Generative AI is a branch of artificial intelligence that creates new content (text, images, audio, video or code) instead of only classifying or predicting from existing data.
How it works
Most generative models learn patterns from very large collections of examples. A large language model (LLM), for instance, is trained to predict the next word in a sentence. Done at enormous scale, that simple task produces a model that can write, summarise, translate and reason through problems.
Image generators typically use diffusion models: they learn to turn random noise step by step into a picture that matches a text description.
What it is good at
- Drafting and summarising documents, emails and reports.
- Writing and explaining code.
- Answering questions over a company's own documents.
- Creating design concepts, marketing images and prototypes quickly.
Where it goes wrong
- It can state false information confidently, which is often called hallucination.
- It reflects biases in its training data.
- It needs careful handling of private and copyrighted data.
That is why organisations need people who understand both how to build with these models and how to evaluate them.
How to start learning
Start with Python and the basics of machine learning, then learn how LLMs work, prompt design, retrieval-augmented generation (RAG) and evaluation. Building small projects, such as a chatbot over your own notes, teaches more than reading alone.