Five friends walk into a bar…
Series 1: AI Basics — Chapter 2: Differences between AI, ML, DL, GenAI & LLMs
The layers of technology within AI and what they mean.
This is the 2nd of seven articles that are dedicated to helping non-technical readers understand Artificial Intelligence (AI) and what it could mean for them.
As a non-technical person myself, I plan to share what I’m learning along the way and will do my best to simplify these seemingly complex topics.
With a lot of information out there, I’ve structured my writing into multiple Series (categories of topics) and Chapters (concepts within each topic). This is the first series on ‘AI Basics’ and contains seven chapters.
Now, let’s get into it.
Chapter 1: Fundamentally Understanding AI
Chapter 2: Differences between AI, ML, DL, GenAI & LLMs [this article]
Chapter 3: A Brief History of Artificial Intelligence
Today, we break down important concepts to understand the differences between AI, ML, DL, GenAI & LLMs. Comprehension of how each of these components within AI differs from each other is key to gaining a solid understanding of current and future developments in this field.
1. What is Artificial Intelligence (AI)?
As mentioned in the previous post, AI is fundamentally a high-level discipline advancing the simulation of human intelligence in machines through a set of approaches to perform tasks such as visual perception, speech recognition, decision-making, and language translation — skills we humans possess.
Simplified version: Imagine AI as a smart helper that can do tasks for you, like a robot friend. It’s like having a magical assistant who can understand what you need and help you with things like making decisions or giving advice.
Within its realm, AI comprises subfields of specialized technologies over the years — Machine Learning, Deep Learning, Generative AI, and Large Language Models (LLMs) — that function together or independently based on the task at hand.
2. Machine Learning (ML)
ML is a subset of AI that focuses on developing algorithms to learn from data, identify patterns, and make predictions or decisions based on its learnings. It uses different methods to learn such as supervised learning, unsupervised learning, and reinforcement learning (definitions below). For example, ML is used to predict patient readmission in healthcare. Other examples are recommended shows from Netflix based on what you’ve watched or are currently watching and recommended products on an e-commerce website.
What is supervised learning? Supervised learning, also known as supervised machine learning, is a subcategory of machine learning. It is defined by its use of labeled datasets to train algorithms to classify data or predict outcomes accurately. Supervised learning helps organizations solve a variety of real-world problems at scale, such as classifying spam in a separate folder from your inbox.
What is unsupervised learning? Unsupervised learning, also known as unsupervised machine learning, uses machine learning algorithms to analyze and cluster unlabeled datasets. These algorithms discover hidden patterns or data groupings without the need for human intervention. Its ability to discover similarities and differences in information make it the ideal solution for exploratory data analysis, cross-selling strategies, customer segmentation, and image recognition.
What is reinforcement learning (RL)? RL is a technique that trains software to make decisions to achieve the most optimal results. It mimics the trial-and-error learning process that humans use to achieve their goals. RL algorithms use a reward-and-punishment paradigm as they process data. They learn from the feedback of each action and self-discover the best processing paths to achieve final outcomes. RL is a powerful method to help AI systems achieve optimal outcomes in unseen environments.
Simplified version: ML is like teaching a computer to learn from examples, just like how you learn from practice. It’s similar to showing a robot many pictures of animals and teaching it to recognize which ones are cats and which ones are dogs.
3. Deep Learning (DL)
DL is a subfield of ML that uses something called neural networks with many layers to simulate the complex decision-making of the human brain. It excels at tasks involving unstructured data like text, images, audio, and video. DL applications include image recognition for things like medical diagnosis and autonomous driving.
What is unstructured data? In the modern world of big data, unstructured data is the most abundant. It’s so prolific because unstructured data could be anything: media, imaging, audio, sensor data, text data, and much more. Unstructured simply means that it is datasets (typical large collections of files) that aren’t stored in a structured format. It might be human-generated, or machine-generated in a textual or a non-textual format. Unstructured data just happens to be in greater abundance than structured data.
Simplified version: DL is like having a computer brain with many layers that help it understand things better. Each layer learns different aspects, like recognizing shapes or colors. It’s as if the computer can see and understand things in a more detailed way, like telling apart different types of fruits.
4. Generative AI (GI): generates content from the data it’s been trained on
GI is a specialized field within AI that focuses on creating systems capable of generating or creating new data such as text, images, or music in a human-like manner. GI is the field that caught the world’s attention on 30 November 2022 when OpenAI released ChatGPT to the world.
Simplified version: GI is like having a creative machine that can make or generate new things, such as stories, pictures, or even videos now. It’s as if you have a robot artist who can paint beautiful pictures, write interesting stories all on its own, shoot short films, and speak like you just by learning from examples.
5. What are Large Language Models (LLMs)
LLMs are advanced artificial intelligence algorithms specializing in understanding and generating human language text. These models analyze vast amounts of data to comprehend natural language patterns and generate new content based on what is asked of them. LLMs have evolved from traditional language models by dramatically expanding the data they use for training, resulting in a significant increase in their capabilities.
They typically consist of at least one billion parameters or more, which are variables that the model uses to generate text based on its training data. These models play a crucial role in various applications from sentiment analysis to conversational AI, making them essential tools for businesses seeking to harness the power of artificial intelligence in their operations.
Simplified version: Imagine LLMs as super smart storytellers who can write amazing stories and have conversations just like humans. They are like magical writers who learn from reading lots of books (millions!) and talking to people, allowing them to create new stories, and articles, or even have chats with you. These language models are like having a friend who knows everything about human history, can talk about anything under the Sun, and help you understand complex information in a simple way.
GenAI and LLMs are key pieces of technology being exponentially improved as consumers can directly access them through tools like ChatGPT, Claude, Copilot, and Gemini.
I will be writing dedicated chapters for these two technologies to provide a deeper understanding.
