Learn it
Artificial intelligence is computers doing tasks that seem to need intelligence. Most modern AI is machine learning: instead of writing every rule, we give examples (data) and the computer finds patterns. AI can be very useful and also make mistakes or reflect bias in its data.
🧰 Tool: any free online Python editor such as Online Python or Trinket, or install Python from python.org and use IDLE.
Code examples
Rule-based vs learning
# Rule-based: a human writes the rules
def is_spam_rules(message):
return "free money" in message.lower() or "click now" in message.lower()
# Learning-based (idea): count words seen in labelled examples
examples = [("win free money now", True), ("see you at practice", False),
("click now for a prize", True), ("homework is due friday", False)]
spam_words = {}
for text, is_spam in examples:
for word in text.split():
spam_words[word] = spam_words.get(word, 0) + (1 if is_spam else -1)
def is_spam_learned(message):
return sum(spam_words.get(w, 0) for w in message.lower().split()) > 0
print(is_spam_learned("free prize now")) # True
print(is_spam_learned("practice friday")) # FalseProject: AI spotter
- List 5 places you meet AI (recommendations, maps, voice assistants…).
- For each, guess what data it learned from.
- Add 4 more labelled examples to the spam learner and test it.
Challenge: Find a message that fools the learner and explain why.