By the end of this Article, you'll be able to:
Define generative AI and how it differs from other AI types
Recognise the key characteristics and technological foundations of generative AI
Up to now, we’ve talked about posture; How you relate to AI.
How do you work with it instead of reacting to it?
What skills keep you dangerous while the landscape shifts?
Now it’s time to lift the hood.
Not to turn you into an engineer, but to remove the mystery.
Because a lot of fear comes from not knowing what’s actually happening.
Let’s change that.
What the Heck is Generative AI?!
You probably interact with generative AI every day:
Writing tools.
Image generators.
Chatbots.
Assistants.
But “generative” isn’t just a buzzword.
It marks a shift in what AI can do.
Traditional AI systems analyse and classify.
For example:
Is this email spam or not spam?
Does this image contain a cat?
Which option best matches this category?
Generative AI does something different.
It thinks, deduces and creates.
It can write a new email.
Draft a paragraph.
Generate an image, a plan, a response, a line of code.
That jump from classification to creation is the fundamental change.
OK..... And?
When AI only classified things, it stayed in the background.
Helpful. Invisible.... but limited.
Generative AI sits much closer to human work:
language
ideas
reasoning
creativity
Which is why it feels exciting... but entirely unsettling.
It’s not just processing information anymore.
It’s participating in work that used to feel human.
However!
That doesn’t mean it understands you.
But it does mean you need to understand it.
How this works (in plain English)
Most of the AI tools you’re using are powered by something called large language models (LLMs).
They’re called:
language models because they work with text
"large" because they contain billions of internal parameters.
values that shape how they respond
A useful mental model:
Imagine reading an enormous amount of human writing.
Websites, books, conversations, code.
not to memorise facts, but to learn patterns.
Patterns like:
Which words tend to follow others
How ideas are structured
How tone shifts with context
That’s what these models do.
They learn the shape of language and knowledge.
This didn't happen overnight
Today’s generative AI exists because three things arrived at the same time.
1. Better learning architectures
The big breakthrough came around 2017 with something called the transformer architecture.
You don’t need to remember the name.
But it unlocked something very important:
the ability to track relationships across long stretches of text
understanding context, not just keywords
This made coherent language generation possible.
2. Massive amounts of data
Modern AI models are trained on vast collections of human-created text:
websites
documentation
public writing
code
This diversity gives them broad coverage.
not wisdom, but range.
3. Huge increases in computing power
Specialised hardware and large computing clusters made it possible to train models at this scale.
Without this, none of the above would matter.
When these three came together, researchers noticed something unexpected.
Their Power level is over 9000!!!!!!!!
As models got larger and trained on more data, their performance improved.
But then something surprising happened.
New abilities emerged.
Not explicitly programmed.
Things like:
reasoning step by step
adapting to new tasks with minimal instruction
following complex directions
This is part of why AI sometimes feels smarter than you expect, and other times confidently wrong.
It’s powerful, but not grounded in understanding the way humans are.
What’s Actually Happening When You Use AI
When you type into a chatbot, the AI isn’t:
Searching a database of answers
Retrieving a stored response
It’s generating text in the moment, based on:
What you wrote
The recent conversation
Patterns it learned during training
There’s also a limit to how much it can “hold” at once (context window).
Think of this as working memory.
It includes:
Your prompt
The AI’s replies
Any information shared in the current conversation
This matters because:
AI doesn’t remember things outside that context unless given access
It doesn’t have unlimited awareness
It can miss things if they fall outside the window
Understanding this explains a lot of “why did it forget?” moments.
The Three Things That Make Generative AI Powerful
Pulling it all together, modern generative AI stands out because of three characteristics:
Scale: It has absorbed patterns from an enormous amount of human-created text.
In-context learning: It can adapt to tasks based on examples you provide.
Emergent capabilities: As models grow, new abilities appear that weren’t directly designed.
This combination makes AI feel flexible, creative, and sometimes unpredictable...
Which brings us back to fluency.
OK... So Now What !?
If you think AI is:
a database → you’ll trust it too much
a magic brain → you’ll defer judgment
a threat → you’ll avoid learning
But if you understand it as:
a powerful pattern-based system that generates plausible outputs within limits
Then your role becomes clear.
You are:
the source of intent
the judge of quality
the owner of outcomes
Which is why the skills we’ve been building matter.
Delegation. Description. Discernment. Diligence.
They’re how humans stay in control when the tools get strong.
Next time, we’ll get more practical.
We’ll look at what generative AI can and can’t do well, and where people most often misuse it... sometimes without realising.
Until then,
I love you.