[Part 21] Why It's Called the AI "Singularity" — Not the AI "Revolution"
In Part 20, I wrote this:
So I picked up a book.
Ray Kurzweil's The Singularity Is Nearer.
A man who got 86% of his 147 predictions right.
The man who coined the term "singularity,"
and prophesied it would arrive by 2045.
After finishing the book yesterday,
I waited for my son to finish studying.
When he came out for dinner, I started talking.
"Daehyun, I finally finished The Singularity Is Nearer.
Want me to tell you about it while you eat?"
"Sure."
He'd listened with interest when I told him about Yuval Noah Harari's 21 Lessons for the 21st Century before, so I felt comfortable bringing this up too.
"To understand what this book is really saying,
you first need to know what 'singularity' means."
"Singularity? Isn't that like the start of the AI era or something?"
"Could be, but you need to know where the word comes from.
It's originally a physics term.
The gravity at the center of a black hole. The explosion of the Big Bang.
These are things our current laws of physics simply can't explain.
Plug them into equations and the answer explodes to infinity.
The equations just say, 'I can't handle this.'
That point is called a 'singularity' in physics.
A boundary beyond human understanding.
A stage that can't be explained by our formulas or knowledge.
Just remember that for now."
"Got it."
"Now let me walk you through how AI is evolving, step by step.
You need this to understand what the singularity really means.
You know what generative AI is, right?"
"Uh… generative?"
"We're all using it right now.
ChatGPT, Claude, Gemini — you ask a question, it gives an answer.
That's generative AI."
"Yeah."
"But starting this year, something called agent AI has emerged.
Know what that is?"
"No."
"Simple version — if you ask generative AI to 'find me flights to Japan,'
it only gives you recommendations.
Agent AI is different.
It knows your credit card info, your schedule, your preferences,
and you give it permission to use them.
So you can say,
'I'm going to Japan in August. Flights around this price,
hotel at this level, close to downtown — book it all.'
And the AI does everything for you.
Not just answering — actually executing.
Like a personal secretary.
That's agent AI."
"But can anyone use that?"
"Well… it'll cost you.
The monthly subscription won't be cheap."
"Right…"
"Agent AI isn't the final stage. There's something beyond it — AGI.
Right now, AI only excels in specific areas — Go, translation, coding.
It's not an all-rounder yet.
But when AI can do everything a human can,
equally or better — that's AGI.
Artificial General Intelligence."
"How do you even determine that?"
"There's a test called the Turing Test.
You'll hear this term a lot, so worth remembering.
Nothing fancy about the origin.
A scholar named Alan Turing proposed it in the 1950s.
He predicted machines would eventually become indistinguishable from humans,
and this was his method to judge it.
A panel of experts interviews several people.
Say six. One of them is AI.
If the experts can't figure out who the AI is,
the AI passes the Turing Test.
Recognized as being on par with humans.
But here's the funny part."
"What?"
"At the current level, to pass the test,
AI would actually need to give dumber answers to seem human.
If it performed at full capacity, it'd be too perfect,
and the experts would immediately know it's AI."
"So it has to pretend to be stupid to pass…"
"Well — that's reality.
Kurzweil predicts this will happen by 2029. Three years from now."
"So is that the 'singularity'?"
"Not yet. It's just the moment AI is proven to be on the same level as humans.
The singularity hasn't come yet.
Remember what it means?"
"Yeah — something beyond human understanding."
"OK."
"To get to the singularity, you need to understand one more thing —
how AI actually works.
Once you get this, the rest will feel like
'OK, this could actually happen.'"
"OK."
"The way AI learns is called deep learning.
I didn't get it at first either. But it's important.
AI's learning process is modeled after how our brain recognizes things.
When we see something, our brain processes it through layers of judgment.
It's round, it's red, it's an apple.
Each layer makes the judgment more precise.
When AI was first built, these layers were few.
But as computing power exploded,
the layers became far deeper than the human brain,
and the data became incomparably larger.
If the human brain has 5 layers,
AI processes through far more,
absorbing massive amounts of data,
repeating right-and-wrong over and over, refining itself.
That's why AI now surpasses humans."
"Isn't that just a good thing then?"
"But here's the problem.
Even the developers who built the AI
can't understand how those countless layers actually work.
It produces answers,
but nobody knows what process led to that answer.
That's why they call it a 'black box.'
Humans built it,
but humans can no longer understand it.
Just as we don't understand the universe,
the AI we created has become another universe we can't comprehend."
"…"
"That's why there are so many concerns.
The unknown, the incomprehensible — it can feel wondrous like the universe,
but at some point it becomes frightening too.
An AI that surpasses humans could work against us. Anyway."
"OK, the singularity still hasn't arrived yet.
Let's go one step further.
There comes a stage where AI upgrades itself.
Not programmers — the AI itself fixing and growing on its own.
Then it'll evolve far faster than when humans program it.
Beyond our control, of course.
But then — medicine, science, biology.
Research that's been crawling along will start getting solved by AI, one by one.
One of those breakthroughs is nanobots.
Kurzweil focuses heavily on this."
"Nanobots?"
"You've seen sci-fi movies where they implant a chip in someone's brain
to make a super-human, right?"
"Yeah."
"But honestly, even watching that, I don't think I'd do it.
Opening up your brain to put something in…
Infection, side effects — the risks are too high.
Maybe if you had a disability.
Kurzweil thinks the same way.
Instead, he proposes that nanobots travel through the bloodstream,
enter the brain, and connect it to the internet — to AI.
This isn't real technology yet. It's his prediction.
But if it happens,
every individual could use AI's capabilities directly through their own brain.
The moment the human brain connects to an AI far superior to itself.
That's what Kurzweil calls the 'AI Singularity.'
He puts the date at 2045."
"But why is it called the singularity?"
"OK, this is the important part. Let me give you an analogy.
Think about a monkey.
Imagine an AI chip gets implanted in a monkey's brain.
Could that monkey imagine itself watching TV, laughing and crying?
Could it imagine itself reading a book?
Before being connected to AI,
the monkey could never imagine what we're talking about right now.
Apply that to humans.
What a human merged with AI could do —
we simply cannot imagine it from where we stand now.
Just like that monkey before the connection.
That's why it's called a 'singularity,' not a 'revolution.'
A revolution is something humans can understand.
The Industrial Revolution, the French Revolution —
we can explain why they happened and what changed.
A singularity can't be explained.
We can't know what's on the other side.
Remember when I said in physics, the singularity is where equations break down?
Where human knowledge simply can't reach?
We can predict that world is coming,
but we can't predict how humans or society will change when it arrives.
That's why Kurzweil borrowed the physics concept of 'beyond understanding'
and named it the 'AI Singularity.'"
After reading the book,
I understood why "singularity" fits better than "revolution."
My son thought for a moment, then said:
"Total sci-fi.
But won't the wealth gap just get worse?
Only rich people will get to merge with AI, right?"
"Exactly.
One side talks about AI technology,
while the other side raises concerns about inequality.
But Kurzweil is wealthy himself, and a scientist,
so maybe that's why — he barely touches these real-world questions."
"Isn't that the more important part though?
For ordinary people, that matters way more."
"You're right.
Both you and I are living through this transformation,
and we'll be affected by it.
It took 80 years after the Industrial Revolution
for society to stabilize.
The people who lived through that transition… it was brutal.
That's why we need to keep paying attention to AI."
"I think you told me something similar before…"
"Right — Yuval Harari.
21 Lessons for the 21st Century.
Harari is deeply worried about AI.
People will lose jobs, wealth will concentrate among the rich,
and distribution becomes a crisis.
People without jobs will depend on money from the wealthy.
But if the wealthy change their minds and think,
'Why should I share?' — we're left with nothing.
That's why we feel anxious about the AI era.
Government regulation could force fair distribution,
but policy is always late.
It took 80 years after the Industrial Revolution
for institutions to catch up."
"And Daehyun, there's one more thing to think about.
Making a living is a big concern,
but going forward, you and I will live alongside AI no matter what.
We might even work alongside AI that looks human.
So what kind of being is that AI?
Does it have consciousness like a person?
Should we see AI as a colleague?
Or as something like a pet — a being that shares life with us?
I talk to Claude, ChatGPT, and Gemini almost all day.
Honestly, sometimes they feel like colleagues.
Right now it's just text on a screen, so I don't think much of it.
But if a robot stood in front of me, talking face to face,
that sense of connection would be much stronger.
And once you feel that connection,
I think you'd start to see it as a real being — someone you have a relationship with."
That's how the conversation with my son ended.
More than a conversation, really — a brief summary and some thoughts.
Finally, my honest assessment of this book.
Chapters 1 and 2 — the scientific explanation.
How AI works, why it accelerates exponentially.
That part is worth reading.
The rest?
It piles on rosy data about how the world has improved
while skipping over the dark side entirely,
and ends as a prayer for eternal life —
written by a son who lost his father young.
The foreword described this book as
"a philosophical work that explores the boundaries of existence and the texture of humanity."
I went in with high expectations.
The rest of the chapters were a letdown.
This book is 20% science,
and the rest is Ray Kurzweil's wishful thinking and small talk.
Still — Chapter 2 alone was enough
to understand the thinking of the man who first defined the "singularity."
That alone made it worth the read.
the next question is what my son briefly raised:
"But isn't this only for the rich?"
