The STEM lab: computing, data and AI

29 topics in four themes, with ideas pitched for every age. The lab also has four hands-on activities: binary lamps, program the robot, teach a machine and a qubit playground.

Computing

Instructions, loops, binary and how computers think.

Giving a robot precise instructions: Follow a list of instructions step by step to see where a robot ends up.
Loops: repeating instructions: Work out what a repeat loop does, and how many times things happen.
Inputs, processes and outputs: A program takes an input, does something to it, and gives an output. Work out the output, or the input.
Sorting step by step: Follow a simple sorting algorithm: swap neighbours that are in the wrong order.
Spotting bugs: Find what is wrong with a short program and choose the fix.
Binary numbers: Convert between binary and ordinary (denary) numbers, and add in binary.
Data types: Name the data type of a value: whole number, decimal, text or True/False.
Hexadecimal and colours: Read hexadecimal (base 16) numbers, as used for colours on screens.
Tracing code: Follow a short piece of code line by line and work out the final value of a variable.
AND, OR and NOT: Work out whether a logic statement is True or False.
Binary search: See how halving the list each time finds an item quickly in a sorted list.
How fast does the work grow?: Compare how the number of steps in an algorithm grows as the amount of data grows.
Storage and file sizes: Calculate how many bits and bytes are needed to store text, pictures and sound.
Secret codes: Caesar cipher: Encrypt and decrypt messages by shifting letters, and use remainders to wrap around the alphabet.
Information and bits: Measure information in bits: how many yes/no questions are needed to pin down one outcome.

Data science

Collecting, checking and making sense of data.

Data detective: Read a table of survey results and answer questions about it, like a data scientist.
Odd one out in data: Spot an outlier in a data set and see how much it changes the mean.

AI and machine learning

How machines learn from examples.

How computers learn from examples: Look at labelled examples and find the feature that tells two groups apart, like a machine learning program does.
Is the data fair?: Spot when data or a sample is unfair, and see why that matters for AI.
Nearest neighbour: how a machine decides: Use the nearest-neighbour idea to classify something new from labelled examples.
A tiny artificial neuron: Work out the output of a simple artificial neuron: multiply the inputs by weights, add them up, and compare.
How good is the model?: Work out how accurate a machine learning model is from its right and wrong answers.
Probability and the spam filter: Use counts to find the chance that an email is spam, given that it contains a word.
Learning by going downhill: Take a step of gradient descent, the method that trains neural networks.
A neural network layer: Calculate the output of a layer of a neural network using a weights table (matrix), inputs, bias and ReLU.

Quantum ideas

The strange, true rules of the very small.

Quantum ideas: Meet the big ideas of quantum physics: energy comes in packets, and nature at tiny scales works with probabilities.
Qubits and probability: Use the amplitudes of a qubit to find the chances of measuring 0 or 1.
How many states can qubits hold?: See how the number of possible measurement outcomes doubles with each extra qubit.
Quantum gates: Follow what the X and H gates do to a qubit, and work out the chance of each result.