Original subtitles

Human life expectancy

has been growing steadily for over 100 years.

At the beginning of the 20th century,

on average, people lived for 47 years,

today, it's 79.

And it's believed that the first person

to live to 150 has already been born.

So, we sent Isobel Yeung to learn about advances in science and medicine

that are poised to radically expand our life expectancy even further.

In southern Japan, living a long life

is one of the greatest human achievements.

We're just driving through the streets of Okinawa,

joining a Kajimaya celebration, which is this lady-- Hi!

This lady turning 97.

Okinawa actually has one of the oldest populations in the world.

When someone turns 97, it's a sort of a turning point,

when you return to your innocence,

and everyone seems to be really happy.

So many Okinawans live past 100,

that the entire island chain is known as a "blue zone."

Meaning, it's a longevity hot spot,

where the rate of centenarians per capita

is among the highest on Earth.

Aw, that's so cool!

God, there's so many old people in this town.

Everyone's just trying to get a glimpse of her,

congratulate her, get to shake hands with her.

She's like a little local celeb.

What's your secret?

Okinawans have far lower rates

of cancer, heart disease and dementia than Americans,

and the reasons why are no longer a mystery.

For over 40 years, Dr. Makoto Suzuki

has been chronicling the details

of the oldest lives in Okinawa.

These are all your studies? Yeah.

Wow.

This guy looks pretty happy and healthy.

Just 100.

How old is this person?

Gosh, she looks amazing. Yeah.

Who is the oldest person you've studied? Uh...

111? Mmm.

Why is Okinawa unique in terms of its longevity?

This Okinawan study is part of an emerging field

of longevity research taking place worldwide

in the hopes of unlocking the biological mysteries of the longest lived.

One of the most renowned is the Institute for Aging Research

at the Albert Einstein College of Medicine,

directed by Dr. Nir Barzilai.

So, how do you plan to allow people to live a long and healthy life?

We have over 50 labs

that are looking at the biology of aging

from several perspectives,

and we're trying to coordinate the efforts,

that can allow us to target the aging process

and, kind of, slow it as much as we can.

In this lab, they're focusing on

how can we turn old cells

into younger, functional cells.

Besides cells, Barzilai and his team

are looking at ways to hack our metabolisms

and even DNA,

in order to fend off age-related illnesses.

They've even begun preparing for the first-ever human clinical trial

of a drug that they believe will slow the aging process.

Metformin is approved by the FDA for treatment of diabetes.

Why Metformin?

Because people with diabetes who take Metformin

have much less cardiovascular disease,

much less cancers,

much less cognitive decline,

and significantly less mortality than people without diabetes.

Metformin has effects on processes like

inflammation,

cellular survival,

stress defense.

What we're predicting, we'll delay variety of diseases

including mortality, cardiovascular disease,

Alzheimer's, cancer by about 30% in this study,

the way we planned it.

Wow.

That's huge.

If Dr. Barzilai's estimates are correct,

average life spans could jump by up to 25 years,

but while Metformin goes through the expensive FDA approval process,

dozens of age-defying products are already being sold.

None of these are defined as medicine,

which enables them to skirt the FDA's requirements.

In that gray area, between being medicinal or not,

are stem cell clinics.

Hi Isobel!

Pleased to meet you. I'm Dr. Halland.

Welcome to the clinic.

I'll be with you in just a moment.

Thanks!

Stem cell treatments employ a controversial form

of body hacking that many think can help you stay forever young.

Small little pinch, okay?

We do regenerative medicine here.

We help people feel better by using the body's own healing abilities.

Dr. Chen is harvesting this woman's fat

so he can collect the mysenchymal stem cells inside it.

The basic building block of our body are stem cells.

As we get older, the amount of viable and healthy stem cells is less.

We can now cell-shock cells,

and so these stem cells actually wake up to repair the body.

The fat solution is then shaken up like a martini

to prepare the stem cells,

or as Dr. Chen calls them,

medicinal signaling cells for activation.

I think this is phenomenal.

I'm very optimistic.

This woman is receiving stem cell treatment

in her knees, which are arthritic.

What stem cells do is they basically recruit other cells into the area.

It's almost like leaving a honing signal.

So, now the body says,

"Hey! Oh, I'm injured. I actually should start healing."

Any better?

Look at you go!

The FDA has only approved a few stem cell treatments

because of the limited clinical trial data.

But Dr. Chen believes that stem cells not only have orthopedic benefits,

they also may extend our lives,

and he's not just selling the idea, he's living it.

So, right now, I'm having my blood taken.

That will then get processed in a centrifuge,

and then we're gonna connect it into the IV.

For me, it's for systemic health.

I feel like if you're constantly in a state of healing,

the likelihood of getting sick is extremely low.

I mean, I don't really get sick at all.

- What we're producing is platelet-rich plasma... - Mm-hmm.

...and that's pretty commonly known as PRP.

All the regenerative elements are in there.

Okay, so what's gonna happen here?

He's going to just connect me to the NAD solution,

replenishing the mitochondria function.

And you do that simultaneously

whilst you're putting the plasma back into you?

- Correct, plus the laser. - Okay.

I'm putting a fiber-optic cable in my vein.

So, as my heart pumps the cells into the blood, it passes through the laser,

picking up energy, its photo-chemistry.

How long do you think you're gonna live for?

I mean, I'm happy with 100.

More than 500 of these boutique-size clinics

are currently operating in the US, without the need for government approval.

But dwarfing these businesses are the longevity ventures

being formed by today's wealthiest tech entrepreneurs,

who are pouring hundreds of millions of dollars

into disrupting human life itself.

We don't have to die a physical death.

Our consciousness can continue for a longer period of time.

Perpetual life, that's what we believe in.

One of the largest is run by Craig Venter,

who lists mapping the first human genome,

and creating the first forms of synthetic life

as just some of his achievements.

He's now cofounded the company Human Longevity.

I hear that you're known as the Steve Jobs of the biotech world.

That's a good positive one, so I like that one.

I describe it, I'm sort of the orchestra conductor, you know.

So, I have close to 600 scientists here in the three different organizations.

And in the biotech world, there's a lot of researchers and scientists

but you're an entrepreneur, essentially.

I think to be a good scientist today, you have to be entrepreneurial.

Especially in this field...

where it costs so much.

Venter has led multiple biotech firms and foundations

dedicated to the human genome,

including Synthetic Genomics,

and the J. Craig Venter Institute.

But at Human Longevity, you can use all that research

to map your own genome...

Good morning! Good morning.

...for the bargain price of $25,000.

Access granted.

Instead of simply getting your blood, and your temperature taken,

HLI's testing takes all day.

This is an electronic floor.

It has thousands of censors,

and this is gonna analyze your movement, and your gait.

So, I want you to count out loud, backwards,

by three's, and I want you to start at 200.

200, 197...

194, 185...

182.

Finger tapping test.

The objective is to tap a key as quickly as possible.

We actually need to collect a stool sample.

And you'll just slide the sample across,

rather than pushing it through... Okay.

...just slide it across.

We're used to thinking of ourselves as intact.

- So, it's a very different view of ourselves. - Mm-hmm.

Three, two, and one, holding.

When we sequence your genome,

we'll find around 8,000 extremely rare variants,

and those are the things more likely to be linked

to unique traits or even diseases.

And what proportion of people

who have been through the program have something wrong with them?

We find in about 40 percent of people

serious, potentially life-threatening medical issues.

If you saw bright spots pop up in these darker regions,

then we would know that there's a problem there.

Okay, but these are all normal-looking.

No? Don't scare me! Don't do that! I don't mean to scare you. I just-- you know, I'm not--

I'm not to interpret these, that's for the radiologist.

A few weeks later, you're given a prognosis

for not just your present state of health,

but your future as well.

This is the big day when you get all your results back.

I know, I'm a little bit nervous to be honest.

We made an avatar of you

with our multiple camera system. Awesome.

So, overall it's good news.

We didn't find anything truly wrong.

Now, we only found hints of early disease.

Slightly higher than normal liver fat,

insulin sensitivity,

and a slight tendency for metabolic syndrome and diabetes.

Does that mean this should act as a warning

in terms of my lifestyle? Absolutely.

You have control, and that's what most people

don't realize this data gives you.

While the data slapped me with some long-term issues to mull over,

the real breakthrough here, is that it can be used

to specifically isolate and discover deadly diseases,

which drastically lower humanity's average life expectancy.

With cancer, we've been finding that at stage zero, stage one,

and early stage two, where it's completely treatable.

We've found several people that have episodic atrial fibrillation,

that is one of the major causes of stroke.

Just putting them on anticoagulants

can be lifesaving in that case.

We need to have interventions to overcome defects

in our genetic code that could lead to early death.

But beyond early detection is a bigger goal.

Venter is coupling the data collected from Health Nucleus participants

with tens of thousands of sequenced genomes

to create the first predictive model of health analysis.

Why is data the answer here?

The complexity of the genome,

6.4 billion letters of genetic code,

is enough on its own that you need a lot of data to analyze it.

So, we use massive computing,

but we also use machine learning,

where tools in the computer

can look at data and learn faster than us mere mortals can.

We're starting to understand the patterns in the code.

Everyday, our system gets more and more and more powerful.

I think we're at the tipping point.

Aging can be manipulated.

It's science now, not science fiction.

Does the answer to longer life live within us?

I think so. I think there's

a lot of things that we haven't unlocked in human physiology.

The next thing is how do you activate it.

We're learning things faster

than we've ever learned in the history of science now.

Anybody that's born in this next decade,

has a greatly increased chance of living into triple digits.

Where do you think this is all going?

I think in two to 300 hundred years,

there'll be extensive editing of every human genome,

everything will be based on genetic predictions,

making people stronger, smarter,

and healthier and living longer.

The world spent more than $71 billion on robotics in 2015,

and the market for intelligent machines

is expected to more than triple by 2020.

Now, robots can already perform complex human tasks,

but as Hamilton Morris discovered,

their real value may lie in whether they can become smart enough

to actually think on their own.

I'm at IREX,

the world's largest robotics convention in Tokyo, Japan,

riding on this robotic unicycle through the exhibitions,

which showcase every imaginable variety of robot.

There are robotic snakes for looking through drains.

There's rescue robots,

climbing robots,

breast examination robots.

This body armor connects me to the robot

which mirrors all the movements.

I put my hand in the robot's hand,

and then held my own hand,

using my hand to control the robot.

Definitely has an uncanny valley effect where it's

just human-like enough

and almost repulsive in a way.

Some of the most popular robots

were human-like machines built for dangerous tasks in disaster recovery.

Though impressive, it was surprising how slow

and limited these robots were at moving like humans.

I visited a robotics lab in Tokyo

to talk to Shigeo Hirose,

a pioneer in robotics technology, about why that's the case.

Why is that this mechanical innovation moves slowly

relative to the innovation in software and electronics?

Essentially, for robots to become more functional,

they need bigger brains, not better mechanics,

and the key to unlocking that brain power

is developing artificial intelligence, or AI.

Even the tech giant Google is said to be moving away from the physical

and doubling down on AI.

AI is the science of programming computers to perceive their environment

and make rational, cognitive decisions in order to achieve a goal,

and it's one of the most rapidly progressing

and sought after technologies in the world.

At Harvard's Center for Brain Science,

Dr. David Cox is hoping to model AI

off real biological functions

by mapping the brains of animals.

A major drive of the experiments we're doing

is to understand how brains learn,

and if we understand how brains learn,

we need to have the brains be learning.

And we have rooms full of these kinds of racks,

and basically what this is, it's like a video arcade for rats,

and if you look inside,

you can see there are different little ports that they can lick.

Basically, those are the buttons on the game controller for the rats

and there's a monitor in the back of the rig,

and we can train the rats to distinguish between different things.

Once the rats are trained,

Dr. Cox observes the neuronal activity

inside the rats' brains while they're thinking.

We've injected a virus into their brain

that carries a genetically engineered protein,

which lights up a little bit brighter

when we shine a laser on it when the cell is active.

So, this is a way that we can actually watch the animal think.

The most intelligent rats' brains

are divided into slices thousands of times thinner than a human hair,

and then imaged through a powerful electron microscope,

and then modeled in 3-D space.

So, what we're doing is we're marrying behavior,

what the animal can do, its actual...

its learning, and how its improving

with the activity of those neurons,

and then marrying that with their neuroanatomy.

And from all of this massive amount of data,

we're gonna try and really understand

what kinds of algorithms might these neuronal systems be computing.

What allows biological systems to do all of these things

that we haven't yet figured out how to make artificial systems do.

That's gonna unlock a huge renaissance

in practical robotics, practical machine learning,

computer vision, things like that.

One of the major milestones in creating human-level intelligence,

is for machines to attain self-awareness.

Columbia University's Creative Machines Lab may have already done it.

When you look at this robot, you can see it has four legs,

but the robot itself does not know that it has four legs.

All it knows is that it has a couple of motors, a couple of sensors.

So, if self-awareness works,

this machine should gradually learn what it is.

As the robot comes to life, and takes its first steps,

it tries to understand what it is.

We turn it on, and we can watch what it thinks it is.

We can actually look at its memory,

and understand how it's modeling itself.

Does it recognize that it has four tentacles or eight joints?

In the beginning, it's completely wrong,

but the robot very quickly knows that they're wrong,

and it basically hones its self-image

to a point where it pretty much matches

its experiences in reality.

Using that self-image, it can learn how to walk

by playing out lots of forms of locomotion in its mind.

These robots learn over time

to simulate themselves

in a future situation they haven't actually experienced.

In other words, they don't have to learn by doing,

they can learn by thinking.

If these robots are in the infancy of AI,

their counterparts at UC Berkeley

are going to kindergarten.

I'm in the arms of BRETT,

the Berkeley Robot for the Elimination of Tedious Tasks.

Using trial and error,

BRETT learns how to fold laundry,

assemble LEGO blocks

and fit pegs into holes.

It goes from scratch.

It has no notion of how its arms move,

no notion of anything except for, like, a goal,

and it does some random set of actions,

and depending on whether its doing good or bad,

it evaluates that and keeps updating its policy

to do better and better and better.

And like a child, BRETT can watch someone perform a task,

like tying a knot, and actually learn it.

Now it has to use both its hands.

There we go.

So, that's how it's supposed to work, yeah.

To make robots like BRETT even smarter,

some AI developers are using gaming principles

to teach their AIs to strategize.

In 2014, Google reportedly spent more than $500 million

to acquire DeepMind,

one of the world's leading AI developers,

led by gaming prodigy

Demis Hassabis.

What is the significance of games in this sort of research?

Games were invented to be a kind of microcosm of aspects of life.

They're, um, a very efficient platform

to sort of develop AI algorithms and test them.

They can have some of the richness of the real world,

but they are in more constrained ways.

Advancements in computer science have been tested with games before,

like super-computer Deep Blue's stunning defeat

of world chess champion, Garry Kasparov in 1997.

Whoa! Kasparov has resigned.

That was something well beyond my understanding.

Now DeepMind has set out to beat

the most complex board game in the world.

The ancient Chinese game of GO.

Can you talk a little bit about what makes this a big deal, why it's important.

There's more possible board positions in GO than there are atoms in the universe.

If you ask a great GO player,

why they made a particular move,

they'll often answer that it just felt right.

That's quite different to when you ask a great chess player,

they'll usually be able to tell you precisely the calculations they made.

GO's a lot more about feel.

These great GO players use all of their years of experience

to just sort of come up with the right pattern-matching move

that fits for that position.

In other words, it was believed you need human intuition,

or else you can't win.

No AI program has ever defeated a GO world champion before.

To see if DeepMind successfully built artificial human intuition,

we traveled to South Korea, where Google organized a GO tournament

with a one-million dollar prize,

pitting one of the world's top human players

against AlphaGo.

That was Lee Sedol, who is the current GO champion walking by.

This is the final game of the GO championship.

This may be the biggest board game competition in history,

because this is a human champion

versus the most sophisticated computer-based

GO playing system that has ever been created.

Lee Sedol has about ten minutes left on the clock.

The game is slowly nearing its end.

I wonder if we have a resignation here?

It could be that Lee Sedol's resigned.

Yeah. I'm getting word Lee has resigned.

Wow.

It's over. Lee Sedol has lost.

AlphaGo won four out of five games.

It is the champion GO player.

AlphaGo's big win reinforced the idea

that we may be on the brink of transformative artificial intelligence,

but some of the world's most renowned thinkers

are warning that there's a high risk of unintended consequences.

The development of full artificial intelligence

could spell the end of the human race.

With artificial intelligence, we are summoning the demon.

He's the guy with the pentagram and the holy water,

and he's like, "Yeah, you sure you can control the demon?"

I don't think it's inherent

that as we create super intelligence,

that it will, necessarily, always have

the same goals in mind that we do.

Oxford philosopher Nick Bostrom,

believes that the benefits of super intelligence

must be reconciled with these risks.

Think of all the things that...

the human species might have been able to achieve in the fullness of time.

Maybe you would have a cure for aging.

Maybe you would have space colonization,

all these kinds of science fiction-like things,

but all of those things, I think,

might happen very quickly done on digital timescales.

In the long-term, I think we have an additional

and very different and distinct and unique problem.

Which is, how could you ensure that this super-intelligent system

will actually do what you intended for it to do?

How could you control an intelligence vastly greater than your own?

There's always this concern

of some kind of telescoping, singularity situation

where a sufficiently intelligent computer

is able to build more intelligent computers,

which leads to some kind of spiraling effect.

I think if you ask most AI researchers or experts,

they'll probably give you the same answer:

that a lot of these more science fiction scenarios are just that, science fiction.

Most technologies, they're kind of inherently neutral,

but it depends on how society uses them and deploys them

that ends up determining whether they end up being for good or for bad.

As researchers work to balance

AI's risks and possibilities,

much of the science fiction of the past is becoming today's reality,

and the world's of artificial intelligence and robotics

are on track to achieve breakthroughs in technology and science,

that have never been possible for humans before.

What I think is the most important thing is

to use AI to advance science and medicine.

Things like diseases,

climate, even things like particle physics, macroeconomics.

These are all hugely complex systems now that we're trying to deal with

and we're trying to get an understanding of.

I kind of dream about the day when AI's good enough

that we can have this idea of AI scientists that work,

hand-in-hand, with human experts

to allow us to make breakthroughs

much faster than we're able to at the moment.

And do you think that will happen soon?

I think that could happen in the next 20 years. Yeah, sure.