What Is AI? — The Core Technical Ideas
Machine learning, neural networks, and large language models explained
The Core Technical Ideas
In Part 1, we defined AI as systems that learn from data rather than following hardcoded rules. Now let’s look at the key technical
concepts that make that possible
Machine Learning
Machine learning (ML) is the engine behind most modern AI. It is a method of training a computer to recognize patterns in data and make predictions, without being explicitly programmed for each task.
There are three main approaches:
• Supervised learning: The model is trained on labeled examples — say, thousands of photos tagged as “cat” or “dog.” It learns the visual features associated with each label and can then classify new, unseen images.
• Unsupervised learning: The model is given data without labels and asked to find structure on its own, such as grouping similar customers together based on purchasing behavior.
• Reinforcement learning: The model learns by trial and error, receiving rewards or penalties for its actions. This is the approach behind game-playing AI and many robotics systems.
Neural Networks
A neural network is a type of machine learning model loosely inspired by the structure of the human brain. It is made up of layers of interconnected “neurons” — mathematical units that take in inputs, apply weights, and pass signals forward.
A simple network might have an input layer, one or more hidden layers that transform the data, and an output layer that produces the final prediction. During training, the network adjusts its internal weights millions of times until it reliably maps inputs to the correct outputs.
Neural networks are the foundation of deep learning, which simply refers to networks with many layers. The “deep” in deep learning describes the depth of the network, not the sophistication of the insight.
Large Language Models
Large language models (LLMs) are a specific kind of neural network trained on vast amounts of text. Their job is to predict the next word in a sequence — given “the cat sat on the,” the model learns that “mat” is a likely continuation.
This seemingly simple task, repeated at enormous scale, produces surprisingly capable systems. LLMs can write essays, answer questions, summarize documents, translate between languages, write code, and carry on conversations. The chatbots and AI assistants that have captured public attention in recent years are almost all powered by language models.
What makes them powerful is their ability to generalize across many tasks from a single training process, rather than being built for one specific job.
In Part 3, we will cover training and inference, generative AI, and the important limitations and considerations everyone should know.