Python is the language of AI. If you've spent any time reading about machine learning, you've seen Python code everywhere — in tutorials, GitHub repos, research papers, and courses. Eventually, you'll want to read it, run it, and maybe modify it.
The good news: you don't need to become a Python expert. You need a focused subset of the language. This guide teaches you exactly that — the minimal Python required to understand and work with AI code — with examples drawn from real ML workflows.
Before anything else, get Python installed.
The recommended way: Install Miniconda, a lightweight Python distribution manager. It makes managing packages and environments straightforward.
Once installed, open a terminal and type:
python --version
If you see Python 3.x.x, you're ready. You can write and run Python in:
For AI learning, start with Google Colab. It's free, runs in your browser, and has GPU access.
In Python, you store values in variables. No type declarations needed.
name = "neural network" # string (text)
layers = 4 # integer (whole number)
accuracy = 0.94 # float (decimal)
is_trained = True # boolean (True/False)
Why this matters for AI: Model hyperparameters, accuracy scores, and configuration values are stored as variables. You'll see things like learning_rate = 0.001 or epochs = 50 constantly.
A list is an ordered collection of items. You can store multiple values in one variable.
scores = [0.82, 0.87, 0.91, 0.95]
labels = ["cat", "dog", "bird"]
# Access by index (starts at 0)
print(labels[0]) # "cat"
print(labels[-1]) # "bird" (last item)
# Add an item
labels.append("fish")
# Number of items
print(len(scores)) # 4
Why this matters for AI: Training histories, lists of labels, batches of data — all stored in lists (or NumPy arrays, which work similarly).
Dictionaries store key-value pairs. Think of them like labelled boxes.
model_config = {
"learning_rate": 0.001,
"batch_size": 32,
"epochs": 10,
"optimizer": "adam"
}
# Access a value
print(model_config["learning_rate"]) # 0.001
# Update a value
model_config["epochs"] = 20
Why this matters for AI: Model configuration, training arguments, and results are almost always represented as dictionaries. When you see config = {...} in AI code, this is what's happening.
Loops let you repeat an action for each item in a collection.
# for loop — iterate over a list
fruits = ["apple", "banana", "cherry"]
for fruit in fruits:
print(fruit)
# range — loop a specific number of times
for i in range(5):
print(i) # prints 0, 1, 2, 3, 4
Why this matters for AI: Training loops iterate over data for multiple epochs. Data preprocessing loops clean each row in a dataset. You'll see for epoch in range(epochs): in virtually every training script.
Functions are reusable blocks of code. You define them once and call them many times.
def calculate_accuracy(correct, total):
return correct / total
result = calculate_accuracy(85, 100)
print(result) # 0.85
Functions can have default values for parameters:
def train_model(data, epochs=10, learning_rate=0.001):
print(f"Training for {epochs} epochs at lr={learning_rate}")
train_model(data) # uses defaults
train_model(data, epochs=20) # override one
train_model(data, epochs=20, learning_rate=0.01) # override both
Why this matters for AI: Every ML library is made of functions. model.fit(), model.predict(), train_test_split() — these are all function calls. Understanding how functions work (including default arguments) unlocks your ability to read library documentation.
accuracy = 0.92
if accuracy > 0.90:
print("Model is performing well")
elif accuracy > 0.75:
print("Model is acceptable")
else:
print("Model needs improvement")
Why this matters for AI: Conditional logic appears in data validation, early stopping criteria, and decision-making during inference.
Python's real power comes from its libraries. You import them at the top of your file.
import numpy as np # numerical computing
import pandas as pd # data manipulation
import matplotlib.pyplot as plt # plotting
The as np part creates an alias — you type np.array() instead of numpy.array(). This is standard convention.
Why this matters for AI: Every AI workflow starts with imports. The three above — NumPy, Pandas, and Matplotlib — appear in virtually every data science notebook.
NumPy is the foundation of numerical computing in Python. Its core object is the array.
import numpy as np
# Create arrays
a = np.array([1, 2, 3, 4, 5])
b = np.array([10, 20, 30, 40, 50])
# Operations apply element-wise
print(a + b) # [11, 22, 33, 44, 55]
print(a * 2) # [2, 4, 6, 8, 10]
print(np.mean(a)) # 3.0
# 2D arrays (matrices)
matrix = np.array([[1, 2], [3, 4], [5, 6]])
print(matrix.shape) # (3, 2) — 3 rows, 2 columns
Why this matters for AI: Neural network weights, image pixel data, word embeddings — all stored as NumPy arrays. When you see X_train.shape == (1000, 28, 28), that's a NumPy array of 1,000 images, each 28×28 pixels.
Pandas is for working with tabular data — spreadsheets in Python.
import pandas as pd
# Load a CSV file
df = pd.read_csv("data.csv")
# Explore the data
print(df.head()) # first 5 rows
print(df.shape) # (rows, columns)
print(df.columns) # column names
print(df.describe()) # statistics
# Select a column
ages = df["age"]
# Filter rows
adults = df[df["age"] >= 18]
# Check for missing values
print(df.isnull().sum())
Why this matters for AI: Real-world datasets come as CSV files. Pandas is how you load, explore, clean, and prepare data before training a model.
Here's a complete Python script that loads data, splits it, trains a simple model, and evaluates it:
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score
# 1. Load data
df = pd.read_csv("customers.csv")
# 2. Separate features and target
X = df[["age", "income", "years_customer"]]
y = df["will_churn"]
# 3. Split into training and test sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
# 4. Train a model
model = LogisticRegression()
model.fit(X_train, y_train)
# 5. Evaluate
predictions = model.predict(X_test)
accuracy = accuracy_score(y_test, predictions)
print(f"Accuracy: {accuracy:.2%}")
Read through this line by line. With the concepts above, you can understand every single line of it. That's the goal — not to write ML code from scratch, but to be able to read, understand, and modify it.
To be practical with AI code, you don't need to immediately learn:
These become valuable as you go deeper, but they're not prerequisites for working with AI tools and tutorials.
The best way to learn is to run code, break it, fix it, and experiment. Pick a Kaggle dataset on a topic you find interesting and start exploring it with Pandas.
Python fluency for AI is not about memorising syntax. It's about recognising patterns — knowing what a loop looks like, what a function call does, what an import brings in. With the concepts in this guide, you have the foundation to read AI tutorials, follow along with code examples, and begin experimenting yourself.
Want a structured path through AI fundamentals? AI Educademy offers free courses that take you from the basics of what AI is through to building and training your own models — no prior experience needed.
Start with AI Seeds — a structured, beginner-friendly program. Free, in your language, no account required.
Machine Learning voor beginners: alles wat je moet weten (gids 2026)
Machine learning voor beginners eenvoudig uitgelegd — leer wat ML is, hoe het werkt, belangrijke algoritmen en hoe je gratis kunt beginnen met praktische voorbeelden.
The Pace of AI Isn't the Problem - Our Readiness Is
AI is advancing faster than ever, but the real gap isn't in the technology - it's in how prepared we are to understand and use it. Originally published on LinkedIn.