A rtificial intelligence Revision Notes

Artificial Intelligence (AI) is a specialised area within computer science focused on designing computer systems that can perform tasks usually requiring human intelligence. Unlike traditional programming, where explicit instructions are given for every action, AI systems aim to replicate cognitive processes such as learning from data, reasoning through problems, and adapting to new situations.

For example, a computer program that can play chess by evaluating possible moves and predicting outcomes demonstrates AI by mimicking human strategic thinking.

• AI is a subfield of computer science.
• It focuses on creating systems that simulate intelligent human behaviour.
• AI systems perform tasks that normally require human cognition.

AI systems are designed to imitate various aspects of human intelligence. These include: learning from experience, understanding natural language, recognising patterns, reasoning logically, and making decisions.

For instance, a voice assistant that understands spoken commands and responds appropriately uses AI techniques such as natural language processing and pattern recognition.

• Learning: AI systems improve performance by analysing data.
• Reasoning: AI can draw conclusions from available information.
• Perception: AI can interpret sensory inputs like images or sounds.
• Language understanding: AI can process and generate human language.

AI technologies are applied in many areas to automate complex tasks. Examples include:

Autonomous vehicles that interpret sensor data to navigate safely.

Fraud detection systems that analyse transaction patterns to identify suspicious activity.

• AI enables automation of tasks requiring intelligence.
• Applications span robotics, data analysis, and natural language processing.
• These systems often improve over time through learning.

Traditional computer programs follow fixed instructions and cannot adapt beyond their programming. In contrast, AI systems can learn from data and improve their performance without explicit reprogramming.

For example, a spam filter using AI can learn to identify new types of unwanted emails by analysing examples, whereas a traditional filter relies on fixed rules.

• Traditional programs execute fixed instructions.
• AI systems adapt and learn from data.
• Learning ability is a key feature of AI.

AI represents a significant advancement in computer science by enabling machines to perform tasks that require intelligence. Understanding AI helps in developing systems that can assist or replace humans in complex tasks, improving efficiency and opening new technological possibilities.

Recognising AI as a field helps students appreciate the challenges and techniques involved in creating intelligent systems.

• AI is a major field within computer science.
• It focuses on replicating intelligent human behaviour.
• AI systems have broad applications and impact.

Data collection is the foundational step for any AI system. It involves gathering relevant and sufficient information from various sources such as sensors, databases, or user inputs. For example, a voice assistant collects audio data from a microphone to process spoken commands. The quality and quantity of data directly affect the AI's performance, as incomplete or biased data can lead to incorrect outputs.

• Data must be relevant and representative.
• Sources include sensors, databases, and user inputs.
• Example: Image recognition AI collects thousands of labelled images.

Once data is collected, AI systems apply specific rules or algorithms to process and interpret this data. These rules define how data is filtered, transformed, and used to generate outputs. For instance, an AI in a spam filter uses rules to identify keywords and patterns that indicate unwanted emails. Rules ensure that data is handled consistently and ethically, preventing misuse or incorrect processing.

• Rules guide data processing and interpretation.
• They ensure ethical and effective use of data.
• Example: Filtering rules in email spam detection.

Reasoning allows AI to make decisions or draw conclusions based on processed data. This involves logical inference, pattern recognition, and problem-solving. For example, a chess AI analyses the current board state and reasons about possible moves to select the best strategy. Reasoning enables AI to simulate human-like decision-making rather than just following fixed instructions.

• Enables decision-making based on data.
• Involves logical inference and pattern recognition.
• Example: AI recommending products based on user behaviour.

A key feature distinguishing AI from traditional programs is its ability to learn from experience and adapt its behaviour. Machine learning techniques allow AI to improve performance by recognising patterns in data without explicit programming for every scenario. For example, a self-driving car AI learns to better identify pedestrians by analysing thousands of driving situations. This continuous adaptation helps AI systems handle new or changing environments effectively.

• AI improves through experience and data exposure.
• Learning methods include supervised and unsupervised learning.
• Example: Email filters updating rules based on user feedback.

Artificial intelligence systems are designed to replicate aspects of human intelligence by processing information and making decisions or predictions. These systems simulate intelligent behaviour by combining data inputs, processing mechanisms, and output generation. Understanding the components that make up AI systems is essential to grasp how they function in real-world applications.

• AI systems simulate human-like decision-making.
• They process data using algorithms to produce outputs.
• Components include data input, processing algorithms, and output mechanisms.

Expert systems are a type of AI that simulate the decision-making ability of a human expert in a specific domain. They rely on a knowledge base containing facts and rules derived from human expertise. An inference engine applies logical rules to the knowledge base to deduce new information or make decisions.

For example, an expert system in medical diagnosis might contain rules such as 'If the patient has a fever and a sore throat, then consider the possibility of a throat infection.' The inference engine uses these rules to infer conclusions based on the patient's symptoms.

Expert systems do not learn from new data; they operate strictly within the rules and facts programmed into their knowledge base.

• Knowledge base stores domain-specific facts and rules.
• Inference engine applies logical reasoning to the knowledge base.
• Decisions are made by matching facts to rules.
• Example: Medical diagnosis system using symptom-based rules.

Machine learning is an AI approach where systems improve their performance by learning from data rather than relying solely on predefined rules. These systems use algorithms that identify patterns in large datasets and adjust their internal parameters to make better predictions or decisions over time.

For instance, a machine learning system for email spam detection is trained on thousands of emails labelled as 'spam' or 'not spam.' The system learns features that distinguish spam emails and uses this knowledge to classify new emails automatically.

Unlike expert systems, machine learning models can adapt to new data, improving accuracy as more data becomes available.

• Learns patterns from training data using algorithms.
• Adjusts internal parameters to improve performance.
• Can generalise to new, unseen data.
• Example: Spam email filter trained on labelled emails.

Both expert systems and machine learning rely on several fundamental components to function effectively: data, algorithms, and computational resources.

Data: The raw information input into the system. Expert systems use structured facts and rules, while machine learning requires large datasets for training.

Algorithms: Procedures or formulas that process data. Expert systems use logical inference algorithms; machine learning uses statistical and optimisation algorithms to find patterns and improve models over time.

• Data provides the basis for decision-making or learning.
• Algorithms process data to simulate intelligence.
• Computational power enables handling complex calculations.
• Expert systems use rule-based algorithms; machine learning uses adaptive algorithms.

Expert systems and machine learning represent two distinct approaches to AI. Expert systems are transparent and interpretable because their rules are explicit, but they lack flexibility and cannot learn from new data. Machine learning systems are flexible and improve with experience but can be opaque, making their decision processes harder to interpret.

Choosing between these approaches depends on the application requirements, such as the need for explainability or adaptability.

• Expert systems: rule-based, no learning, transparent.
• Machine learning: data-driven, learns from data, less transparent.
• Trade-off between interpretability and adaptability.
• Application context determines suitable AI approach.