Definition
A street light switches on by itself every evening. Is it smart?
Automation is a machine carrying out fixed instructions without a person's help. Artificial intelligence goes further: the machine decides what to do in situations its instructions did not spell out, often by learning from examples.
In simple words
A washing machine runs the same programme every time you press the button. That is automation. A spam filter meets emails nobody has ever seen and still decides which ones are spam. That is AI. Automation repeats; AI adapts.
Why it matters
Many products are called 'AI' when they are only automation. Knowing the difference stops you being fooled by advertising, and helps you choose the right tool: a simple rule is cheaper and more predictable when the situation never changes.
How it works
The one-question test
Ask: can it deal with a case nobody wrote an instruction for? If it only ever does what was decided in advance, it is automation. If it handles new cases by itself, it uses AI.
| Fixed automation | Artificial intelligence |
|---|---|
| A timer switches the light on at 6 pm | A camera learns when the room is in use and switches lights on and off |
| A lift goes to the floor you press | A lift system learns rush hours and waits where people will need it |
| An ATM gives cash when the PIN is right | A bank system spots an unusual payment and asks you to confirm |
Try it yourself
Fixed rule against a learner
The same messages go to both. The rule never changes, and the learner improves after each answer. Who ends up ahead?
Fixed rule or learned behaviour?
Decide whether each machine follows a fixed rule or has learned from examples.
A music app suggests songs you end up liking
A keyboard predicts your next word
A photo app finds every picture of your grandmother
A water tank pump stops when the tank is full
A microwave stops after 2 minutes
A traffic signal changes every 60 seconds
Common mistakes
- Calling everything automatic 'AI'. Automatic and intelligent are not the same.
- Thinking AI is always the better choice. Fixed rules are often better when nothing changes.
- Thinking AI has no instructions at all. People still decide the goal, the data and the limits.
Check yourself
Summary
- Automation follows fixed instructions; AI handles situations the instructions did not cover.
- Test: can it deal with a case nobody wrote a rule for?
- Both are useful. Choose fixed rules when nothing changes, and AI when things keep changing.
Finished reading?
Sources: Russell & Norvig, Artificial Intelligence: A Modern Approach