Every AI term in this course, translated back to the stall. Listed A to Z.
AI agent
At the stall: The chef who runs the whole service.
AI that takes a goal, plans steps, uses tools, and acts, checking its own progress along the way, instead of answering one question.
Algorithm
At the stall: Picking the recipe.
The method a machine follows to learn from data. Same ingredients, different recipes, different dishes.
Attention
At the stall: Watching the right ingredient at the right moment.
The mechanism inside modern AI that weighs which parts of the input matter most for the word it is writing now.
Bias
At the stall: A stall that only learned from one neighbourhood.
Systematic unfairness in a model's answers, inherited from skewed training data. Bias in, bias out.
Classification
At the stall: Fresh prawn or spoiled prawn.
Teaching a machine to sort things into named groups from labelled examples. The most common job in machine learning.
Closed model
At the stall: Food delivery.
A model you can only use through its owner's service. You send a request, receive the result, and never see the kitchen.
Clustering
At the stall: The regulars sort themselves into groups.
Letting a machine find natural groups in data nobody has labelled. Nobody tells it the groups; it finds them.
Computer vision
At the stall: How a stall reads a photo.
Machines learning to recognize what is in an image, from the face that unlocks your phone to prawns on a sorting belt.
Data cleaning
At the stall: Washing, peeling, deveining.
Fixing and removing bad raw data: duplicates, errors, missing values, anything that would ruin the dish. Most of the work in a real AI project sits here.
Data collection
At the stall: Going to the wet market.
Gathering the raw information an AI learns from, from wherever it lives: databases, documents, sensors, websites.
Defining the problem
At the stall: Decide the dish before you shop.
Choosing exactly what you want the AI to do before gathering anything. A hawker does not fire the wok at random; the dish comes first.
Deployment
At the stall: Serving the plate.
Putting the finished model to work in the real world, where real people use it and real consequences follow.
Embodied AI
At the stall: The day the recipe gets hands.
AI connected to a physical body, with sensors and movement, that must survive gravity, heat, and slippery prawns.
Evaluation
At the stall: Taste and adjust.
Testing the trained model on examples it has never seen, the way a cook tastes before serving. Too bland means back to the wok.
Feature engineering
At the stall: Brewing the prawn stock.
Turning raw data into the concentrated signals a model actually learns from, the way prawn heads become stock. You never serve the heads; you serve the essence.
Fine-tuning
At the stall: Coaching the chef to cook your style.
Extra training that teaches an existing model a specific style, format, or task, using far less data and money than building one from scratch.
Foundation model
At the stall: The master chef, trained over years.
A very large AI, trained at great cost on enormous amounts of text or images, that arrives already skilled. Most AI work today starts from one and adapts it.
Generative AI
At the stall: The chef who invents new dishes.
AI that creates new text, images, or sound, rather than sorting or predicting what already exists.
Hallucination
At the stall: Confidently serving the wrong dish.
When an AI states something untrue with complete confidence, because it is predicting plausible words, not checking facts.
Large language model
At the stall: The chef who has read every recipe.
A model trained on enormous amounts of text, whose core skill is predicting the next word well enough to write whole passages.
Model drift
At the stall: The regulars' tastes slowly change.
What happens when the world changes and the model does not: slowly, its answers stop fitting. The fix is monitoring and retraining, not blame.
Model training
At the stall: Firing the wok.
The step where the machine studies the prepared data again and again, adjusting itself each pass until its answers stop improving. Shorter than everyone thinks, once the prep is done.
Moravec's paradox
At the stall: The wok hei problem.
The finding that tasks humans find hard, like chess, are easy for AI, while tasks we find easy, like walking or tossing a wok, are brutally hard. The skill is in the body.
Multi-agent system
At the stall: A team of specialists in one kitchen.
Several agents with different jobs passing work to each other, the way a wok chef, a soup cook, and an order-taker run one stall.
Neural network
At the stall: Many cooks, one wok.
A model built from many small, simple decision units stacked in layers. No single unit understands anything; together, they do.
Open source
At the stall: The open kitchen.
A model released with weights, training code, and data details, so anyone can rebuild or deeply change it. You watched it cook.
Open weights
At the stall: The dish plus the ingredients list.
A model you can download, run yourself, and even adjust, while the training data and full method stay secret.
Pre-training
At the stall: Culinary school.
The enormous first training run that gives a model its general skill. Expensive, slow, and usually someone else's job.
Prompting
At the stall: How you order changes what you get.
Writing the instruction you give a generative model. Small changes in wording change the answer, which is why it is a skill.
Regression
At the stall: How many plates on a rainy Friday.
Teaching a machine to predict a number from past patterns: sales, crowds, prices, temperatures.
Reinforcement learning
At the stall: Earning tips by trial and error.
Training by consequences: the machine tries things, earns rewards or penalties, and keeps what paid off. No instructions, only feedback.
Retrieval-augmented generation
At the stall: Handing the chef your recipe card.
Giving a model the exact documents it needs at the moment you ask, so it answers from your facts instead of guessing. Bring your own recipe.
Robot sensors
At the stall: The five senses of a robot chef.
Cameras, microphones, and force and heat sensors that let a robot perceive the physical world, each standing in for a sense a hawker never thinks about.
Sim-to-real
At the stall: Practising on a virtual stall first.
Training a robot in a simulation, where failing costs nothing and time runs fast, before it ever touches the real wok.