Machine Learning

In order to explore the bigger questions emerging in the field of AI, it is important to first get a solid understanding of what AI actually is. That is exactly what this post is about.

A simple breakdown of the AI buzzwords we keep hearing

The general idea of “AI” keeps getting passed around everywhere, it seems to be ever present, but how many of us actually reflect on what we’re actually talking about when we think of AI. It is one thing to discuss AI applications in daily life, business and industry, but in order to truly gain an edge in AI, no matter from which field you come from, whether its business, computer science, arts or finance, it is crucial to make sense of what AI is comprised of, what makes it work, what this coded “intelligence” really entails. This understanding grants us a perspective and knowledge of how we can apply AI in what really inspires us, helping us bring our ideas to life, when we know what aspect of AI we can use to implement in a project, business solution, or startup idea. This is what makes the difference between someone who “watches AI” versus “does AI”. So let’s zoom out and answer a simple question: 

What actually is AI? And how do all the pieces like ML, DL, NLP, and CV fit together?

Let’s break it down.

AI: Artificial Intelligence – The Big Umbrella

In strict terms, AI is a broad field of computer science focused on building machines that can simulate human intelligence - think reasoning, learning, planning, and perception. Through this blog guide we will use analogies and examples to help break down the discussed concepts.

Analogy: If your brain is like a complex decision-making machine, AI is a digital attempt to recreate something similar.

Example: When Spotify recommends you a song, that’s AI. When your car parks itself, that’s AI too. At first glance, AI might seem like a far-fetched, overly complex field - something best left to computer scientists and tech experts. However, this thinking results in the loss of an incredible opportunity.  In reality, many of the most impactful contributions to AI today are coming from people who don’t come from traditional technical backgrounds. Some of the world’s most innovative AI startups have been launched by entrepreneurs with backgrounds in business, psychology, design, or the arts. They didn't build the AI models themselves - but they understood the technology and its value, identified real-world problems, and brought teams together to solve them. Understanding the core concepts of AI opens doors - not only to technical development, but to creative problem-solving, entrepreneurship, and leadership in one of the most transformative fields of our time.  And the good news is, these seemingly complex technical concepts aren’t as out of reach as they might appear - read on, and you’ll see just how accessible they can be.

AI: The Technical Bit - Algorithms, Data & Probability

At its core, AI is built on algorithms - step-by-step procedures that process information, learn from patterns, and make decisions. Most modern AI systems rely heavily on statistics and mathematical probability to make educated guesses or predictions.  AI systems are trained on large datasets, and they try to detect patterns in that data. Then, using those patterns, they make predictions about new, unseen situations.

Example:  

• A spam filter sees thousands of emails labeled as "spam" or "not spam" → It learns which features (like “free money!!!”) are common in spam → Then it flags future emails based on what it learned.

• Back to the Spotify example; a recommendation engine like Spotify learns which types of songs you listen to → It looks for patterns and suggests new music with similar features. The technical bits behind the scenes involve:

• Optimization algorithms that try to minimize error

• Probabilistic models that estimate likelihoods (the mathematical probability part)

• Feedback loops that let the system improve over time

Now, having explored the basic underlying concepts of what fuels AI, let's look at what is under the “AI Umbrella”..These are the subfields that give AI its intelligence - each with its own focus, methods, and real-world applications.

Let’s start with the most fundamental one, something you’ve probably heard of quite a lot: Machine Learning - the brains behind AI.

ML: Machine Learning – The Brains Behind AI

ML is a subset of AI that focuses on teaching machines how to learn patterns from data, without being explicitly programmed. The technical bit: At its core, machine learning relies on mathematical models and statistical algorithms that learn from data by identifying patterns and relationships. These models are trained on labeled or unlabeled datasets and optimized through processes such as gradient descent and loss minimization. The goal is to make accurate predictions or decisions when exposed to new, unseen data - without needing to hard-code specific rules. Machine learning encompasses various approaches, including supervised learning (learning from labeled data), unsupervised learning (finding hidden structure in unlabeled data), and reinforcement learning (learning through trial and error based on feedback).

Analogy: Imagine teaching a child how to recognize dogs by showing them hundreds of dog pictures. Eventually, they “get it” - that's machine learning.

Example: Spam filters in your email inbox learn what spam looks like and filter it accordingly.

DL: Deep Learning – The Neural Network Layer

Deep Learning is a subset of ML that uses algorithms called neural networks, inspired by how the human brain works. To expand on that with more technical insights: deep learning uses artificial neural networks with multiple layers (often called deep neural networks) to model complex patterns in data. These networks consist of interconnected nodes (neurons) that process input through layers, gradually extracting higher-level features. Deep learning excels in tasks like image recognition, language processing, and speech recognition, where traditional algorithms struggle with raw, unstructured data.

Analogy: If ML is like learning with flashcards, DL is like absorbing knowledge through a full sensory experience, but more complex, and more powerful.

Example: Facial recognition on your phone, or Netflix knowing exactly what type of thriller you like? That’s DL in action.

NLP: Natural Language Processing – Teaching Machines to Understand Us

NLP is all about making computers understand, interpret, and respond to human language. It combines linguistics, computer science, and machine learning to help machines process and generate human language. NLP models rely on techniques like tokenization, part-of-speech tagging, named entity recognition, and language modeling. 

To quickly cover the definitions of the underlying techniques of NLP:

• Tokenization: Splitting text into smaller units like words or phrases (called tokens) to make it easier for a computer to process.

• Part-of-Speech Tagging: Labeling each word in a sentence with its role, like noun, verb, or adjective.

• Named Entity Recognition (NER): Identifying proper names in text, like people, places, dates, or organizations.

• Language Modeling: Predicting the next word in a sentence based on the words that came before it. Recent advancements, such as transformer-based architectures (for example the beloved GPT used in ChatGPT), have enabled machines to understand context, nuance, and intent with remarkable accuracy - this powers chatbots, voice assistants, and translation apps.

Analogy: NLP is the translator between us and machines.

Example: ChatGPT, Google Translate, Siri, autocorrect - all use NLP.

CV: Computer Vision – Giving Machines “Eyes”

Computer Vision enables machines to see and understand images and video the way humans do. It uses image processing, pattern recognition, and deep learning to extract meaning from visual data. Techniques like object detection, image classification, and semantic segmentation allow machines to identify and interpret elements within images or video. 

Analogy: Think of it as giving eyes to a robot and then teaching it how to recognize what it sees.

Example: Self-driving cars identifying people walking on the streets, or your phone unlocking when it sees your face.