π Topic 01 of 6 Β· Chapter 06 Β· AI & Robotics
AI Fundamentals β ML, Deep Learning, NLP
Narrow AI, ML, deep learning, neural networks, NLP, computer vision β complete notes for UPSC & PSC exams.
π€ AI β Types
| Type | Description | Example | Status |
|---|---|---|---|
| Narrow AI (ANI) | Designed for specific tasks; cannot generalise | ChatGPT, AlphaGo, Siri, facial recognition | Exists today |
| General AI (AGI) | Human-level intelligence across all domains | β | Does not exist yet |
| Superintelligence (ASI) | Surpasses human intelligence in all areas | β | Hypothetical; future concern |
π€ AI Hierarchy
- Artificial Intelligence (AI) β broad field; machines performing human-intelligence tasks
- Machine Learning (ML) β subset of AI; systems learn from data without explicit programming
- Deep Learning β subset of ML; uses multi-layered artificial neural networks
- Natural Language Processing (NLP) β AI understanding human language; ChatGPT, Google Translate
- Computer Vision β AI interpreting visual information; facial recognition, medical imaging
- Reinforcement Learning β AI learns by trial and error; reward-based; AlphaGo, autonomous vehicles
π€ Neural Networks
- Inspired by the human brain β interconnected nodes (neurons)
- Input layer β Hidden layers β Output layer
- Deep Learning = many hidden layers = “deep” neural network
- Applications: image recognition, speech recognition, language translation, drug discovery
- Convolutional Neural Networks (CNNs) β for image processing
- Recurrent Neural Networks (RNNs) β for sequential data (text, speech)
- Transformers β architecture behind ChatGPT, BERT, GPT-4
π Key Exam Points:
β’ AI > ML > Deep Learning (hierarchy)
β’ Narrow AI = specific tasks = exists today; AGI = human-level = doesn’t exist
β’ ChatGPT = OpenAI = Large Language Model (LLM) = NLP = Narrow AI
β’ AlphaGo = DeepMind = defeated world Go champion (2016) = Reinforcement Learning
β’ Turing Test = Alan Turing (1950) = test for machine intelligence
β’ AI > ML > Deep Learning (hierarchy)
β’ Narrow AI = specific tasks = exists today; AGI = human-level = doesn’t exist
β’ ChatGPT = OpenAI = Large Language Model (LLM) = NLP = Narrow AI
β’ AlphaGo = DeepMind = defeated world Go champion (2016) = Reinforcement Learning
β’ Turing Test = Alan Turing (1950) = test for machine intelligence