines capable of performing tasks that typically require human intelligence. Or AI is the branch of computer science by which we can create intelligent machines which can behave like a human. ✓ Think like humans, and able to make decisions. 5  Artificial Intelligence exists when a machine can have human-based skills such as: ✓ Learning, ✓ Reasoning, and ✓ Problem solving An AI machine can perform tasks without being specifically pre-programmed to do so. 6  Agent and Environment An AI system is composed of an agent and its environment. Agent: is anything that can be viewed as: ✓ Perceiving its environment through sensors, and ✓ Acting upon that environment through actuators/effectors. Examples of Agent: ✓ A Human-agent has eyes, ears, and other organs which act as sensors, and hands, legs, mouth, and other body parts acting as actuators. ✓ A Robotic agent has cameras and infrared range finders which act as sensors and various motors acting as actuators/effectors. 7  Intelligent agents must be able to set goals and achieve them. An agent: ✓ Assess its environment, ✓ Make predictions, and then ✓ Evaluate its predictions and ✓ Adapt based on its assessment 8  Machine perception is the ability to use input from sensors (such as cameras, microphones, etc.) to deduce aspects of the world. For example, computer vision. Computer vision is an AI field that empowers computers and systems to extract meaningful information from digital images, videos, and other visual inputs, enabling them to take actions or make recommendations based on that information Computer vision describes the machine's understanding of images and videos. 9  High-Profile Examples of AI Autonomous vehicles: ✓ Vehicles like drones and self-driving cars that navigate and operate without human intervention. Medical diagnosis: ✓ AI systems that analyze medical data to assist in diagnosing diseases and conditions. Creating art: ✓ AI algorithms capable of generating artistic content, including poetry and visual art. Proving mathematical theorems: ✓ AI systems that use mathematical algorithms to prove complex theorems and solve mathematical problems. 10  Playing games: ✓ AI programs, like AlphaGo that can play and excel at strategic games against human opponents. Search engines: ✓ AI-driven algorithms that retrieve relevant information from the internet based on user queries. Eg. Google search engine Online assistants: ✓ Virtual assistants that use AI to perform tasks and provide information based on voice commands. Eg. Siri Image recognition in photographs: ✓ AI technology capable of identifying and categorizing objects, people, or scenes in digital images. 11  Spam filtering: ✓ AI systems that automatically detect and filter out unsolicited or unwanted emails. Prediction of judicial decisions: ✓ AI models that analyze legal data to predict outcomes of legal cases or decisions. Targeting online advertisements: ✓ AI algorithms used to analyze user data and behavior to deliver personalized advertisements. 12 Enabling Technologies for the Advancement of AI The arrival of the Internet, The advent of Big Data, Smart mobile phones, Social media, Cheaper and more powerful hardware: ✓ such as Graphical Processing Units (GPUs) 13 Relationships of AI, ML, and DL 14  AI (Artificial Intelligence): An umbrella discipline that covers everything related to making machines smarter. Refers to the general broader concept of creating machines/software that can perform tasks requiring human intelligence. Aims to simulate human-like cognitive processes such as problem-solving, reasoning, learning, and decision-making. Encompasses various techniques such as rule-based systems, natural language processing (NLP), and computer vision. The ultimate goal of AI is to mimic human intelligence across different tasks. 15  Machine Learning (ML): A subset of AI that focuses on algorithms learning from data to improve performance on specific tasks. Algorithms learn from examples rather than being explicitly programmed. Enables machines to learn and make predictions by recognizing patterns or relying on past experiences. Can make decisions and improve over time without human intervention. Generally, it’s an AI system that can self-learn based on the algorithm. ML techniques are used in a wide range of applications, from recommendation systems and fraud detection to image recognition and language translation. ML includes supervised, unsupervised, and reinforcement learning techniques or 16 algorithms.  Deep Learning (DL): A specialized form of ML that handles large datasets and is inspired by the human brain's structure. Associated with learning from examples and is inspired by the way the human brain filters information. Uses artificial neural networks composed of multiple layers to process and transform data. DL is particularly effective for tasks like image recognition, NLP tasks such as speech recognition, machine translation, text classification and summarization, question- answering systems, and autonomous driving. 17  Artificial Neural Networks: A computational model inspired by the structure and function of the human brain's neural networks. Consists of interconnected nodes, called neurons (nodes), organized in layers. Each neuron receives input, performs a computation on it, and produces an output. These neurons are arranged in layers, typically including an input layer, one or more hidden layers, and an output layer. “Deep” in deep learning refers to the presence of multiple layers in the network. DL systems help a computer model filter the input data through layers to predict and classify information. 18  Need for Artificial Intelligence 1. To Create Expert Systems that exhibit intelligent behavior that can learn, demonstrate, explain, and advise its users. ✓ An expert is an individual with extensive knowledge/skills in a particular area. An Expert system: ✓ A computer program utilizing AI technologies to emulate the judgment and behavior of a human/organization with expert knowledge and experience in a specific field. ✓ In essence, an expert system simulates the decision-making capabilities of a human expert. 19  Examples of Expert Systems: ✓ MYCIN: An expert system used to identify various bacteria causing severe infections. It also recommends drugs based on the patient’s weight. ✓ DENDRAL: An expert system for chemical analysis, predicting molecular structures. ✓ PXDES: An expert system designed to predict the degree and type of lung cancer. ✓ CaDet: One of the top expert systems for identifying cancer in its early stages. 2. To create helping machines that can find solutions to complex problems like humans do. 20  Goals of Artificial Intelligence Replicate human intelligence: ✓ Develop systems that can think, learn, and make decisions like humans. Solve knowledge-intensive tasks: ✓ Address complex problems that require extensive knowledge and reasoning. Establish an intelligent connection between perception and action: ✓ Enable machines to interpret sensory data and respond appropriately. Build machines to perform tasks requiring human intelligence: Examples: ✓ Planning surgical operations. ✓ Driving a car in traffic. ✓ Proving mathematical theorems. Create Systems Exhibiting Intelligent Behavior: ✓ Develop systems capable of learning autonomously, demonstrating acquired knowledge, and providing explanations. 21 Comprises of Artificial Intelligence 22  Advantages of Artificial Intelligence High Accuracy with Fewer Errors: ✓ AI systems exhibit high accuracy and fewer errors as they base decisions on pre-existing data and experiences. High Speed: ✓ AI systems can make decisions quickly, exemplified by AI defeating human champions in chess due to rapid decision- making. High Reliability: ✓ AI machines can perform repetitive tasks with consistent accuracy. Useful for Risky Areas: ✓ AI can be deployed in dangerous situations such as bomb defusal or deep-sea exploration, reducing human risk. Digital Assistant: ✓ AI enhances user experience by providing digital assistants, such as those used by e-commerce websites to suggest products based on customer preferences. Public Utility: ✓ Applications include self-driving cars for safer travel, facial recognition for security, NLP for search engines, spelling checkers for digital assistants like Siri and Google Assistant, and language translation services like Google Translate. 23  Disadvantages of Artificial Intelligence High Cost: ✓ AI requires expensive hardware and software, along with significant maintenance to stay up-to-date with current demands. Limited Creativity (Can’t think outside the box”): ✓ AI systems are confined to their programming and training, lacking the ability to think outside the box. No Feelings and Emotions: ✓ AI lacks emotional intelligence, preventing emotional connections with humans. Increased Dependence on Machines: ✓ Overreliance on AI may reduce human mental capabilities. Lack of Original Creativity: ✓ AI can’t match human creativity and imagination, as it can’t generate new ideas beyond its programming. 24  Levels of Artificial Intelligence 1. Rule-Based AI Involves programming explicit rules to guide the system's behavior. Characteristics: ✓ Uses simple if-then coding statements to make decisions. ✓ Produces predefined outcomes based on a set of human-coded rules. ▪ “IF X happens then do Y” ✓ Requires a set of facts/data and a set of rules for manipulating that data. Examples: ✓ Expert System: ▪ Helps a medical doctor choose the correct diagnosis based on symptoms. ✓ Chess Program: ▪ Selects tactical moves to play a chess game. ✓ Intelligent Thermostat: ▪ Adjusts temperature based on programmed rules like “if the temperature is below 65°F, then turn on the heater”. 25 2. Context-Awareness and Retention: Involves AI systems that can understand and adapt to the context of a situation. Characteristics: ✓ Considers previous interactions and adjusts responses accordingly. ✓ Trained on the best human knowledge and experience in specific domains. Examples: ✓ Google Assistant: Understands a user's previous queries to provide relevant answers. ✓ Chatbots: Simulate human conversation and learn from interactions to improve over time. Eg. ChatGPT ✓ Robo-advisors: Provide financial advice based on user interactions. 26 3. Domain Specific Expertise: AI systems with specialized knowledge in a particular field. Characteristics: ✓ Can make decisions and perform tasks within a specific domain. ✓ Often exceed human capabilities in their specialized area. Examples: ✓ Medical Diagnosis Systems: Analyze patient data to provide accurate diagnoses. ✓ Google DeepMind's AlphaGo: Defeated 18-time world Go champion Lee Sedol, showcasing domain-specific expertise. 27 4. Reasoning System (Machine): AI systems that perform logical reasoning and draw conclusions from available information. Characteristics: ✓ Concepts of intellect, intentions, knowledge, and logic. ✓ Can reason, interact, and negotiate with humans and other machines. Examples: ✓ Legal Expert System: Uses legal rules to provide legal advice or predict case outcomes. Current Status: still in development. 28  5. Self-Aware Systems / Artificial General Intelligence (AGI): AI systems possessing human-like general intelligence, capable of understanding, learning, and applying knowledge across a wide range of tasks. Characteristics: ✓ Have a form of self-awareness. ✓ Perform tasks without specific programming. Current Status: No such systems exist today. 29  6. Artificial Super Intelligence (ASI): AI that surpasses human intelligence in virtually every aspect. Characteristics: ✓ Capable of outperforming the smartest humans in all domains. ✓ Potential to solve complex problems like poverty, hunger, and climate change. Current Status: Hypothetical, posing ethical and existential questions. 30  7. Singularity and Transcendence/Excellency: Singularity: A hypothetical point where AI's growth becomes uncontrollable and irreversible, leading to rapid advancements. Transcendence/Excellency: AI surpassing human intelligence to a level that is incomprehensible and unimaginable to humans. Characteristics: ✓ Potential for humans to connect minds and communicate like a "human internet." ✓ Possible interactions with other forms of life and natural activities. Current Status: Controversial and debated; some believe it could be achieved by 2045. 31  Types of Artificial Intelligence AI can be categorized into different types based on their capabilities and functionalities. AI Capabilities: ✓ Refer to the inherent abilities and skills that an AI system possesses. ✓ These abilities result from the technologies, algorithms, and design principles built into the AI system. ✓ AI capabilities encompass a wide range of cognitive and computational tasks the AI system can potentially perform. ✓ These tasks often derive from fields such as machine learning, natural language processing, and computer vision. ✓ These capabilities describe the fundamental abilities of the AI system, regardless of the specific tasks it's designed to carry out. 32  AI Functionalities: ✓ AI functionalities are the specific applications, tasks, or operations that an AI system can perform based on its capabilities. ✓ These functionalities are the practical ways in which the AI system's abilities are put to use to achieve specific goals or provide value to users or applications. ✓ Utilize the underlying capabilities of the AI system to provide tangible results or perform specific tasks. 33  AI Types: Based on Capabilities: 1. Weak or Narrow AI: The most common and currently available AI. It’s limited to a specific or narrow area. It’s able to perform a dedicated task with intelligence. Can’t perform beyond its field or limitations, as it’s only trained for one specific task. Examples: ✓ Virtual Personal Assistants: Siri, Google Assistant, and Amazon Alexa are examples of narrow AI. They can perform tasks like setting reminders, sending messages, or providing weather updates. ✓ Image Recognition: AI systems that can identify objects or people in images, such as facial recognition software used in social media platforms. ✓ Language Translation: Applications that provide language translation services, like Google Translate, are narrow AI as they focus on translating text between languages. 34 2. Strong or General AI: General AI is a type of intelligence that could perform any intellectual task with efficiency like a human. Refers to artificial intelligence that possesses human-like cognitive abilities. This AI would have the ability to understand, learn, and apply knowledge across a wide range of tasks, similar to how humans can. It would possess a form of consciousness and self-awareness, allowing it to perform any intellectual task that a human can do. It’s used to make a system that could be smarter and think like a human on its own. Currently, there is no such system; it remains a theoretical concept and has not been achieved. It may arrive within the next 20 or so years. 35 3. Super AI: It’s an outcome of general AI. Refers to an intelligent machine that is smarter than human intelligence and ability. Super AI, also known as Artificial Superintelligence (ASI), goes beyond human intelligence in all aspects. This type of AI surpasses human cognitive abilities and has the potential to outperform the best human minds in practically every field, including creativity, problem-solving, scientific research, and so on. It’s still a hypothetical concept of AI. 36  AI Types: Based on Functionality: 1. Reactive Artificial Intelligence: It’s the simplest type of AI; React to some input with some output. Refers to AI systems that can make decisions based solely on predefined rules and programmed responses. There is no learning that occurs, and this is the first stage of any AI system. Don't possess memory or the ability to learn from past experiences. They excel at the specific tasks for which they are designed but lack adaptability or the ability to handle new situations. Examples: ✓ Chess-playing AI: Deep Blue, the AI developed by IBM, defeated world chess champion Garry Kasparov in 1997. Deep Blue utilized a vast database of chess moves and expert-defined rules to make strategic decisions. ✓ Automated Manufacturing: Robots used in manufacturing processes that perform repetitive tasks based on pre-defined instructions are examples of reactive AI. 37 2. Limited Memory Artificial Intelligence: Capable of learning from historical data and experiences to some extent. They can make decisions based on both pre-programmed rules and patterns learned from past data. Can store past experiences or some data for a short period of time. Uses the stored data for a limited time only. However, their learning and decision-making capabilities are confined to a specific domain, and they don't possess a full understanding of the world. Examples: Self-driving cars are one of the best examples of limited memory AI systems. These vehicles use limited memory AI to process real-time sensor data and historical data to navigate roads and make driving decisions. These cars can store: ✓ The recent speed of nearby cars, ✓ The distance of other cars, ✓ Speed limits, and other information to navigate the road. Recommendation Systems: Online platforms like Netflix and Amazon use limited memory AI to suggest movies or products based on users' past preferences and behaviors. 38 3. Theory of Mind Artificial Intelligence: Theory of Mind AI refers to when machines acquire decision-making capabilities equal to those of humans. It involves understanding the mental states of other entities, including their emotions, intentions, beliefs, and desires. This type of AI would be able to comprehend and predict the actions of humans and other AI systems based on their perceived mental states. It is still not developed, but natural language-understanding AI systems that can accurately interpret and respond to human emotions and intentions in text or speech could be considered to have a rudimentary form of Theory of Mind. 39 4. Self-Aware Artificial Intelligence (artificial superintelligence): Self-awareness AI, often referred to as AI superintelligence, represents an advanced level of AI that not only possesses human-like cognitive abilities but also exhibits self-awareness, consciousness, sentiments, and the ability to introspect. This level of AI would surpass human intelligence in all respects. It is the future of AI. This AI concept remains theoretical and speculative, and its development raises significant ethical, philosophical, and technical challenges. 40  How Humans Think Human intelligence, or the cognitive process, is composed of three main stages: 1. Observe and input the information or data into the brain. 2. Interpret and evaluate the input that is received from the surrounding environment. 3. Make decisions as a reaction to what you received as input, interpreted, and evaluated. 41  Mapping Human Thinking to AI Components In the first stage, humans acquire information from their surrounding environments through human senses, such as sight, hearing, smell, taste, and touch, and through human organs, such as eyes, ears, and other sensing organs. In AI models, this stage is represented by the sensing layer, which perceives information from the surrounding environment. This information is specific to the AI application. There are sensing agents such as: ✓ Voice recognition for sensing voice, ✓ Visual imaging recognition for sensing images. Thus, these agents or sensors take on the role of the hearing and sight senses in humans. 42  The second stage is related to interpreting and evaluating the input data: ✓ In AI, this stage is represented by the interpretation layer, i.e., reasoning and thinking about the gathered input that is acquired by the sensing layer. The third stage is related to taking action or making decisions: ✓ After evaluating the input data, the interacting layer performs the necessary tasks. ✓ Robotic movement control and speech generation are examples of functions that are implemented in the interacting layer. 43  Influencers of Artificial Intelligence 1. Big Data: Refers to Massive volumes of structured and unstructured data from various sources such as social media, sensors, and transactions. Role in AI: Provides raw material for training and improving machine learning algorithms. Larger and more diverse datasets enhance the accuracy of AI models. Examples: ✓ Recommendation Systems: Platforms like Netflix analyze users' viewing habits, preferences, and ratings to recommend tailored content. More data leads to more accurate recommendations. ✓ Real-time Data Analysis: Major hospital networks use real-time data and analytics dashboards to save millions and reduce patient length-of-stay. ✓ Chatbots in Financial Firms: Deployment of chatbots for customer support increases accounts. ✓ Product Recommendations: AI increases product quotes by 60% in insurance and travel services by referring products to customers. ✓ Equipment Health Assessment: AI improves productivity and reduces costs by assessing equipment health. 44 2. Advancements in Computer Processing Speed and New Chip Architectures ✓ Impact: Improved processing power enables faster and more efficient execution of AI algorithms. ✓ Technologies: GPUs (Graphics Processing Units) and TPUs (Tensor Processing Units) accelerate AI tasks like deep learning. Example: ✓ Deep Learning: Neural networks benefit from faster processing speeds, enhancing tasks like image recognition and natural language processing. This enables real-time applications such as autonomous vehicles and instant language translation. 45 3. Enabled Processing of Large Amounts of Data at High Speed Importance: Crucial for AI systems to learn patterns, make predictions, and generate insights. Benefit: Improved processing capabilities handle complex calculations and extract meaningful information from diverse data sources. Example: ✓ Healthcare Diagnostics: AI algorithms analyze medical images (e.g., MRIs, CT scans) to detect patterns indicating diseases like cancer, aiding radiologists in making more accurate diagnoses. 46 4. Cloud Computing and APIs Cloud Computing: Delivery of on-demand services (servers, storage, databases, networking, software, analytics, and intelligence) via the internet on a pay-per-use basis. ✓ Role: Provides scalable and accessible computational resources, crucial for AI projects requiring significant power. ✓ Services Enabled: Data analysis, social media, video storage, e-commerce, and AI capabilities. 47  Application Program Interfaces (APIs): ✓ A set of rules and protocols for building and interacting with software applications. APIs allow different software systems to communicate with each other. ✓ They define the methods and data formats that applications can use to request and exchange information. ✓ Standardized interfaces for integrating AI services into applications, facilitating communication between software components. Examples: ✓ Natural Language Processing: Cloud-based APIs (e.g., Google Cloud Natural Language, IBM Watson's Language Translator) add language understanding and translation functionalities to apps. 48 Cloud AI Services: ▪ IBM Watson AI services via IBM Cloud. ▪ Amazon AI services via Amazon Web Services (AWS). ▪ Microsoft AI tools via MS Azure cloud. ▪ Google AI services via Google Cloud Platform. 49 5. Emergence of Data Science Data science refers to the extraction of insights and knowledge from data using statistics, machine learning, and domain expertise. Role: Data scientists shape AI projects from data collection and preprocessing to model development and interpretation. Example: ✓ Fraud Detection: Data scientists create ML models to analyze transaction data and identify patterns associated with fraudulent activities, detecting anomalies in real-time transactions. 50  Applications of Artificial Intelligence Agriculture: ✓ Optimization: AI optimizes crop management, monitors plant health, and improves yield prediction. ✓ Technologies: Agriculture robotics, soil and crop monitoring, predictive analysis. ✓ Example: Agrobots equipped with computer vision can identify and remove weeds, reducing the need for herbicides. Healthcare ✓ Medical Diagnosis: AI aids in medical diagnosis, drug discovery, personalized treatment, and administrative tasks. ✓ Efficiency: Makes better and faster diagnoses than humans. ✓ Example: AI-powered medical imaging helps radiologists detect abnormalities in X-rays and MRIs. 51  Education: ✓ Personalized Learning: AI assists in personalized learning, automated grading, and creating adaptive educational content. Examples: ✓ Online Learning Platforms: AI-driven platforms like Coursera offer personalized course recommendations based on learner behavior. ✓ AI Chatbots: Act as teaching assistants, communicating with students. Finance and E-commerce: ✓ Applications: AI is used for fraud detection, algorithmic trading, customer service chatbots, and product recommendations. Examples: ✓ Chatbots: E-commerce sites like Amazon use AI chatbots for customer support and product inquiries. ✓ Adaptive Intelligence: Helps shoppers discover products. 52  Gaming and Entertainment: ✓ Gameplay Enhancement: AI enhances gameplay with realistic characters, dynamic environments, and procedural content generation. Examples: ✓ NPCs: Non-player characters in video games exhibit intelligent behavior and adapt to player actions. ✓ Content Recommendations: Services like Netflix use AI for program and show recommendations. Data Security: ✓ Cybersecurity: AI identifies and responds to cybersecurity threats, analyzing patterns to prevent attacks. ✓ Bug Detection: AI can be used to determine software bugs. Example: AI-driven systems monitor network traffic to detect unusual behavior that could indicate a breach. 53  Social Media: ✓ Content Management: AI powers content recommendation, sentiment analysis, and image recognition. ✓ Trend Analysis: Analyzes data to identify the latest trends, hashtags, and user requirements. Example: Facebook's AI identifies and suggests tags for people in photos. Travel & Transport: ✓ Optimization: AI optimizes route planning, vehicle maintenance, and customer service. Examples: ✓ Ride-sharing Apps: Apps like Uber use AI algorithms to match drivers and riders efficiently. ✓ Travel Arrangements: Suggests hotels, flights, and best routes. 54  Automotive Industry: ✓ Applications: AI is used for autonomous driving, predictive maintenance, and vehicle safety systems. Examples: ✓ Self-driving Cars: AI enables features like lane-keeping and adaptive cruise control in Tesla's Autopilot. Robotics: ✓ Task Performance: AI enables robots to perform tasks in manufacturing, healthcare, exploration, and more. ✓ Intelligent Robots: Can perform tasks based on experiences without being preprogrammed. Example: ✓ Surgical robots like the da Vinci Surgical System assist surgeons in performing complex procedures with precision. 55  Artificial Intelligence Tools and Platforms AI has evolved a diverse range of tools to solve some of the most challenging problems in computer science. These tools encompass: Search and Optimization: ✓ Function: Refines websites to enhance their visibility on search engines, increasing online reach and impact. ✓ Techniques: Ensure websites rank higher in search results, driving more traffic and engagement. Logic: Computability Theory: Explores the boundaries of what machines can compute. Impact: Contributes to the development of more efficient and powerful computational methods. Probabilistic Methods for Uncertain Reasoning: Function: Makes decisions and draws conclusions under uncertainty. Techniques: Assess probabilities and weigh various factors to make informed judgments.. 56  Classifiers and Statistical Learning Methods: Function: Identifies patterns and relationships within data. Techniques: Categorize and predict outcomes, aiding decision-making across various applications. Neural Networks: Function: Emulates the human brain to recognize intricate patterns within extensive datasets. Applications: Image recognition, language processing, and data analysis. Control Theory: Function: Guides the behavior of systems towards specific objectives using feedback loops. Impact: Ensures systems adapt dynamically to changes, achieving desired goals with precision. Languages: Automatic Manipulation of Natural Language: Function: Enables software to comprehend, generate, and interact with human language. Applications: Chatbots, language translation, and voice assistants. 57  The most common AI platforms: Microsoft Azure Machine Learning ✓ A cloud-based platform offering tools and services for building, deploying, and managing machine learning models. ✓ Allows data scientists and developers to create predictive models, automate workflows, and integrate AI into applications using various programming languages and frameworks. Google Cloud Prediction API (AI Platform Prediction) ✓ A service for building and deploying machine learning models. ✓ Simplifies the creation of predictive models and offers scalable infrastructure for hosting and serving models to make predictions on new data. IBM Watson ✓ A suite of AI-powered services and solutions. ✓ Includes natural language processing, computer vision, chatbots, and more. ✓ Enables businesses to integrate advanced AI capabilities into their products, services, and processes. TensorFlow ✓ An open-source machine learning framework developed by Google. ✓ Allows developers and researchers to create, train, and deploy machine learning models across various platforms. ✓ Widely used for research and production, supporting deep learning and other machine learning tasks. https://www.tensorflow.org/tutorials/quickstart/beginner 58 https://www.datacamp.com/tutorial/tensorflow-tutorial  Deep Learning Libraries and Frameworks: ✓ PyTorch: An open-source deep learning framework developed by Facebook's AI Research lab. https://pytorch.org/tutorials/beginner/pytorch_with_examples.html https://www.youtube.com/watch?v=V_xro1bcAuA ✓ Keras: A high-level neural networks API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano. https://keras.io/ Reinforcement Learning Platforms: ✓ OpenAI Gym: A toolkit for developing and comparing reinforcement learning algorithms. https://openai.com/index/openai-gym-beta/ ✓ Unity ML-Agents: An open-source project that enables games and simulations to serve as environments for training intelligent agents. https://github.com/Unity-Technologies/ml-agents 59  AI in Edge Computing: ✓ NVIDIA Jetson: A platform for AI at the edge, providing powerful computing for deep learning and computer vision applications. ✓ Intel Movidius: Offers neural compute sticks and vision processing units for deploying AI at the edge. AI Ethics and Fairness Tools: ✓ IBM AI Fairness 360: A toolkit that helps detect and mitigate bias in machine learning models. ✓ Google's What-If Tool: An interactive visual interface for exploring machine learning models and their fairness. 60  Collaboration and Experimentation Platforms: ✓ Kaggle: A platform for data science competitions and collaborative projects, offering datasets, notebooks, and a community of data scientists. https://www.kaggle.com/ ✓ Google Colab: A free Jupyter notebook environment that runs in the cloud, allowing for easy sharing and collaboration on AI experiments. https://colab.research.google.com/ 61  Sample AI Applications Commuting: Google’s AI-Powered Predictions: Uses AI to predict traffic conditions, optimize routes, and provide real-time navigation assistance. Ridesharing Apps (Uber and Lyft): Utilize AI algorithms to match riders with drivers efficiently, optimize routes, and predict demand. Commercial Flights (AI Autopilot): AI autopilot systems assist pilots in navigating and managing flights, improving safety and efficiency. 62  Email: Spam Filters: AI-driven filters identify and block unwanted emails, ensuring only relevant messages reach your inbox. Smart Email Categorization: AI sorts emails into categories such as Primary, Social, Promotions, and Updates, enhancing organization and productivity. 63  Social Networking: ✓ Facebook: Automatically highlights faces in photos and suggests tags using facial recognition technology. ✓ Pinterest: Utilizes computer vision to identify objects in images (pins) and recommend similar pins. AI is also used for spam prevention, search and discovery, ad performance, monetization, and email marketing. ✓ Instagram: Uses machine learning to understand the contextual meaning of emojis, replacing slang terms (e.g., a laughing emoji for "lol"). ✓ Snapchat: Introduced facial filters (Lenses) in 2015, which track facial movements to add animated effects or digital masks that adjust dynamically. 64  Online Shopping: ✓ Search: AI powers search engines on platforms like Amazon to quickly return relevant products based on user queries. ✓ Recommendations: AI generates personalized product recommendations through artificial neural networks, showing suggestions like “customers who viewed this item also viewed” and “customers who bought this item also bought”. 65  Mobile Use: ✓ Voice-to-Text: Converts spoken words into text using AI, a standard feature on smartphones. ✓ Smart Personal Assistants: AI-driven assistants perform tasks through voice commands: ✓ Siri (Apple) and Google Assistant (Google): Perform internet searches, set reminders, and integrate with calendars. ✓ Alexa (Amazon): An AI-powered personal assistant that creates to-do lists, orders items online, sets reminders, and answers questions. Integrated with Echo and Dot smart speakers for home automation, music playback, and more. ✓ Cortana (Microsoft): AI assistant pre-loaded on Windows computers and Microsoft smartphones, performing similar tasks as Siri and Google Assistant. 66 End of Chapter Three! 67