Jargon Buster i. AI, Machine Learning, Deep Learning, Generative AI
- 18 hours ago
- 4 min read
AI, Machine Learning, Deep Learning, Generative AI, and LLMs are all terms commonly thrown around in tech writing, news and media, but they each refer to slightly different things and shouldn't be used interchangeably... even if they sometimes are.
This guide explains what each one actually means, how they relate to each other, and why it matters which is which.
What is AI, exactly?
Artificial intelligence is the broadest of the four terms. It describes any computer system capable of performing tasks that would typically require human intelligence: understanding language, recognising images, making decisions, or solving problems.
This category is wider than most people assume. A rule-based chatbot that responds "I didn't understand that" to anything outside of a predetermined script counts as AI. So does a simple chess program from the 1970s. Neither is particularly impressive, but both qualify as AI.
What most early AI systems had in common was that a human wrote all the rules. The program followed instructions. It didn't learn anything on its own.
What is machine learning, and how does it differ from AI?
Machine learning is a subset of AI. It changed the way these AI systems are built.
Instead of a programmer writing explicit rules, a machine learning model is trained on data. You show it many examples, and it works out the patterns itself. Feed a machine learning model enough labelled images, and it learns what distinguishes a cat from a dog, without being told which features to look for.
The model can generalise to new examples it hasn't seen before, and it adapts as it encounters more data. A traditional rule-based system can't do that.
The key distinction is not capability but method: were the rules written by a human, or learned from data?
What is deep learning?
Deep learning is a subset of machine learning. It uses a specific type of model called a neural network, loosely inspired by the structure of the human brain. They consist of a connected network of nodes where each node performs calculations before passing data off to the next one.
Nodes are grouped into layers. The "deep" in deep learning refers to the number of layers, which can run into the hundreds.
The significant advantage of deep learning is how well it handles unstructured data such as text, images, and audio. Earlier machine learning methods often required careful preparation of data before it was given to the model, but Deep learning can learn directly from the raw input data.
Most of the AI that gets people's attention today sits on deep learning foundations. Image recognition, speech-to-text, language translation, and large language models all use it.
Deep learning is what made modern AI practical at scale. Most of the 'impressive' systems we use today are built on a foundation of deep learning.
What is generative AI, and where does it fit in?
This is where the hierarchy gets slightly less tidy. Generative AI doesn't describe a new level in this stack. It describes a type of output.
Generative AI is AI that produces new content: text, images, audio, video, or code. ChatGPT generates text. Midjourney generates images. Both use deep learning under the hood, and both are AI in the broadest sense. The "generative" label tells you what they do, not how they're built.
This is what distinguishes them from a model that classifies whether an email is spam. That model also uses deep learning and machine learning. It makes a decision. It doesn't create anything.
Note: "generative AI" has also become a marketing term. When a product claims to be "powered by generative AI," it most likely means it uses a large language model or an image generation model.
What is a Large Language Model?
A Large Language Model (LLM) is a specific type of generative AI. It uses a deep learning model to generate text. Chat GPT, Google Gemini, and Claude are all examples of an LLM.
Why are these terms used interchangeably?
Largely because of marketing. "AI" became the label to reach for because it sounds impressive. Once that happened, it got applied to almost anything involving a model or an algorithm.
There's also a technical reason. The categories genuinely overlap. Generative AI uses deep learning. Deep learning uses machine learning. Machine learning is AI. Calling a washing machine "AI" might not be wrong, but it's definitely not using the same technology as ChatGPT.
The problem is that this overlap becomes misleading when precision matters.
Why does it matter which term you use?
Understanding the distinctions helps you ask better questions and calibrate your expectations.
If a company says their product "uses machine learning," that tells you something specific. There's a model. It was trained on data. If they say it "uses AI," that tells you almost nothing. It could mean a rule-based model written in 2009 or a state-of-the-art large language model. You can't tell from the label.
As IBM's explainer on AI and machine learning puts it: AI is the field, machine learning is the method, and deep learning is the technique. Getting these straight helps you engage with the technology on its own terms, rather than its marketing.
There's a simpler benefit, too. Hype becomes much easier to spot. A lot of the noise around AI comes from people treating one term as a catch-all for all four.
Final thoughts
AI is the umbrella. Machine learning is a method within it. Deep learning is a technique within that. Generative AI describes what a system does with its output.
Understanding the distinctions won't make you a developer, but it will make you a more informed user of the tools, and a harder target for marketing that relies on the confusion.
If you want to put this into practice, The Fundamentals of Prompting covers how to get more out of the AI tools you're already using day to day.
Sources:
Comments