💚

Beginner-Friendly Topic

Take your time - it's perfectly normal to re-read this topic 2-3 times. Try the interactive code editor below to run code yourself. Use the Q&A section to check your understanding before moving on. You've got this! 🚀

How an AI System Works

Every AI system follows a similar basic pattern: collect data, train a model, use the model to make predictions, and evaluate how well it works. Understanding this cycle helps you understand any AI system.

12 min•By Priygop Team•Updated 2026

The AI System Lifecycle

Here is the basic flow of any AI system:

1Collect Data: Gather examples of inputs and the correct answers
2Train the Model: The AI studies the data and learns patterns
3Evaluate: Test the model on new data to see how accurate it is
4Deploy: Put the model into a real application where users can use it
5Monitor and Improve: Watch how the model performs and update it over time

Let us walk through this with a real example.

Diagram

Loading diagram…

Deep Learning ⊂ Machine Learning ⊂ Artificial Intelligence

Example: Building an Email Spam Detector

Step 1 - Collect Data: Gather 10,000 emails labeled as spam or not spam
Step 2 - Train: The AI reads all 10,000 emails and learns which words and patterns appear more in spam
Step 3 - Evaluate: Test the trained AI on 2,000 emails it has never seen. Measure accuracy
Step 4 - Deploy: Integrate the trained model into Gmail so it filters real emails
Step 5 - Monitor: Track if the spam filter starts missing new types of spam and retrain if needed

Simple Code: The Concept

# This is a simplified illustration of the AI training concept
# We will build real models starting from Module 3

# Step 1: Data (inputs and correct labels)
training_data = [
    ("Win a free prize now!", "spam"),
    ("Meeting tomorrow at 3pm", "not_spam"),
    ("Urgent: click here to claim your reward", "spam"),
    ("Can you review this document?", "not_spam"),
    ("You have been selected for a lottery!", "spam"),
    ("Team lunch is on Friday", "not_spam"),
]

# Step 2: The AI "learns" by counting patterns
spam_words = {}
for email, label in training_data:
    if label == "spam":
        for word in email.lower().split():
            spam_words[word] = spam_words.get(word, 0) + 1

# Step 3: Use the learned patterns to predict
def predict_spam(email):
    score = sum(spam_words.get(word, 0) for word in email.lower().split())
    return "spam" if score > 2 else "not_spam"

# Step 4: Test on new email
new_email = "Congratulations! You won a free prize!"
print(predict_spam(new_email))  # Expected: spam

Try It Yourself

Try It YourselfHTML

HTML Editor

✓ ValidTab = 2 spaces

# Run the spam detector and try different emails
training_data = [
    ("Win a free prize now!", "spam"),
    ("Meeting tomorrow at 3pm", "not_spam"),
    ("Urgent click here to claim your reward", "spam"),
    ("Can you review this document?", "not_spam"),
    ("You have been selected for a lottery!", "spam"),
    ("Team lunch is on Friday", "not_spam"),
]

spam_words = {}
for email, label in training_data:
    if label == "spam":
        for word in email.lower().split():
            spam_words[word] = spam_words.get(word, 0) + 1

def predict_spam(email):
    score = sum(spam_words.get(word, 0) for word in email.lower().split())
    return "SPAM" if score > 2 else "NOT SPAM"

# Try these - and add your own!
test_emails = [
    "You won a free prize! Click now!",
    "Can we schedule a call tomorrow?",
    "Urgent: claim your lottery winnings",
    "Great work on the project report",
]

for email in test_emails:
    print(f"'{email}' -> {predict_spam(email)}")

Result

Click ▶ Run to see the result

Edit the code on the left, then click Run (Ctrl+Enter)

HTML|30 lines|975 chars|✓ Valid syntax

UTF-8

Key Takeaways

Every AI system follows a similar basic pattern: collect data, train a model, use the model to make predictions, and evaluate how well it works.
Step 1 - Collect Data: Gather 10,000 emails labeled as spam or not spam
Step 2 - Train: The AI reads all 10,000 emails and learns which words and patterns appear more in spam
Step 3 - Evaluate: Test the trained AI on 2,000 emails it has never seen. Measure accuracy

Topics in This Module

The AI System Lifecycle

Here is the basic flow of any AI system:

1Collect Data: Gather examples of inputs and the correct answers
2Train the Model: The AI studies the data and learns patterns
3Evaluate: Test the model on new data to see how accurate it is
4Deploy: Put the model into a real application where users can use it
5Monitor and Improve: Watch how the model performs and update it over time

Let us walk through this with a real example.

Diagram

Loading diagram…

Deep Learning ⊂ Machine Learning ⊂ Artificial Intelligence

Example: Building an Email Spam Detector

Step 1 - Collect Data: Gather 10,000 emails labeled as spam or not spam

Step 2 - Train: The AI reads all 10,000 emails and learns which words and patterns appear more in spam

Step 3 - Evaluate: Test the trained AI on 2,000 emails it has never seen. Measure accuracy

Step 4 - Deploy: Integrate the trained model into Gmail so it filters real emails

Step 5 - Monitor: Track if the spam filter starts missing new types of spam and retrain if needed

Simple Code: The Concept

# This is a simplified illustration of the AI training concept
# We will build real models starting from Module 3

# Step 1: Data (inputs and correct labels)
training_data = [
    ("Win a free prize now!", "spam"),
    ("Meeting tomorrow at 3pm", "not_spam"),
    ("Urgent: click here to claim your reward", "spam"),
    ("Can you review this document?", "not_spam"),
    ("You have been selected for a lottery!", "spam"),
    ("Team lunch is on Friday", "not_spam"),
]

# Step 2: The AI "learns" by counting patterns
spam_words = {}
for email, label in training_data:
    if label == "spam":
        for word in email.lower().split():
            spam_words[word] = spam_words.get(word, 0) + 1

# Step 3: Use the learned patterns to predict
def predict_spam(email):
    score = sum(spam_words.get(word, 0) for word in email.lower().split())
    return "spam" if score > 2 else "not_spam"

# Step 4: Test on new email
new_email = "Congratulations! You won a free prize!"
print(predict_spam(new_email))  # Expected: spam

Try It Yourself

Try It YourselfHTML

HTML Editor

✓ ValidTab = 2 spaces

# Run the spam detector and try different emails
training_data = [
    ("Win a free prize now!", "spam"),
    ("Meeting tomorrow at 3pm", "not_spam"),
    ("Urgent click here to claim your reward", "spam"),
    ("Can you review this document?", "not_spam"),
    ("You have been selected for a lottery!", "spam"),
    ("Team lunch is on Friday", "not_spam"),
]

spam_words = {}
for email, label in training_data:
    if label == "spam":
        for word in email.lower().split():
            spam_words[word] = spam_words.get(word, 0) + 1

def predict_spam(email):
    score = sum(spam_words.get(word, 0) for word in email.lower().split())
    return "SPAM" if score > 2 else "NOT SPAM"

# Try these - and add your own!
test_emails = [
    "You won a free prize! Click now!",
    "Can we schedule a call tomorrow?",
    "Urgent: claim your lottery winnings",
    "Great work on the project report",
]

for email in test_emails:
    print(f"'{email}' -> {predict_spam(email)}")

Result

Click ▶ Run to see the result

Edit the code on the left, then click Run (Ctrl+Enter)

HTML|30 lines|975 chars|✓ Valid syntax

UTF-8

Topics in This Module