As we start to lean on AI frameworks in our products (such as qualitative analysis using Gen AI in our Community Engagement Platform) and services, the conversations around adoption are becoming more frequent. A common discussion point is understanding the distinction between Machine Learning and Generative AI.
You may have found yourself caught in the crossfire of these conversations where these terms (and others) get thrown around with little consideration of what they actually mean or the difference between them. This understanding is kind of important if you're looking to bring them into your organisation.
The purpose of this article is to provide a simple explanation of each technology, its meanings, how it differs and how they all relate. The underlying frameworks are, of course, extremely complex with many of their principles above my paygrade, so with that, I'll keep it simple!
Machine Learning
Machine Learning (ML) as a principle is pretty self-explanatory and certainly not a new concept. It's a program that ingests large amounts of data and makes informed predictions based on input. For example, every time you use Netflix or scroll through your social media feed, ML is working behind the scenes to record those actions and provide better predictions around the content you like. It learns from your behaviour, so each time you skip a show or like a post, it gathers a bit more data to improve your content personalisation.
This happens through either supervised learning (where we tell it what's what, like tagging) or unsupervised learning (where it figures things out on its own).
Artificial Intelligence
Artificial Intelligence (AI) is the broader picture. If ML is learning from experience, AI is taking that learning and applying it to solve problems. It's the concept of making machines capable of tasks that typically need some form of human intelligence. So for example, every time your phone recognises your face or you've purchased a car that parks itself, that's AI in action. It takes all the patterns that ML has learned and uses them to make decisions and perform tasks.
Generative AI
Then we have Generative AI (Gen AI), which has got everyone excited over the last 12 months, and made a lot of very smart NVIDIA nerds very wealthy. While traditional AI and ML are great at recognising patterns and making decisions, Gen AI can create entirely new content such as text, images, music and code.
It's the difference between a system that can spot a spelling mistake and one that can write an entire article. Tools like ChatGPT and Claude (which I've used to help draft this article) are prime examples as they don't just understand what you're asking for, they can create something new in response.
Large Language Models
Large Language Models (LLMs) are fascinating because they sit at the intersection of all these technologies, acting as the interpreter between humans (your chat inputs, for example) and machines. Imagine them as incredibly well read assistants who’ve somehow managed to read and remember most of the text on the internet (a bit like that mate we all have who remembers every useless fact *looks in the mirror* but actually useful).
These models have learned from massive amounts of data, including billions of books, articles, websites and documents. Through this learning, they’ve developed the ability to understand context and nuance in a way that previous AI just couldn’t manage. It’s the difference between a calculator that can solve a math problem and a tutor who can explain why the answer makes sense.
You interact with these models when using services like ChatGPT, and you may have noticed that you can choose between different versions of their LLMs. Unlike those painful customer service chatbots we’ve all dealt with, modern LLMs actually get what you’re asking. They’re not just matching keywords or following a script, they’re switched on enough to essentially read the room and adapt their style on the fly.
Take something complex like quantum computing: the same LLM can explain it as a boardroom insight, a technical deep-dive for developers, or a simple analogy, depending on how you ask:
- For a simple analogy: "Explain quantum computing like I’m 10 years old."
- For a business context: "How could quantum computing disrupt industries like finance or healthcare?"
- For a technical deep-dive: "What are the key differences between classical computing and quantum computing?"
- For a creative twist: "Write a short sci-fi story about a future shaped by quantum computing."
The Relationship Between These Technologies
The relationship between these technologies is like building blocks that stack together to create modern AI applications. AI is the broad concept of machines doing smart things, while LLMs are specifically designed for understanding and generating human language. They're the reason why modern AI can have a natural conversation with you instead of just responding to rigid commands like the old days.
Let's break it down:
- Machine Learning: provides the foundation for the ability to learn from data.
- AI: uses this learning capability to solve problems and make decisions.
- Generative AI: takes it further by creating new content.
- LLMs: connect these technologies by converting human language into actions, combining ML’s learning, AI’s decision-making, and Gen AI’s creative abilities.
With the consumer release of Gen AI platforms, you can see these technologies working together seamlessly in everyday applications. When you use ChatGPT, for instance, you’re experiencing the combined power of Machine Learning (pattern recognition), AI (decision making), and Generative AI (content creation), all made possible through a Large Language Model.
If you want to dive further into these concepts, just ask Claude or ChatGPT – logic suggests they probably know what they’re talking about 😉
Tools Used to Create This Article
- Claude: To co-write the article
- Claude: To design the technical diagram
- Claude: To write the header image prompt, based on the copy
- Midjourney: To generate the header image (chaos: 25, weirdness: 100)
- ChatGPT: To peer review, with help from Jeff Crume's excellent video on AI, ML, and Deep Learning (Transcript)
- ChatGPT: To create the HTML markup for the post (because I had it open)
