字幕列表 影片播放
Thanks to Brilliant for supporting this episode of SciShow.
Go to Brilliant.org/SciShow to learn more.
[♪ INTRO]
In 2019, Google researchers announced that they had
achieved quantum supremacy.
That does not mean that they're ushering in a new sci-fi future.
As far as we know.
It sounds very grandiose.
Instead, it describes what might be the first
useful quantum computer, an entirely new way
of performing calculations that's better than anything
we've got right now.
But not everyone thinks that Google actually got there,
or that quantum supremacy is even a
thing worth worrying about.
To understand what this means, we've got to go
way back to the basics of how computers work.
If you've been told one thing about a computer,
it's that, deep down, everything is just ones and zeros.
And, amazingly, this very simple fact is more or less true.
The principle behind your everyday modern computer
dates back to a landmark paper published
by British computer scientist Alan Turing in 1936.
He described a theoretical device we now call
the Turing machine, capable of solving any problem
with just three simple actions.
The machine could read a zero or one from a bit of memory,
like a strip of paper.
It could also write a zero or one to that bit of memory,
or it could move to an adjacent bit.
Mathematicians have proven that by combining
those three actions with a set of rules about when to use each one,
the Turing machine is capable of solving any mathematical problem.
It was theoretical at the time, but now every modern computer
is basically just an extremely complicated, very small,
really wonderful Turing machine.
These devices now are starting to be called classical computers.
Quantum computers, on the other hand, wait for this,
have bits called qubits that can represent a zero, a one,
or any combination of the two.
To understand what that means, we gotta use quantum physics'
most famous -- or infamous -- problem.
Yes, the cat one.
This thought experiment, first proposed
by Austrian physicist Erwin Schrödinger,
imagined a cat locked in a box with a poison device.
At a random time, the device would activate and kill the cat.
Until you open the box, you can't know whether
the dirty deed has actually happened, so, in a sense,
the cat is both alive and dead while hidden from view.
We've simplified this.
Now, Schrödinger's whole point with this is that
the world we're familiar with -- the classical world --
doesn't work like this.
But the quantum world does.
The cat is both alive and dead,
but only with quantum particles.
Now let's get back to qubits.
When you read the value of a qubit, you only ever get a zero or a one,
just like the cat can only ever be alive or dead in the end.
But which you get is all up to chance.
Every qubit has a probability of being a zero
and a probability of being a one,
a combination called a superposition.
But what the chances of each are is based on
how the qubit is set up.
What's more, the values of different qubits
can be linked together in a process called quantum entanglement.
That means that if you measure the state of
one entangled qubit, you also get information
about its buddies.
This adds up to, well, math.
By entangling qubits in certain combinations,
engineers can solve some of the same kinds of problems
that they can with classical computers.
It's the combination of superposition and entanglement
that gives quantum computers their theoretical power.
Basically, they should be able to do the same things,
but way faster.
In principle, a quantum computer can perform
a calculation very quickly by representing all possible outcomes
at once and then finding the correct one.
Which brings us back to quantum supremacy.
It's the idea that this approach can solve some kinds
of problems that classical computers can't --
in a practical amount of time, that is.
But unlike the physics that actually makes quantum computers work,
the idea of supremacy, of being better than classical computers,
is pretty imprecise.
Like, what counts as an impractical amount of time?
Also, quantum computers might be better only for
certain specific kinds of problems -- so does that matter?
Google's announcement that they had achieved
quantum supremacy has put these questions
front and center.
They constructed a device consisting of 54 qubits,
made of tiny loops of superconducting wire
and capable of representing around ten quadrillion states.
With it, they created a quantum random number generator
and generated a million numbers in just 200 seconds.
And after running some tests on the world's
most powerful supercomputer, they concluded
that the machine would take about 10,000 years
to do the same thing.
But it didn't take long for a research group at IBM to respond,
claiming they could program the same supercomputer
to do the simulation in two and a half days,
while also providing extra accuracy.
This is why it would be nice to have
a more solid definition of quantum supremacy.
2.5 days is not 10,000 years, but it's still
about 1000x slower than 200 seconds.
But we also don't necessarily need
a quantum random number generator.
Classical ones work fine.
So even if this is quantum supremacy, does that matter?
We at SciShow are not qualified to say,
but what's clear is that quantum computers are getting better.
And that has profound implications
for the world we live in.
Take, for instance, cryptography.
Every time you log into your computer or check your email,
your data is protected by encryption.
That encryption only works because classical computers
can't efficiently solve certain kinds of math problems.
The code protecting your bank account, for instance,
isn't unbreakable -- it would just take so long
that the bad guys don't bother.
But what if something that today takes ten thousand years
suddenly takes two hundred seconds?
That's the kind of change quantum computing represents.
In a way, you can think of these machines
totally like classical computers in the 1950s.
They fill rooms, require tons of power,
and are only useful for certain kinds of problems.
But year by year, they're getting smaller and more powerful.
We can debate how much progress has been made,
but progress is being made.
If history is any indication,
we will get there sooner or later.
Whenever quantum computers do become a thing,
they're going to need quantum computer programmers.
And you can learn computer science for yourself
with the courses on Brilliant.org.
Like the in-depth course on data structures,
which is all about the fundamental ways computers store
and manipulate data.
By the end of the course, you'll know exactly how
computers store data easily and access it quickly.
Brilliant has dozens of courses like this one.
In addition to computer science,
they cover science, engineering, and math.
Each one is designed to be hands-on
and to keep you engaged the whole way through.
Courses are even available offline via their Android and iOS apps
so you can keep learning on the go.
The first 200 people to sign up at Brilliant.org/SciShow
will save 20% on an annual premium subscription.
So if you've been telling yourself you want to learn to code,
here's your chance!
And by checking them out, you're helping support SciShow,
so thanks!
[♪ OUTRO]