Types of AI; Main Domains of AI Technology

Types of AI

AI can be classified in two major ways:

1. Based on capabilities (what it can do)

Based on functionality / consciousness (how it thinks)

Classification: Based on Capabilities

7.1 Reactive Machines / Purely Reactive

• Basic form of AI.
• Operate purely based on present data — no memory, no learning.
• Respond to specific inputs with fixed outputs — cannot adapt.
• Cannot form memories or use past experiences.

Examples:

• IBM Deep Blue (1997) — defeated Garry Kasparov in chess. Could identify pieces and make predictions, but could not store memories or learn from past games.

Google AlphaGo (early version) — used pattern recognition without learning from previous games.

Characteristics:

Input → Fixed Response
(No past memory, No future learning)

7.2 Limited Memory

• Can learn from past data to improve future responses.
• Most modern AI applications fall under this category.
• Uses historical data for decisions but has no long-term memory — erased after use.
• Machine learning models, autonomous systems and robotics rely on limited memory.

Examples:

• Self-driving cars — observe roads, traffic signs, nearby cars; make decisions based on past experiences and current conditions.
• Chatbots — remember recent conversations to improve flow and relevance of replies.
• Recommendation engines — remember recent browsing history.

Characteristics:

Past Data + Current Input → Decision
(Short-term memory only)

7.3 Theory of Mind

• Aims to understand human emotions, beliefs, intentions, desires and interact socially.
• Still in development — no complete machine built yet.
• Would need to understand that people and creatures have thoughts and emotions affecting behavior.

Potential Applications:

• Human-robot interaction detecting emotions and empathizing

Collaborative healthcare robots understanding patient distress

• AI companions for elderly care

Fictional examples: Avengers (JARVIS), I Robot, Transformers, Ra.One

7.4 Self-Awareness

• Advanced stage of AI — self-consciousness and awareness.
• Would understand and react to emotions AND have its own consciousness.
• Ultimate goal of AI — not yet achieved.

Potential Applications:

• Autonomous systems making moral and ethical decisions
• AI pursuing its own goals based on world understanding

Capability Progression

Level 1: Reactive Machines   → No memory     (Deep Blue)
Level 2: Limited Memory      → Short memory  (Self-driving car, ChatGPT)
Level 3: Theory of Mind      → Understands emotions (In research)
Level 4: Self-Awareness      → Conscious AI  (Future goal)

Classification: By Intelligence Level (Narrow → General → Super)

Narrow AI Examples: Siri, Alexa, Google Search, Face ID, ChatGPT (for specific tasks) AGI: Would pass the Turing Test for any task; hypothetical Super AI: The "technological singularity" — immense theoretical concern

Main Domains of AI Technology

                +-----------------------------------------+
                |       ARTIFICIAL INTELLIGENCE           |
                +-----------------------------------------+
                    |              |               |
              +-----+         +---+---+       +---+---+
              |               |               |
         DATA SCIENCE     COMPUTER         NATURAL
                           VISION        LANGUAGE
                                        PROCESSING

8.1 Data Science

Data Science is an inter-disciplinary field that uses scientific methods, processes, algorithms and systems to extract knowledge and insights from both structured and unstructured data.

Related to data mining, machine learning, big data.

Key Techniques:

• Statistical analysis and probability
• Machine learning and deep learning
• Data visualization (charts, dashboards)
• Big data processing (Hadoop, Spark)

Database management (SQL, NoSQL)

Data Science Application Examples:

Data Science Lifecycle:

Define Problem → Collect Data → Clean Data → EDA → Model → Evaluate → Deploy → Monitor

8.2 Computer Vision

Computer Vision enables machines to interpret and make decisions based on visual inputs (images and videos). Uses deep learning (primarily CNN).

How it works:

Image Input → Preprocessing → Feature Extraction (CNN) → Classification/Detection → Output

Applications:

8.3 Natural Language Processing (NLP)

NLP is an AI method for communicating with intelligent systems using a natural human language (e.g., English).

NLP is a sub-field of AI where computers can understand and process human language. Its objectives are:

• Read — understand written text
• Decipher — interpret meaning beyond words
• Understand — comprehend context and intent
• Make sense — extract actionable insightss