6.3.3 | ARTIFICIAL INTELLIGENCE

ON THIS PAGE
6.3.3 Explain the basic operation and components of AI systems to simulate intelligent behaviour
 Limited to: 
– expert systems
– machine learning 
• Expert systems have a knowledge base, a rule base, an inference engine and an interface
• Machine learning is when a program has the ability to automatically adapt its own processes and/or data

ALSO IN THE TOPIC
 6.1.1 SENSORS, MICROPROCESSORS AND ACTUATORS
6.1.2 AUTOMATED SYSTEMS IN ACTION
6.2.1 ROBOTICS
6.2.2 CHARACTERISTICS OF A ROBOT
 6.2.3 ROBOT ROLES
 6.3.1 WHAT IS AI
​6.3.2 CHARACTERISTICS OF AI
YOU ARE HERE | ​6.3.3 OPERATIONS AND CONTROLS OF AI
AUTOMATED SYSTEMS TERMINOLOGY
AUTOMATED SYSTEMS ANSWERS

---

HOW AI EMULATES INTELLIGENT BEHAVOIUR

Artificial Intelligence (AI) is often hailed as the frontier of technological innovation, but at its core, it's about simulating the cognitive functions we associate with human minds. Here's a brief overview of how AI systems emulate intelligent behaviour.

• Learning from Experience | Just as humans learn from experiences, AI systems learn from data. The more data an AI system processes, the better it gets at making predictions or decisions. This is particularly evident in machine learning, a subset of AI, where algorithms adjust themselves in response to patterns in data.
• Problem Solving | AI systems are designed to tackle complex problems, sometimes problems that are too vast or intricate for human minds. For instance, analysing vast datasets to find patterns or optimizing routes for delivery trucks across a city.
• Reasoning and Decision Making | AI can simulate human reasoning. Expert systems, a type of AI, use a 'knowledge base' and a 'rule base' to mimic the decision-making abilities of a human expert. They draw conclusions based on predefined rules and knowledge.
• Perception | AI systems can interpret the world around them by recognizing objects, speech, and text. Computer vision, for instance, allows machines to interpret and make decisions based on visual data.
• Interaction | Chatbots and virtual assistants like Siri or Alexa simulate human interaction. They understand natural language, respond to queries, and even recognize emotions in some advanced applications.
• Adaptation |AI systems can adapt to changes in their environment or in the data they process. For example, a recommendation system might suggest different products as it learns more about a user's preferences.
• Automation of Tasks | AI can perform repetitive tasks, analyse large datasets, and even create content, simulating human efficiency but at a much larger scale and speed.

AI systems are intricate assemblies of data, algorithms, and computational power. Expert systems and machine learning are two key concepts that AI is built upon. These components, when harmoniously integrated, enable machines to perceive, reason, learn, and interact in ways that were once the exclusive domain of humans. As technology advances, these components are continually refined, making AI systems more efficient, accurate, and versatile.

SECTION 1 | COMPONENTS OF AN AI SYSTEM

AI systems can simulate intelligent behaviour by using various components and methods. Some of the basic components and operations of AI systems are:

• Data | Data is the raw material that AI systems use to learn and improve. Data can come from various sources, such as sensors, cameras, text, speech, or the web. Data can be structured (such as numbers, tables, or graphs) or unstructured (such as images, videos, or natural language). AI systems need large amounts of data to train their algorithms and models.
• Algorithms | Algorithms are the set of rules or instructions that AI systems follow to process data and perform tasks. Algorithms can be based on logic, mathematics, statistics, or probability. Algorithms can be simple (such as sorting or searching) or complex (such as machine learning or deep learning).
• Models |Models are the representations or abstractions of reality that AI systems use to make predictions or decisions. Models can be based on data, algorithms, or both. Models can be descriptive (such as clustering or classification) or generative (such as regression or synthesis).

SECTION 2 | OPERATIONS OF AN AI SYSTEM

While these are not mutually exclusive categories, as some components can also be involved in some operations. For example, algorithms can be used for learning, reasoning, or problem-solving. However, operations could come under the following 5 categories.

• Learning | Learning is the process by which AI systems improve their performance and adapt to new situations by using data and feedback. Learning can be supervised (where the system is given the correct answers or outcomes), unsupervised (where the system finds patterns or structures in the data), or reinforcement (where the system learns from its own actions and rewards).
• Reasoning | Reasoning is the ability of AI systems to draw conclusions or make inferences from data and models. Reasoning can be deductive (where the system applies general rules to specific cases), inductive (where the system infers general rules from specific cases), or abductive (where the system finds the best explanation for a given situation).
• Problem-solving | Problem-solving is the ability of AI systems to find solutions to complex or novel problems by using data, models, learning, and reasoning. Problem-solving can involve planning (where the system devises a sequence of actions to achieve a goal), search (where the system explores different options to find the best one), or optimization (where the system finds the optimal solution among many alternatives).
• Perception | Perception is the ability of AI systems to sense and interpret information from the environment. Perception can involve vision (where the system recognizes objects, faces, or scenes), speech (where the system understands spoken language or generates speech), audio (where the system analyses sounds or music), or touch (where the system feels textures or shapes).
• Language | Language is the ability of AI systems to communicate with humans or other machines using natural language. Language can involve understanding (where the system comprehends written or spoken language), generation (where the system produces written or spoken language), translation (where the system converts one language to another), or dialogue (where the system engages in a conversation with a user)

SECTION 3 | EXPERT SYSTEMS

Expert systems are a branch of artificial intelligence (AI) designed to simulate the decision-making ability of a human expert. These systems use a combination of knowledge and inference to solve problems in a specific domain, such as medical diagnosis, engineering, or troubleshooting technical issues.

Features of Expert Systems

1. Knowledge Base:
   
   • Stores facts and rules about a particular subject area.
     
   • Example: Symptoms and diagnoses for a medical expert system.
     
2. Inference Engine:
   
   • Processes the information in the knowledge base to draw conclusions.
     
   • Uses logical reasoning to apply rules and deduce new information.
     
3. User Interface:
   
   • Allows users to interact with the expert system.
     
   • May include graphical or text-based interfaces for inputting data and viewing results.
     

How Expert Systems Work

1. Input: The user provides data or answers to questions.
   
2. Processing: The inference engine matches the input with rules in the knowledge base.
   
3. Output: The system provides advice, recommendations, or solutions based on the matched rules.
   

Applications of Expert Systems

1. Medical Diagnosis:
   
   • Systems like WebMD provide possible diagnoses based on symptoms entered by users.
     
2. Technical Support:
   
   • Used to troubleshoot problems in IT or engineering systems.
     
3. Financial Decision-Making:
   
   • Helps evaluate investment opportunities or assess creditworthiness.
     
4. Agriculture:
   
   • Recommends farming practices based on soil conditions and weather.
     

Advantages of Expert Systems

• Consistency: Provides uniform advice based on established rules.
  
• Efficiency: Handles large amounts of data quickly.
  
• Availability: Accessible 24/7.
  
• Cost-Effective: Reduces the need for a human expert in repetitive tasks.
  

Limitations of Expert Systems

• Lack of Creativity: Cannot think outside the predefined rules.
  
• Maintenance: Requires regular updates to remain accurate.
  
• Dependence on Data Quality: Relies on a well-structured knowledge base.
  
• Limited Scope: Focused on specific domains, not general-purpose problem-solving.
  

Key Terminology

1. Knowledge Base: A repository of facts and rules used by an expert system.
   
2. Inference Engine: The component that applies logical reasoning to the knowledge base.
   
3. Rule-Based System: A system that uses "if-then" rules to make decisions.
   
4. User Interface: The medium through which users interact with the system.
   
5. Heuristics: Techniques used to make decisions based on experience rather than formal logic.
   

SECTION 4 | MACHINE LEARNING

Machine learning (ML) is a subset of artificial intelligence that enables systems to learn and improve from experience without being explicitly programmed. ML focuses on developing algorithms that can analyze data, identify patterns, and make decisions or predictions.

Features of Machine Learning

1. Data-Driven:
   
   • Relies on large datasets to identify patterns and improve accuracy.
     
   • Example: Training a system to recognize images of cats using thousands of labeled cat pictures.
     
2. Algorithm-Based:
   
   • Uses algorithms to process data and make predictions or classifications.
     
   • Examples include decision trees, neural networks, and support vector machines.
     
3. Continuous Improvement:
   
   • Learns and adapts over time as new data becomes available.
     

How Machine Learning Works

1. Data Collection: Gather data relevant to the problem.
   
2. Training: Use a dataset to train the model by feeding input-output pairs.
   
3. Testing: Evaluate the model's performance on unseen data.
   
4. Prediction: Deploy the trained model to make predictions or classifications on real-world data.
   

Types of Machine Learning

1. Supervised Learning:
   
   • The system learns from labeled data.
     
   • Example: Predicting house prices based on features like size and location.
     
2. Unsupervised Learning:
   
   • The system identifies patterns in data without labels.
     
   • Example: Grouping customers into segments based on purchase behavior.
     
3. Reinforcement Learning:
   
   • The system learns by interacting with an environment and receiving feedback.
     
   • Example: Training a robot to navigate a maze by rewarding successful moves.
     

Applications of Machine Learning

1. Image Recognition:
   
   • Identifying objects, faces, or handwriting in images.
     
2. Natural Language Processing (NLP):
   
   • Understanding and generating human language, such as in chatbots and translation services.
     
3. Recommendation Systems:
   
   • Suggesting products, movies, or music based on user preferences.
     
4. Autonomous Vehicles:
   
   • Enabling self-driving cars to make decisions based on sensor data.
     

Advantages of Machine Learning

• Adaptability: Can handle diverse tasks and improve with experience.
  
• Automation: Reduces the need for manual intervention in repetitive tasks.
  
• Scalability: Can analyze vast amounts of data efficiently.
  

Limitations of Machine Learning

• Data Dependency: Requires large amounts of high-quality data.
  
• Computationally Intensive: Needs significant processing power and resources.
  
• Bias Risk: Can inherit biases present in the training data.
  
• Black Box Nature: Difficult to understand how some models make decisions.
  

Key Terminology

1. Algorithm: A set of rules for solving a problem or completing a task.
   
2. Model: A mathematical representation of data used for predictions or classifications.
   
3. Training Data: Data used to teach a machine learning model.
   
4. Testing Data: Data used to evaluate a model’s performance.
   
5. Overfitting: When a model performs well on training data but poorly on unseen data.
   

SECTION 5 | DIFFERENCES AND INTERPLAY

While expert systems and machine learning are distinct technologies, they often complement each other in the development of advanced AI models. Understanding their differences and how they work together is crucial in grasping the full potential of artificial intelligence.

Key Differences

1. Knowledge Source:
   
   • Expert systems rely on a predefined knowledge base, created by human experts.
     
   • Machine learning models derive knowledge from data, identifying patterns and relationships.
     
2. Decision-Making Process:
   
   • Expert systems use rule-based logic to make decisions.
     
   • Machine learning systems use statistical models and algorithms to predict outcomes.
     
3. Adaptability:
   
   • Expert systems are static and require manual updates.
     
   • Machine learning systems improve and adapt as new data is introduced.
     
4. Scope:
   
   • Expert systems are focused on solving specific domain problems.
     
   • Machine learning can handle a broader range of tasks, including those without clear rules.
     

How They Work Together

1. Enhancing Expert Systems with Machine Learning:
   
   • Machine learning can be used to update and expand the knowledge base of an expert system by analyzing new data.
     
   • Example: A medical expert system can integrate ML models to refine diagnosis rules based on patient data trends.
     
2. Hybrid Systems:
   
   • Combining rule-based reasoning with ML algorithms allows systems to leverage the strengths of both approaches.
     
   • Example: Fraud detection systems use expert-defined rules for known fraud patterns and ML models to identify emerging threats.
     
3. Feedback Loops:
   
   • Machine learning can analyze the performance of an expert system and suggest improvements to its rule base.
     

Applications of Combined Systems

1. Healthcare:
   
   • Expert systems provide initial diagnostic support, while ML analyzes patient histories to refine recommendations.
     
2. Customer Support:
   
   • Chatbots use expert systems for predefined responses and ML for understanding complex queries.
     
3. Financial Services:
   
   • Expert systems handle rule-based compliance checks, while ML predicts market trends and risks.
     

Advantages of Integration

• Accuracy: Improved decision-making by combining human expertise with data-driven insights.
  
• Scalability: Ability to handle large and diverse datasets efficiently.
  
• Adaptability: Hybrid systems can evolve with changing requirements and data trends.
  

Expert systems and machine learning each play unique roles in AI. While expert systems bring structured, rule-based reasoning, machine learning provides adaptability and scalability through data analysis. Together, they create powerful AI models capable of solving complex problems in diverse domains.

ALSO IN THIS TOPIC

 6.1.1 SENSORS, MICROPROCESSORS AND ACTUATORS
6.1.2 AUTOMATED SYSTEMS IN ACTION
6.2.1 ROBOTICS
6.2.2 CHARACTERISTICS OF A ROBOT
 6.2.3 ROBOT ROLES
 6.3.1 WHAT IS AI
​6.3.2 CHARACTERISTICS OF AI
6.3.3 OPERATIONS AND CONTROLS OF AI
AUTOMATED SYSTEMS TERMINOLOGY
AUTOMATED SYSTEMS ANSWERS