Artificial Intelligence: Core Concepts

Artificial Intelligence (AI) aims to create machines that mimic human intelligence, enabling them to learn, reason, and make decisions. This cheat sheet covers AI definitions, types, working principles, and key applications.

Core Principles

• AI enables machines to think and act like humans.
• AI learns from data using algorithms and decision-making models.
• Key AI processes include Learning, Reasoning, and Self-Correction.
• AI relies on Deep Learning and Natural Language Processing (NLP).
• AI systems can be categorized by capabilities (Weak/Strong) and functionality (Reactive/Limited Memory).
• Environments for AI range from discrete and known to continuous and unknown.
• Search algorithms like DFS, BFS, and UCS explore problem spaces differently.
• Prompt engineering is crucial for effective human-AI communication.
• Machine learning approaches include Supervised, Unsupervised, Reinforcement, and Hybrid learning.

Key Terms

• Artificial Intelligence (AI): A branch of computer science focused on creating machines capable of tasks requiring human intelligence.
• Learning: AI systems acquiring knowledge from data using algorithms and patterns.
• Reasoning: AI applying logic to make decisions and evaluate possibilities.
• Self-Correction: AI systems analyzing errors and improving accuracy over time.
• Deep Learning: A subset of ML using neural networks to mimic the brain.
• Natural Language Processing (NLP): AI's ability to understand and generate human language.
• Weak AI (Narrow AI): AI designed for a specific task within a limited context.
• Strong AI (AGI): AI aiming to replicate human-level cognitive abilities across domains.
• Prompt Engineering: The technique of crafting inputs to guide AI models for desired outputs.
• Supervised Learning: ML using labeled datasets to train models.
• Unsupervised Learning: ML using unlabeled data to find hidden patterns.
• Reinforcement Learning: AI learning through trial and error with rewards/penalties.

Real World Examples

• Self-Driving Cars: Utilize AI for learning (data from sensors), reasoning (decision-making), and self-correction (improving navigation).
• Virtual Assistants (Siri, Alexa): Examples of Weak AI using NLP to understand and respond to voice commands.
• Recommendation Systems (Netflix, Amazon): Use AI to analyze user behavior and suggest content.
• Chess-playing AI (Deep Blue): A Reactive Machine AI that plays based on current board state and programmed rules.
• Email Spam Detection: Uses machine learning to classify emails based on learned patterns.
• Image Recognition: Employs Convolutional Neural Networks (CNNs) to process visual data.
• Chatbots: Leverage NLP for understanding and generating human-like text responses.

Timeline

• 1950: Alan Turing proposes the Turing Test.
• 2004: John McCarthy defines AI as the 'science and engineering of making intelligent machines'.
• N/A: Development of Weak AI (Narrow AI) systems like virtual assistants.
• N/A: Advancements in Deep Learning and Neural Networks.
• Present: Active research in Strong AI (AGI) and sophisticated prompt engineering.