Start with the big idea
Artificial intelligence (AI) is a broad label for computer systems designed to perform tasks that normally require abilities we associate with human intelligence — such as recognising patterns, understanding language, making predictions or generating content.
AI is not one single technology. It is an umbrella covering many approaches, from rule-based software to modern machine-learning systems.
Artificial intelligence is the broad field. Machine learning is one way to build AI. Deep learning is a type of machine learning using multi-layer neural networks. Generative AI uses learned patterns to create new outputs such as text, images, audio or code.
What is machine learning?
Traditional programming often works like this: a developer writes explicit rules, gives the program data, and the program produces an answer. Machine learning changes the emphasis. Instead of hand-writing every rule, we provide examples or data and an algorithm adjusts internal parameters to find useful patterns.
After training, the model can apply those learned patterns to new inputs. A spam filter, recommendation system or image classifier can all use this general idea.
Where neural networks fit
A neural network is a mathematical model made of connected layers of computations. During training, the strengths of those connections are adjusted so the network gets better at a task. Modern deep-learning systems use many layers and can contain huge numbers of learned parameters.
The name is inspired by biological neurons, but artificial neural networks are simplified mathematical systems — they are not digital replicas of a human brain.
What makes generative AI different?
Many older AI systems mainly classify or predict: “Is this email spam?” or “Which product might this person like?” Generative AI can produce new content. Depending on the model, that could be text, images, video, audio, code or other data.
A useful way to think about it is: the model has learned statistical structure from training data, and it uses that structure to construct a new output in response to an input.
What AI can and cannot do
AI is useful for
- Summarising and transforming information
- Pattern recognition and prediction
- Drafting content and code
- Brainstorming and exploration
- Automation when connected to tools
AI still needs care because
- Outputs can be wrong or fabricated
- Models may miss context
- Training data can contain bias
- High-stakes decisions need human oversight
- Fluent language is not proof of truth
The one sentence to remember
Modern AI is largely about building models that learn useful patterns from data and then use those patterns to predict, classify, generate or act.