All language subtitles for Secrets.Of.Big.Data.S01E01.WEBRip.x264-ION10

Narrator: A Royal Navy Warship

has its location
mysteriously faked,

when it's automated tracking
data is compromised.

Creating a tense standoff
in Russian waters.

Nik Badminton: Every fake data
shadow could result in military

action based on a lie.

Narrator: A LIDAR scan
of the Amazon rainforest

reveals several mysterious
circle shaped depressions

on the jungle floor.

Kris Alexander: Between 3
and 32 mounds are present.

Ramona Pringle: So there must
be some human involvement.

♪

Narrator: An American teenager
is beaten and arrested

after parking in an
area determined

to be a hotspot for crime
by a computer algorithm.

Mhairi Aitken: You're not
predicting crimes anymore,

you're predicting the
likelihood of their arrest.

Anthony Morgan: And that is
when the real problem occurs,

when police stop somebody,
not because they're guilty,

but they're just in the
wrong place at the wrong time.

♪

Narrator: These are the stories
of the future

that big data is bringing
to our doorsteps.

The real world impact of
predictions and surveillance.

The power of artificial
intelligence

and autonomous machines.

For better or for worse,

these are the
Secrets of Big Data.

The Black Sea.

A tense military hotspot
since Russia annexed

the Crimean peninsula from the
Ukraine in 2014,

an act that prompted
international condemnation

and severe economic sanctions.

The British Royal Navy
destroyer, HMS Defender,

sails from Odessa in southern
Ukraine to Georgia.

It should be a routine voyage.

Anthony Morgan: For the British,
the ship's passage is a

Freedom of Navigation Operation,
an exercise of their right

to sail freely in
international waters.

♪

Narrator: In order
to make this journey,

the Defender has to pass by
the southern coast

of the Russian-occupied
Crimean peninsula.

But the Russians have no
intention of allowing

the British passage through
territory it considers its own,

so they follow it.

Anthony Morgan: Two Russian
coast guard ships now shadow

the British warship.

And the crew of the Defender
detect

20 Russian military aircraft
in the surrounding skies.

Narrator: A BBC news crew,
embedded on the HMS Defender,

witness the encounter.

And the Russians have their own
news crew onboard

one of their coast guard ships.

Anthony Morgan: The Russian's
warn the Defender

they'd better change course...

Russian Coast Guard Officer:
Zulu Lima 1-0-1,

change your course to starboard
and follow clear.

Do you read me? Over.

Anthony Morgan: But Commander
Vince Owen says they're just

sailing through Ukrainian waters

in accordance with
international law.

HMS Defender Vince Owen:
Russian Coast Guard ship,

this is British Warship
Delta Three Six.

We are conducting
innocent passage.

We have the right to do so in
accordance with international

law and United Nations
convention on board the sea.

Narrator: The Russians then
make a shocking threat:

if the Defender doesn't leave
their territory,

they'll fire.

It's now a tense
military standoff

with hundreds of lives at stake.

But why are the Russians
behaving this way?

The Defender is sailing
in international waters.

Mhairi Aitken: Something has
provoked the Russian aggression

that the British
aren't aware of,

AIS signals that the Defender
was sending out

a few days before the incident.

Narrator: Automatic
Identification System signals

are part of a global
safety system

that keeps track
of marine traffic.

On June 19th, Russian officials
were alerted that the Defender's

AIS data showed it approaching
their naval base on the Crimean

peninsula, just three kilometres
off-shore from Sevastopol.

Mhairi Aitken: Sevastopol is the
headquarters for Russia's

Black Sea fleet, so a British
encroachment like this

would certainly justify the
Russian threats days later

when it returned to the area.

Narrator: But the ship had
been nowhere near Sevastopol.

Mhairi Aitken: A live YouTube
webcam feed from June 19th

shows the Defender
docked in Odessa.

That's almost 300 kilometers
from the naval base

where the Russians located
it through AIS.

Narrator: How could the
Defender's AIS data show it near

the mouth of the Sevastopol
harbour if it was never there?

The answers lie within the inner
workings of AIS technology.

Mhairi Aitken: AIS is a wireless
radio technology first designed

to prevent collisions on the
world's crowded waterways.

Anthony Morgan: At its simplest
level, AIS is just sending

radio signals between
receivers on ships.

Every few seconds, a ship's
AIS transponder broadcasts

its identification, its route,
its location, and its speed.

Narrator: Since
its launch in 2002,

technological advances have led
to AIS contributions

to weather prediction, search
and rescue missions,

and the prevention of
environmental crimes.

Anthony Morgan: The data travels
through a global AIS network

that includes ships,
receivers, and satellites.

Then it's aggregated by sites
like MarineTraffic and AISHub

and it's posted online as
open-source information.

Narrator: That's where
the trouble begins.

Since 2020, data analysts with
organizations like Skytruth,

an environmental watchdog,
have noticed

a startling and disturbing
occurrence.

Mhairi Aitken: Someone is
feeding fake information

into the global AIS network.

After analysing massive
amounts of data from ships

all over the world,
analysts have proven

that nearly 100 sets
have been falsified.

Narrator: The analysts are able
to compare a ship's AIS data

with satellite images
of the same ship.

In some cases, the vessels are
thousands of kilometers away

from where their AIS tracks
say they're located.

But what's most ominous is that
almost all of the fake AIS data

has misidentified the locations
of NATO and European warships.

Anthony Morgan: If false AIS
locations of military vessels

go unchecked, it can trigger
international incidents,

even war.

Narrator: 19 kilometers off the
coast of the Crimean peninsula,

the HMS Defender sticks
resolutely to its course,

ignoring threats from the
Russian military

to leave its waters.

But the Russians, believing the
fake AIS information,

don't know if the Defender has
other intentions.

Russian Officer:
Delta Three Six.

Keep away from borderline
of Russian Federation.

Do you read me? Over.

HMS Defender Vince Owen:
Russian Coast Guard Ship.

This is British Warship
Delta-Three-Six.

Your actions are unsafe,
unprofessional,

and endanger both
of our vessels.

Are you threatening me? Over.

Anthony Morgan: The British have
no idea that the Russians

are responding to the faked
sighting of the Defender

near their naval base, and so
they prepare to fight.

Narrator: As his officers
man their action stations,

Commander Owen
assures the Russians

he is merely seeking safe
passage through an

internationally recognized
shipping lane.

HMS Defender Vince Owen:
Russian Coast Guard vessel,

this is British Warship
Delta Three Six.

I will continue to follow the
internationally recognized route

nine-zero, into
international waters. Over.

Narrator: But the Russians
keep up their threats.

Russian Officer:
British Warship, if you don't

change the course, I will fire!

Narrator: The Royal Navy crew
are primed and ready

for conflict. Protective gear,
weapons systems loaded.

The two Russian
coast guard ships

are now as close
as 100 metres to the Defender.

And suddenly fire their weapons.

[weapon shots fired]

Mhairi Aitken: They're
just warning shots.

But the Russians are
clearly signaling

that they mean business.

Narrator: The Russians then
take their threat

to the next level.

[jet engines whooshing]

An SU-24 jet drops
four bombs directly

in the HMS Defender's path.

How long can the British
maintain their course

before a battle erupts on the
Black Sea?

[jet engines whooshing]

Although the Defender is ready
to fight back if necessary,

the Commander keeps the ship on
its original course,

and the Russian's stand down.

♪

Anthony Morgan: About half an
hour later, the Defender

has passed through Crimean
waters unscathed.

It is a huge sigh of relief
for everybody involved.

And in the aftermath,
everybody tries to unpack

what exactly happened here?

Narrator: In the days following
the Black Sea confrontation,

the Russians sound off publicly
with loud accusations against

British infringement on
their territories.

Anthony Morgan: The Russian
Defense Ministry called

the British warship's actions
"dangerous"

and a "gross violation"

and it demanded an investigation
of the ship's crew.

Narrator: The skirmish sparked
a war of words

between the two nations.

The British Ministry of Defense
minimized the incident

by saying they believed "the
Russians were undertaking

a gunnery exercise
in the Black Sea"...

Mhairi Aitken: The British Naval
attache in Moscow was warned

by the Russian foreign ministry
that the next time

a similar incident occurs, his
country's forces would exercise

their military might.

Narrator: But in the background,
intelligence experts were trying

to get to the truth about what
had provoked the Russians.

They were shocked to discover
that the HMS Defender's

AIS readings had been faked.

Analysts attempt to figure out
how the data was spoofed

and then introduced into
the global AIS network.

Anthony Morgan: Because of the
unencrypted nature of AIS data,

it's difficult to determine
where the fake signal came from.

But analysts can tell when it
first entered the data stream,

by looking at the first
receiver to pick up the signal.

And it's very telling.

Narrator: As in their
100 previous discoveries,

the analysts find the Defender's
fake data tracks

come from shore-based AIS
receivers, not satellites.

Mhairi Aitken: This eliminates
radio transmissions as a source,

because a satellite would
have picked up on anything

transmitted by a radio.

Narrator: But if the fake AIS
data didn't come from

a radio transmission,
how did it end up

in the aggregated data stream?

And who's spoofing the
data in the first place?

Mhairi Aitken: It really comes
down to who would benefit

from this kind of confrontation?

Who stands to gain from
military tension

between Russia and Britain?

Some pointed the finger at
the Russians themselves.

Narrator: The actions
are consistent

with the disinformation tactics

the Russians have
been accused of using

on a number of occasions.

If they had made a military
strike on the Defender,

they could have cited
the faked AIS sighting

and claimed that NATO ships were
operating in their waters.

The fact that it wasn't true
would be irrelevant.

Is the entire encounter
a clever Russian tactic

to muddy the political waters?

Nik Badminton: Disinformation
is a Russian specialty.

They inject this data
into our systems,

so we can't tell what's true
from what's false.

Narrator: But even the
Russians have been victims

of falsified positions.

♪ [suspenseful music]

Analysts discover that just
days before the

HMS Defender incident,

AIS data showed the
Russian warship Pavel Derzhavin

in Ukraine's territorial
waters near Odessa.

Anthony Morgan: The Pavel
Derzhavin was actually

many miles away at the time.
So that clearly didn't happen.

And during that same time, AIS
data tracked a Russian corvette

going from Kaliningrad
right into Polish waters.

Narrator: But since it would
have launched an

international incident, the
Russian encroachment

almost certainly never occurred.

If the Russians are not behind
the fake AIS data,

who is trying to spark an
international conflict?

When data analysts examine the
falsified positions further,

they discover similarities
in their patterns.

Mhairi Aitken: The shared
characteristics suggest

a common perpetrator.

But it still begs the question
of how the fake AIS data

was introduced to the network.

Nik Badminton: Analysts have
come up with a plausible theory.

They think that people have
taken data from these

AIS systems, they've manipulated
them, and they've fed that

back in through open sourced
data streams,

and that misinformation is
hidden in the sea of data.

Narrator: The researchers'
theory prompts

a frightening conclusion,

the perpetrators could be anyone
connected to the internet.

Could a random hacker be
responsible for it all?

Anthony Morgan: It could be a
hacker, it could be somebody

with a political axe to grind,

it could be a technical
problem...

The reality is we
just don't know.

Narrator: The political threat
from false AIS incidents

keeps worsening.

Analysts discover that five days
before the Defender was spoofed,

an American warship was faked
passing north through Crimea's

Kerch Strait, even though
Russia has closed the strait

to foreign warships.

Mhairi Aitken: And in addition
to the hotly disputed

Black Sea waters near Crimea,
11 fake AIS simulations

of NATO allied warships show
incursions

into Russian waters near
Kaliningrad and Murmansk.

Nik Badminton: Every fake
data shadow could result

in military action
based on a lie.

Skytruth researchers have
confirmed fake AIS positions

for 15 naval vessels across
seven countries,

and they're identifying more
every single month.

Narrator: As the team at
Skytruth continues the hunt

for more phantom warships,

elsewhere, work
has already begun

on devising measures to
fight the growing problem.

♪

Analysts around the world are
studying ways to safeguard

the reliability of AIS data.

Some believe the solution lies
in adding a digital signature

to each AIS message
that is sent.

Anthony Morgan: And that could
make a big difference,

because these systems were
invented before engineers

recognized the importance of
encrypting networks like these.

Narrator: But until
a solution is found,

the threat of war increases as
these Phantom Warships

continue to sail some of the
world's most volatile waters,

perhaps instigating further
international friction.

And the world will
have to hold its breath

that the next incident doesn't
have a more disastrous outcome.

♪

♪ [upbeat music]

♪♪

♪ [suspenseful music]

Narrator: One afternoon in
Manchester, New Hampshire,

18 year old Connor Deleire sits
alone in the driver's seat

of his parked car.

For some reason,
local law enforcement

finds his presence suspicious.

Anthony Morgan: A cop cruising
in the area spots Deleire

and decides to check on him.

And he can't help but
notice immediately

that Deleire is quite nervous.

His hand is shaking as
he hands him his ID.

Narrator: Deleire claims he's
just waiting for a friend

to pick up his niece in
a nearby building.

But the officer doesn't
believe him.

Anthony Morgan: Deleire
starts to get worked up.

The more questions he's asked,
the more agitated he gets.

Narrator: More cops
arrive on the scene

and Deleire is asked
to step out of his car.

When he does, one of the
officers sees,

he's got a holstered handgun.

Anthony Morgan: The presence
of a gun changes everything.

The cops quickly disarm Deleire

and put his gun safely
out of reach.

Narrator: Police insist on
searching Deleire and his car,

but he refuses to cooperate and
the situation escalates quickly.

The teenager's resistance leads
Police to believe

they have caught a dangerous
criminal red-handed.

All thanks to an application
of Big Data called

Predictive Policing.

Mhairi Aitken: Predictive
Policing uses complex algorithms

to analyze massive amounts of
historical data in an attempt

to predict when and where crimes
are about to occur.

Some believe the
technology can actually

prevent crimes
from happening.

Narrator: Police in the US
have been using technology

for decades to assist their
fight against crime.

The Reagan administration
began working with

computer generated crime maps
in the '80s,

and in the '90s the NYPD
invented CompStat,

a data-driven statistical tool,
to reduce crime.

Anthony Morgan: But in
2010, a big shift happens.

The American Justice Department
starts using Google's software.

The same software that they use
to run their search engines

is now being used to run
predictive policing algorithms.

Narrator: Computer systems are
fed historical crime data,

such as location, time,
and type of past crimes,

and then that information is
linked to more people-based data

like criminal records,
substance abuse history,

and even age, gender,
and marital status.

Using this data, algorithms
attempt to determine

where and when offenses are
most likely to occur.

Nik Badminton: Predictive
Policing and accessing data

and using algorithms to
predict where crime is,

is a very attractive prospect

for police forces
across the world.

So these ideas
spread like wildfire

and the idea is to reduce crime.

But we have to wonder: Is
that really what's happening?

Narrator: In 2011, the LAPD
launches Operation LASER

to predict where gun violence
is likely to occur.

Then they target property crimes
with a software called PredPol,

which determines "hot spots"
where crimes will happen.

Anthony Morgan: PredPol becomes
the gold standard predictive

software for dozens
of police departments.

Nik Badminton: Cities like
Chicago have been able to

implement predictive policing
systems quite effectively

to reduce gun crime
and to predict

where violence may happen.

Narrator: The new technology is
so ground-breaking,

it spreads around the world.

Major police departments in
China, Denmark, Germany, India,

the Netherlands, Canada,
and the United Kingdom

are either deploying
predictive policing tools

or testing them for future use.

Anthony Morgan: Police
forces in smaller cities

then begin to invest
in this powerful software.

Narrator: Like Manchester,
New Hampshire,

where Connor Deleire sat in a
parked car completely unaware

of the recent advances
in Big Data.

Mhairi Aitken: Deleire has no
idea he's in one of Manchester's

highest crime areas and that
just by being present

in a "predictive hot spot",
he is a cause for suspicion.

Narrator: After the discovery
of Deliere's gun,

the Police conduct
a pat down search.

An officer feels a suspicious
object on Deleire's person.

The teenager then reaches into
his pocket

and takes some items out,
putting them on the car roof.

Among the items is a fully
loaded gun magazine.

Anthony Morgan: And that's when
the police say Deleire started

thrashing back and
forwards and wouldn't stop.

Narrator: Attempts to arrest
Deleire only result in more

resistance, so police
resort to force,

blasting him with pepper spray.

But the teen isn't
going down easily

and the cops ultimately have to
use a taser to subdue him.

Anthony Morgan: Because police
are in that neighborhood

at that time,
they're nearly certain

that they have stopped a violent
gun crime in its tracks.

Narrator: Deleire's arrest
supports claims

by Manchester police officials

that their Predictive Policing
tools are working

exactly as they should,

by reducing crime
using fewer resources.

Anthony Morgan: There are
many police departments

that are strongly in favor of
predictive policing,

because they believe it means

that they can do more with less.

Narrator: Around the world,
there are many companies with

competing softwares, all vying
to supply police forces

with what they need to prevent
crimes from happening.

Nik Badminton: Capitalism
ultimately loves crime,

there's chaos and
complexity there.

There's data underneath
that they can capture,

and there are systems that they
can use to solve the problems

that are perceived in society.

But it kind of ignores like the

root causes of some
of these problems;

the societal issues
around trauma and poverty.

Narrator: Civil Rights advocates
in the United States also take

issue with predictive policing.

They claim it undermines a
citizens' constitutional rights

- specifically the Fourth
Amendment, which guards

against "unreasonable
search and seizure".

Anthony Morgan:
The Fourth Amendment requires

that police have a "reasonable
suspicion" to stop someone.

Critics argue that Predictive
Policing makes it easier

for police to meet
that standard,

even if somebody is innocent.

And that is when the real
problem occurs,

when police stop somebody,
not because they're guilty,

but they're just in the
wrong place at the wrong time.

Narrator: So was this why
Connor Deleire was arrested?

Manchester police insist that
Deleire displayed

criminal traits, while
waiting in his car,

and their suspicions
of him were justified.

Anthony Morgan: The first
officer said Deleire was looking

into his lap, which they took to
be a sign of potential drug use.

Plus, he was an out of towner
sitting alone in a parked car,

which unfortunately, is also the
MO of somebody looking

for a prostitute.

Since Deleire was in a drug
and prostitution hotspot,

the police felt they had all the

reasonable suspicion
they needed.

Narrator: Even if approaching
Deleire was justified,

critics of predictive
policing claim

that there are bigger
problems with the system.

They believe that by branding
some neighbourhoods as hotspots,

officers expect trouble on
patrol and are more likely

to stop and arrest people
because of prejudice,

without any real justification.

Civil Rights Advocates believe
the root of the issue

lies in the data itself.

Mhairi Aitken: There's a
prejudice in the data being fed

into their algorithms because
predictive policing algorithms

are programmed using
historical data.

They reflect the biases and
prejudices in that data.

Narrator: This is referred to as
'Dirty Data'

which reinforces police bias

toward poor
and minority communities.

If arrest rates stating you're
twice as likely

to be arrested if you're Black
than if you're White

are fed into a predictive
algorithm,

it reinforces the over-policing

of communities of color.

Mhairi Aitken: It quickly
becomes a vicious cycle.

The algorithm is predicting
where crime is likely to occur,

or who is likely to commit a
crime, based on "dirty data",

which reflects the historic
biases or prejudices

or previous police decisions.

But by making predictions
based on this dirty data,

the algorithm is
continuing to perpetuate

a long history of
discriminatory policing.

Narrator: A Predictive Policing
algorithm loop in Florida's

Pasco County leads to the
unwarranted harassment

of a 15-year-old and his family.

After he committed a minor
felony, the system began

to target him as a potential
repeat offender.

Anthony Morgan: After he was
arrested for stealing bicycles

from a garage, the police
algorithm dispatched officers

to his home, to his gym,
to his parents place of work.

They harassed him at home

21 times in five months.

Narrator: And in Australia,
predictive tools

targeting teenagers are
causing an outcry

from concerned citizens and
advocacy groups.

In Melbourne, these algorithms
were supposed to identify people

who were at high risk of
committing crimes,

but ended up surveilling
innocent indigenous children.

Mhairi Aitken: If you say that
a certain group of kids

is more likely to commit
crimes in the future,

you're basically putting
an 'X' on their backs.

You're not predicting
crimes anymore,

you're predicting the
likelihood of their arrest.

Narrator: At a time when
tensions between police

and minority groups are inflamed

in many of the world's
major cities,

Predictive policing is seen by
some as stoking the fire.

Anthony Morgan: The Connor
Deleire case is a troubling

example of just how Predictive
Policing can go awry.

In the end, his
story checked out.

He was simply
waiting for a friend.

Narrator: Deleire also has a
conceal and carry permit,

so his handgun and magazine
are perfectly legal.

Nik Badminton: So Connor Deleire

ultimately was an
innocent suspect,

but he was profiled by
algorithms and designated

to be someone of interest,
and his behaviour,

maybe looking down
in a suspicious way,

made police want to
interact with him.

And they rewarded that
interaction with facial trauma

and concussion when he was a
completely innocent bystander.

Narrator: Police charge Deleire
with only one crime:

resisting arrest.

Mhairi Aitken: In this case,
police presence

at a so-called "hotspot"
created a crime

that simply wouldn't have
happened otherwise.

Nik Badminton: This is a painful
reminder for Connor Deleire

and the community at large

of what predictive policing
can deliver:

wrongful arrest, trauma,
violence,

and the creation of
criminal situations

that maybe don't exist,
or shouldn't really

be looked for in
the first place.

Narrator: With its inherent
ethical problems the future

of Predictive Policing is a hot
button issue around the world.

In the US, the LAPD's
Operation LASER,

which was meant
to predict gun crimes,

has been shuttered because of
dirty data issues.

Chicago's program was terminated
when they were unable

to prove it reduces crime.

In 2020, Santa Cruz, California
became the first US city

to actually ban the municipal
use of predictive policing,

because of fears that it
perpetuated racial inequality.

Anthony Morgan: Despite all of
the prejudices

with Predictive Policing,
many police departments,

including the NYPD, have begun
to intensify their use

of these kinds of tools.

They're hopeful that they can
address these biases

to improve their programs.

Narrator: Many advocates
believe the survival of

Predictive Policing depends
on increased transparency

and greater accountability
from local police departments.

Mhairi Aitken: Communities
need to know how and why

police are using
this new technology.

With more transparency, we may
be able to evaluate collectively

the effectiveness and
direction of these programs.

Narrator: What the future holds
for Predictive Policing

is anyone's guess,

but one thing its critics say is
abundantly clear,

there is work to be done
and the current system

is inherently flawed and those
flaws need to be addressed.

So wherever Predictive
Policing goes,

it looks like controversy is
sure to follow.

♪

♪ [dramatic music]

♪

Narrator: In a remote section
of western Brazil,

deep in the heart
of the Amazon rainforest,

a team of international
researchers is performing

an aerial LiDar survey
when they come across

something peculiar
on the jungle floor.

Anthony Morgan: This part of
Brazil is extremely isolated

and largely uninhabited.

It is dense
rainforest out there.

Narrator: Examining
three-dimensional digital maps

created by the LiDar scan, a
process that involves billions

of lasers being shot
from their helicopter

through the thick canopy and
onto the forest floor,

the team notices a series of
circular depressions

on the ground that don't fit in
with the natural habitat.

Ramona Pringle: The area is
virtually untouched by humans,

so to come across something like
this in the middle of nowhere

is definitely unusual.

Narrator: What are
these strange circles?

The team hopes a thorough
examination of the LiDar data

will provide some answers.

Anthony Morgan: LiDar is
shorthand for

light detection and ranging.

It's essentially a remote
sensing system used to determine

the exact distance to an object,
or in this case,

to the surface of the earth.

Narrator: Whereas sonar uses
sound waves to measure distances

and radar uses radio waves,
LiDar technology relies on

light pulses, making it far more
accurate and versatile.

Kris Alexander: These light
pulses are directed at an object

or the ground and then the
transmitter calculates

the amount of time it takes for
the light to return

and determines the distance.

Narrator: LiDar systems are
extremely accurate.

The margin of error is only
around 15 centimetres vertically

and 40 centimetres horizontally.

Some transmitters are
able to send over

160,000 light pulses per second,

collecting a huge amount of data
in a short time.

This information, combined with
other input recorded

by the system, is then used to
automatically generate

precise 3D images of the
targeted area.

Ramona Pringle: The earliest
experiments with LiDar date back

to the early 1960s,
not long after the laser

was first developed.

Narrator: The first LiDar
systems were designed for

satellite tracking and were
similar to today's technology,

only much more primitive.

By the 1970s, NASA began using
laser-focused remote sensing

in its spacecraft.

Anthony Morgan: The last three
Apollo missions were equipped

with a LiDar altimeter
to scan the terrain

looking for possible
landing sites.

Narrator: But it wasn't until
the late 1980s that LIDAR became

a feasible and extremely precise
tool for scientists,

mostly because GPS became
available for public use.

By the mid-1990s, LIDAR
transmitters were able

to generate up to 25,000 pulses
per second and were mainly used

for topographic mapping of
the earth's surface.

This technological advancement
and increase in usage

led to many new discoveries,
just like the circles

in the Amazon rainforest.

As the research team pores
over the LiDar imaging,

theories about the origin of the
patterns come to light.

Kris Alexander: Maybe the
circles were caused

by ancient meteorites.

Over thousands of years, the
craters could have become

overgrown with vegetation,

resulting in their modern day
appearance.

Narrator: While meteorite
strikes are certainly

a plausible explanation,

there are problems
with this theory.

Ramona Pringle: Most meteorite
craters have a raised,

somewhat consistent rim as a
result of the high velocity

impact that created them.

The Amazon patterns appear to
consist of a series of mounds

that form the outside
of the circles.

Anthony Morgan: It also seems
unlikely that so many meteorites

big enough to make a mark on the
surface of the Earth

would have been found in such a
small geographical area.

Narrator: Though the source of
the circles remains a puzzle,

their discovery should
hardly be surprising.

The Amazon rainforest
is full of mysteries.

In fact, this isn't the first
time that a LiDar survey

of the region revealed
something astounding.

In 2019, a team of scientists
who were scanning the rainforest

to create a biomass map
discovered several groves

of gigantic trees on the border
between Pará and Amapá states

in Northern Brazil.

Ramona Pringle: These trees
are absolutely massive.

Way bigger than anything else
previously found in the Amazon.

Narrator: One specimen,
a red angelim,

stands even taller
amongst the giants.

Measuring in at
88.5 meters tall,

it's a staggering
30 meters taller

than the previous record holder.

This behemoth is roughly the
same height as a 30 story

building and nearly as tall
as the Statue of Liberty.

Anthony Morgan: Researchers
believe that this tree is over

400 years old and is capable of
storing up to 40 tons of carbon.

That's an entire
hectare of rainforest,

or about two and-a-half
football fields.

Narrator: It remains unclear as
to how this particular group

of trees managed
to grow so tall,

but scientists have
thoughts on the matter.

Kris Alexander: These trees
may have taken advantage

of a disruption that weakened
this section of the forest.

Human interference or severe
weather may have created

a void in the canopy and these
trees simply grew

into the space as a
pioneer species.

Narrator: The research team that
had discovered the mysterious

circles takes a deeper dive
into the LiDar maps

and notices something that leads
them to believe

that these circles are not
naturally occurring.

There appears to be several
rectangular outlines nearby.

Ramona Pringle: Near perfect
rectangles don't occur

naturally, so there must
be some human involvement.

Narrator: The team also
observes something else

that catches their attention,

a series of straight lines

that seem to be extending
outwardly from the circles.

They can only conclude that
these patterns

are definitely made by humans.

But what are they, exactly?

Kris Alexander:
The researchers determine that

what we are seeing here is
actually the archaeological

remains of a series of very old
intertwined indigenous villages.

Narrator: Dating from between
1300 and 1700 A.D.,

the villages all have strikingly
similar arrangements,

with elongated mounds
surrounding a central plaza,

like symbols on a clock,

or the rays of the sun.

It is a significant discovery.

There are 25 circular
mound villages

and 11 rectangular
mound villages,

with another 15 or so,
too degraded to be categorized.

Ramona Pringle: The circular
villages have an average

diameter of about 86 metres,
while the rectangular layouts

are a little bit smaller, with
an average length of around

45 metres.

Narrator: Further analysis of
the LiDar data confirms

that the indented lines
stretching from the circles

are meticulously designed roads,

with each village having two
primary avenues

that are deep and wide with
elevated banks,

leading to other villages,

and lesser roads that lead to
adjacent streams.

Most villages have two roads
heading to the south,

and two to the north.

A. Morgan: Many of these
villages are pretty close

together, only about
five kilometers apart.

The main roads connect
many of these communities,

forming an extensive network
deep in the heart of the jungle.

Narrator: It appears the
villages are arranged

to represent distinct
social models,

with no apparent hierarchy.

Some researchers also believe
it's feasible

that this arrangement is meant
to symbolize the cosmos.

Kris Alexander: Between 3 and
32 mounds are present

at each village site.

Some mounds measure up to
3 metres in height,

and extend up to
20 metres in length.

Narrator: Future excavation
on the ground should reveal

exactly what these mounds
were used for,

but the most popular theory is

that they were where the houses
of the villagers were situated

and possibly burial locations.

Ramona Pringle: Overall,
this is an important

archaeological discovery.

For years, experts believed
that this region

was largely unpopulated before
European colonization.

Narrator: With the help
of LiDar technology,

these findings paint a picture
of the long and rich

human history of the area
and highlight

the complex structure of
indigenous societies

long falsely considered
to be primitive.

And as the use of LiDar mapping
continues to proliferate

in the exploration of the
far-flung corners of our planet,

who knows what secrets
are out there,

just waiting to be discovered?

♪ [dramatic music]

♪

Narrator: October 29, 1999.

After gaining strength off
India's east coast for 4 days,

Cyclone Odisha reaches
super-cyclone intensity

and makes landfall, raking the
state of Odisha

with torrential rains,
a 6 metre storm surge,

and hurricane-level winds
of 260 kilometres per hour.

In the coastal town of Puri,
local merchant

Kalpreddy Satyavati flees with
her family and other residents

into a nearby forest.

Kris Alexander: It was
so bad that Kalpreddy

and her three daughters had to
hang onto trees for survival.

She and her family pulled
through, but she's the owner

of a tiny food stall there
in the fishing community.

She lost everything
she owned in the storm.

Narrator: Cyclone Odisha
lays waste to 1.6 million homes

in the state's coastal
towns and villages,

wiping out the region's
sugarcane and rice crops

and resulting in nearly
10,000 lives lost.

The devastation totals
billions of US dollars.

In the two decades since
the deadly storm,

climate change has triggered
a surge

in extreme weather conditions
around the world,

with earthquakes, hurricanes,
floods, and wildfires

wreaking havoc on
people and their property.

Ramona Pringle: It seems like
it's getting worse every year.

In 2020, the global cost of
natural disasters

totalled around
$200 billion US dollars.

Narrator: But work is well
under way to try to reduce

the repercussions
of future cataclysms.

Throughout the world, Big Data
is starting to be used

to predict and lessen the
impact of natural disasters.

Nik Badminton: Researchers can
apply artificial intelligence

machine learning against big
data sets of every disaster

that's happened across history,
and then mixing that with

real-time information from
sensors out in the world

to help us predict where
natural disasters may happen.

Narrator: By playing out
millions of scenarios,

AI is able to identify some of
the contributing factors

that occur right before certain
natural disasters.

In 2018, Google launches the

Flood Forecasting Initiative
in India.

Its systems analyze a mix of
past and present data

about rainfall and river levels,
to create new modeling

that has the power to
predict adverse events.

Kris Alexander: The data is
collected from satellite images,

maps, flood simulations, and
even citizens with stream gauges

who are hired by local
governments

to measure water levels.

Narrator: Google claims their
forecasts can predict

a flood's water level with an
accuracy of plus or minus

15 centimetres, more than
90 percent of the time.

Ramona Pringle:
Once a flood is forecasted,

Google sends alerts through
smartphone notifications,

to people who are located
in at risk areas.

In Bangladesh and India alone,

Google reaches
360 million people,

and during the 2021
monsoon season,

Google sent 115 million
notifications

to 22 million people

and also worked with local
organizations

to reach people with no
smartphones.

Narrator: Some experts believe
Big Data can also achieve

what many scientists have
long-believed is impossible,

predict earthquakes.

Kris Alexander: Take something
like the tiny movements

and sounds that are present
along an earthquake's

tectonic fault lines.

AI experts are hunting within a
massive sea of data

from these shifting tectonic
plates for a single

seismic signal that could reveal
the pattern of events

that occurs before
a major earthquake.

Narrator: A seismologist at the
Los Alamos National Laboratory,

claims to have discovered
such a pattern by examining

a particular seismic signal's
energy with AI algorithms.

Ramona Pringle: This
seismologist has identified

a growth pattern in the
signal's amplitude,

during the cycle
of an earthquake.

He thinks it could be a
breakthrough,

but it's still years away
from real world application.

Nik Badminton: What we're trying
to do is take data

or apply machine learning and
really speculate

where things may happen
and at what time.

Narrator: As research into
earthquake prediction continues,

Google has started to notify
people around the world

when an earthquake begins,
with its alert systems.

It's a new service that
turns user devices

into mini seismometers.

Kris Alexander: The service
relies on the accelerometers

that already exist in
phones and tablets,

which are able to sense
earthquake vibrations

and then alert their
users of danger.

Narrator: Although it can only
give a moment's warning,

those few seconds can make a
difference in giving you time

to drop, cover, and hold on
before the shaking arrives.

Kris Alexander: It's not a
prediction before the event,

but at least people
near the event's epicentre

can receive the alert,
and get to safety.

Narrator: These types of alerts
could have been invaluable

when on November 25, 2020,

two decades after Odisha,

the east coast of India is hit

by yet another devastating
storm,

Cyclone Nivar.

With no warning system in place,
once again, local merchant

Kalpreddy Satyavati
loses all her belongings.

Ramona Pringle: Kalpreddy and
her daughters had no food

or shelter for days afterward.

Her food stall and all of her
supplies were obliterated.

Narrator: But for Kalpreddy,
Big Data now might help

mitigate the impact of these
devastating storms

that affect the world's most
impoverished regions.

SEEDS, an organization
providing disaster relief

and rehabilitation to
vulnerable communities,

is working on a possible
solution.

Kris Alexander: SEEDS received
a grant through Microsoft's

AI for Humanitarian Action
program

and created an AI model
called Sunny Lives.

While a storm is gathering,
Sunny Lives uses machine

learning and advanced data
analytics to study its likely

path and identify the residents
who are most vulnerable to it.

Nik Badminton:
This model looks at road maps,

proximity to bodies of water
and elevation factors.

The model assigns a score to
each residence that indicates

level of risk, and those
that are at the most risk,

can be sent help when
they need it the most.

Narrator: Unfortunately for
Kalpreddy Satyavati,

at the time of Cyclone Nivar,

Sunny Lives
was still in the pilot phase

so it was unable to help.

Ramona Pringle: We have very few
examples of its application

in real life scenarios because
so much of this technology

is in the
proof-of-concept stage.

Narrator: But there are
situations

where individuals are
being used as a source of data

in the aftermath of
natural disasters.

Social media posts containing
valuable metadata

like time references and geotags

can help pinpoint
victim locations,

and shared images can relay
information about

road shutdowns, flooding and
power outages.

Nik Badminton: Social media has
the benefit of being near real-

time and a place for people to
share critical information.

This is data that can be
extracted and analyzed

with artificial intelligence to
determine the hardest hit areas

so that we can send help.

Narrator: In parts of the world,
this is already happening.

Indonesia is hit by scores of
natural disasters each year.

In Jakarta, bots from the
website PetaBencana.id

monitor social media posts
and then engage in real-time

AI-assisted chats with
witnesses and victims.

Kris Alexander: These
interactions make it possible

to get immediate real-time
information from the ground,

and this is crucial to
rescue missions.

Narrator: But critics say
reliance on social media data

is rife with issues.

They cite the challenges in
verifying its accuracy

and validity, as well as an
underlying access problem.

Getting essential information
back to people in need,

when many vital communication
networks are lost

during a natural disaster is
a serious obstacle.

Nik Badminton:
These problems are real

and not easily overcome.

The models that we produce will
definitely get better over time.

And as they do, it's going
to provide more value

to the researchers
that help predict

whether these natural
disasters will happen.

Narrator: In May of 2021, this
developing technology shows

its full potential to the people
of India when Cyclone Yaas

begins to gather strength off
the East coast.

In Puri, it's another dire
threat to Kalpreddy Satyavati.

But this time the SEEDS
Sunny Lives AI model

springs into action.

Ramona Pringle: In just a few
hours, Sunny Lives takes

the predicted path of Cyclone
Yaas and runs satellite images

of the affected areas,
creating risk scores

for all of the residences.

Narrator: SEEDS members then
enter the affected communities

and distribute notices
to almost 1,000 families

in their native languages.

Each notice contains predictions
for the cyclone's landfall,

instructions on how to
secure their homes,

and relocation and
shelter information.

Ramona Pringle: Kalpreddy
Satyavati received the advisory

a day before the cyclone hit.

She secured her food stall
contents in the concrete house

of a neighbour and then packed
up her family and everything

they owned and rushed to a
recently built cyclone shelter.

Narrator: As predicted, the
storm destroyed Kalpreddy's hut.

But the advance warning meant
she and her family were safe,

their belongings saved, and her
food stall was up and running

almost immediately afterward.

The success of the SEEDS program
in dealing with Cyclone Yaas

is hopefully a harbinger
of things to come.

Lives were saved,
damage was minimized

and livelihoods were preserved.

And while there is
still much work ahead,

it's clear that Big Data can
become an indispensable tool

to help alleviate human
suffering

in the face of devastating
natural disasters.

♪