Seeing Humanity through Dystopian AI

Have you noticed—especially this year, since ChatGPT burst onto the scene—that everyone seems fascinated by AI? And yet, in conversations with my family and friends, I keep picking up on this quiet sense of unease. The discussions always seem to circle back to the same question: Will machines take our jobs?

I think this anxiety runs deeper than AI itself. It's rooted in our modern obsession with efficiency—an obsession we can trace all the way back to Taylorism in the late 1800s and early 1900s.

If you don't remember, Taylorism was all about using scientific methods to break work into smaller, simpler tasks. Have each person do one tiny, repetitive thing, and boom—productivity goes up.

In many ways, Taylorism built the world we live in. It fueled the Industrial Revolution. It made us incredibly productive. But here's the thing—in chasing efficiency, haven't we spent the last century slowly turning ourselves into machines?

We didn't need to wait for AI to replace us. We've been doing it to ourselves. In the name of "efficiency," we've willingly become cogs in this giant social machine, doing the same tasks over and over, day after day. We became the assembly line.

So when AI comes along, it forces us to look in the mirror. It makes us confront a question we've been avoiding:

Why do we work in the first place?

I want to offer a different perspective on this.

Hi everyone. I'm Yuan-Sen Ting—an astrophysicist who uses AI to decode the universe.

My team and I use cutting-edge AI to analyze the cosmos. We study billions of galaxies to figure out how much dark matter and dark energy is out there. We probe the inner workings of stars. Recently, we've even been hunting for undiscovered asteroids that could threaten Earth.

And it's not just astronomy. AI has massive potential across all of science. It might be our best shot at solving some of humanity's biggest problems.

The Real Power of AI

Take the pandemic—I know many of us still shudder thinking about it. But here's something remarkable: we can now use AI to predict protein structures from amino acid sequences with stunning accuracy. That's going to dramatically speed up how we develop vaccines.

Or consider climate change. In just the past couple of years, physicists used AI to achieve breakthroughs in nuclear fusion. That's potentially a path to truly sustainable energy.

Yet despite all this potential, most people still see AI with fear and pessimism. And I think that pessimism comes from unfamiliarity. AI feels like this massive, complex, scary thing.

But here's the truth: AI isn't as mysterious as you think. Look at the screen—this is the actual code for GPT-2.

Surprised? This code for the chatbot that's supposed to represent "the future"—I haven't even finished this sentence, and you've already seen all of it.

That's why I've always felt "artificial intelligence" is a misleading name. It sounds like we're trying to create conscious beings. We're not. In academia, we prefer "machine learning," because that's what it actually is.

All we're really doing is teaching computers to find patterns in massive amounts of human data.

The Three Pillars of Machine Learning

There are three main areas. First, natural language processing—teaching machines to talk to us, like ChatGPT does. Second, computer vision—giving machines the ability to see, which you know from self-driving cars. And third, reinforcement learning—that's how we build robot vacuums and game-playing AIs.

So how do machines actually learn? It's surprisingly simple.

For language, we grab tons of text from the internet and randomly hide some words. The computer's job? Guess the missing words. It's like those fill-in-the-blank games we played as kids. Through billions of guesses, the machine learns how words connect and what they mean.

So even ChatGPT, as impressive as it sounds, is basically a very sophisticated parrot.

Once you understand the trick, it's like figuring out how a magic trick works. Magic is still cool—but it's not a miracle. Same with AI.

A Bias About Ourselves

Now, I'll admit—when ChatGPT entered our daily lives, lots of people wondered: Is this the singularity? Is this the moment machines become truly intelligent?

Here's the thing. Imagine a parrot starts mimicking your speech. And it gets better and better at it. That would freak people out, right?

But that shock—at least part of it—comes from a bias we have about ourselves.

Let me ask you: What do you think is humanity's most precious, irreplaceable ability?

Since you took the time to come to a TED talk, I'm guessing you'd say language. The power of words.

Fair enough. If we had to pick one thing that makes humans unique, the prefrontal cortex—which handles language, logic, and decision-making—would be a good candidate.

But here's what most people don't realize: from a biological perspective, this isn't our fundamental trait. The prefrontal cortex is actually a pretty recent upgrade—maybe a million years old, going back to our ape ancestors.

A million years sounds ancient, but compared to the rest of our evolutionary history? It's nothing. Our ability to walk and grasp things? That goes back hundreds of millions of years, to the earliest mammals. That's almost a hundred times longer.

If mammalian evolution were a 24-hour day, we spent almost the entire day optimizing our bodies. Language and logic? Those showed up in the last fifteen minutes.

Language Is Actually Pretty Basic

Why does this matter?

When we see breakthroughs in natural language processing, it's not because we're approaching some sci-fi singularity. It's because language is evolutionarily new—which means it's actually easier for computers to copy.

In other areas—computer vision, reinforcement learning—we're still hitting walls. That's why ChatGPT can write your emails, but we still can't build a robot vacuum that my mom is happy with.

So the real lesson from ChatGPT isn't "wow, machines can do our amazing human thing." It's "huh, maybe that thing we thought was so special... isn't that special after all."

In this talk, the point isn't whether I deliver a great speech. What's actually profound is you—every subtle expression on your face, every time you clap, every glance you exchange. That's infinitely harder to replicate than anything I'm saying up here.

Why Study AI At All?

People often assume that AI researchers like me are part of some "Team Machine"—that we dream of machines replacing humans.

That's a beautiful misunderstanding.

Go back to the very beginning of computer science, to Alan Turing. Before computers even existed, he proposed the Turing Test. Most people think it was about measuring how smart machines can get. But that's not what Turing was after. He was asking: What is humanity? What is intelligence?

I've always seen AI as a fundamentally human-centered science. Yes, we're making machines smarter. But more importantly, we're learning about ourselves by trying to imitate ourselves. AI isn't just a tool for efficiency—it's a mirror that helps us see who we really are.

AI as the Great Equalizer

AI's impact goes beyond science. It can be a force for fairness.

When I was in elementary school, IBM's Deep Blue beat Kasparov, the world chess champion. Everyone was devastated. "Chess is dead," they said.

But look at chess today—it's thriving like never before. Why? Because now, no matter who you are or where you're from, you can play against a machine and get better. AI became the great equalizer.

My team is working on something similar—a "machine astronomer." We're training a language model on astronomy. The goal isn't to replace astronomers. It's to give every curious kid, anywhere in the world, a "professor friend" to explore the universe with.

Walking the Tightrope

Think about the pandemic. That crisis showed us just how fragile human progress can be. There's an old saying from Laozi: "Fortune and misfortune lean on each other." What seems like disaster can lead to opportunity. What seems like blessing can hide danger.

The same applies to AI. Yes, it has two sides. I'm not here to sell you blind optimism. But I am here to say: finding balance takes courage.

Being human is like walking a tightrope. Every step requires care. But standing still won't keep you balanced. You have to keep moving forward.

That's how we'll find the balance we're looking for.

Thank you.