Artificial intelligence (AI) has many definitions. When I asked Bing “What is artificial intelligence?”, it responded with a concise summary: “Artificial intelligence enables machines to think and act like humans.” Although this specific phrase wasn’t directly quoted in any of its five cited references, it effectively synthesized the information into a clear explanation.
Another excellent definition comes from Andrew Ng, founder of Deep Learning and a longtime AI expert formerly at Google, who describes AI as “a huge set of tools for making computers behave intelligently.”
The AI Landscape
AI isn’t one single technology but rather a collection of various tools and approaches working together. For example, Machine Learning is a broad category within AI. It includes subcomponents:
- Deep Learning: A specialized subset of machine learning
- Large Language Models: Systems trained on vast text data
These technologies are components in the larger AI ecosystem.
When we discuss concepts like cognitive AI and generative AI, we’re actually referring to combinations of these tools that together create specific AI capabilities.
Real-World Applications Require Multiple AI Technologies
Consider what makes a self-driving vehicle function. There are multiple technologies that must work together, including:
- Machine Vision: Allows the car to “see” its surroundings
- Image Recognition: Identifies objects like road signs, lane markings, and other vehicles
- Deep Learning: Processes and extracts meaning from visual information
- Robotics: Controls the physical movements of the vehicle
Similarly, a chatbot combines multiple technologies:
- Speech-to-Text (required for voice bots): Converts spoken questions to text
- Intent Recognition: Determines what you’re really asking about (billing questions, technical support, etc.)
- Information Retrieval: Researches or accesses relevant information
- Natural Language Generation: Creates human-like responses
Cognitive AI: Thinking Like Humans
Cognitive AI refers to systems that think and interact more like humans. Key characteristics include:
- Programming through natural language rather than code
- Generating human-like responses
- Not following explicit “if-then” programming
Generative AI: Creating New Content
Generative AI goes further by combining existing information to create something new. However, it’s important to remember that these systems still rely entirely on their training data. They can only work with information they’ve been exposed to.
A good example is Google’s soccer-playing robots. Instead of programming specific soccer moves, engineers simply gave the robots a goal (to score) and constraints (must use feet only, must follow game rules). The robots had to figure out how to play on their own, and they did this successfully.
Summary
AI is not one thing. It’s a collection of interconnected tools rather than a single technology. These tools that can used together in multiple ways to automate existing processes and create new capabilities.
