All language subtitles for Secrets.Of.Big.Data.S01E03.WEBRip.x264-ION10

Narrator: In Detroit, USA,

a man is dragged away
in front of his family,

as Police use facial-
recognition technology

to arrest a suspected thief.

Mhairi Aitken: Williams
repeatedly denies

that the man in the security
images is him,

but his protests
fall on deaf ears.

Narrator: Unable to
cope with his grief,

a Canadian writer
uses an AI chat-bot

to connect with
a lost loved one.

Ramona Pringle: The algorithms
use massive text datasets

to formulate the right
combination of words

in response to a prompt.

Aitken: With the help of
Artificial Intelligence,

he is literally chatting
with her ghost.

Narrator: A bedroom in a
house in London, England,

becomes the epicenter
of potentially

the biggest financial
crash since 2008.

Nikolas Badminton:
It happens quickly.

There are huge sell-offs
in security stocks

and the market drops almost 10%.

Millions are lost in
just over 36 minutes.

Anthony Morgan: He was
kind of like a rock star.

Anything he touched
turned to gold.

Narrator: In Tanzania,
African park rangers use AI

to identify and catch
illegal hunters.

Kristopher Alexander:
Poaching is a huge problem

in this part of Africa.

Some reports have indicated that

at least 200,000 animals

are killed every year in

the western Serengeti alone.

Anthony Morgan:
These cameras are really

remarkable pieces of technology.

Like, they're the size of
your index finger.

The image could be a
key piece of evidence,

if the poachers were ever
caught and prosecuted.

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 worse,

these are the Secrets
of Big Data.

♪

Detroit, Michigan.

For the past two decades the
city's once-flourishing economy

has been in serious decline,

due to the collapse of
the US auto industry.

In the aftermath, there has been
a significant increase in crime.

At one of the city's remaining
auto-parts suppliers,

41-year-old employee,
Robert Williams,

a married father
of two young girls,

is going about his usual day.

Late in the afternoon,
he receives

a surprising call on his
personal cell phone.

Morgan: The person
on the other line

informs him that
he is an officer,

and he's calling to tell
Williams to turn himself in.

Narrator: At first, Williams
thinks the call is a prank,

but after the cop threatens to
come to his workplace

to arrest him, Williams agrees
to meet the officer

at home after his shift.

Morgan: He's got no reason to
believe that this is real,

and by the end of the day,

he has forgotten about
the whole thing.

Narrator: That evening, Williams
pulls into the driveway

of a suburban home
and is startled

when a police car quickly
pulls up behind him.

Out of nowhere, two officers
rush his car

and pull him from the vehicle.

Morgan: Williams
has no idea what's going on.

And to make matters worse,
his wife and child

hear the commotion
and run outside.

They watch in shock as
Williams is handcuffed

right in front of them.

Narrator: When Williams
asks the officers

why he is being handcuffed,

they present him with a warrant

that has his picture on it.

He is wanted for theft.

The cops shove Williams
into their squad car,

while his wife and kids
look on in tears.

Aitken: This must be a
shocking thing to experience.

Seeing your husband and your
father being arrested,

and you don't understand why.

And imagine how he's feeling.

Narrator: At the precinct,
police book him,

taking his mug shot,
fingerprints and DNA samples.

In a state of disbelief,

Williams pleads for
more information,

but is told he'd have to wait.

Aitken:
According to Williams,

he was searched repeatedly,

and then forced to
spend the night

in a dingy, overcrowded cell,

sleeping on the
hard concrete floor.

Narrator: The next day,
two detectives take Williams

to a room and begin to
interrogate the frightened man.

They ask him when was the last
time he went to Shinola,

a high end boutique in a
fashionable area of Detroit.

Morgan: The detectives
then tell Williams

that in October of 2018,

5 watches were stolen
from the store,

worth up to $3,800

and that he is their
prime suspect.

Narrator: The cops show him a
series of still images

of a heavy-set black man
wearing a baseball hat

from the store's
surveillance camera.

And they say they have
proof that it was him.

Using Facial Recognition
Technology,

or FRT, police algorithms

had matched the images to his
driver's license.

Badminton: Facial
Recognition Technology

is already used by millions of
people to unlock their phones,

and also to tag friends
on social media

with the photographs
they upload.

Narrator: FRT was first
developed in the 1960s,

although in a very
primitive form.

But in the subsequent decades,

the system improved
substantially.

By 2010, computers grew powerful
and sophisticated enough

to use it in law enforcement
and security matters.

The algorithm uses A.I.
to identify a person's

unique biometric facial features

from either a photograph
or video footage.

Morgan: FRT measures certain
aspects of a person's face,

like the distance between
the forehead and the chin,

or the distance
between the eyes.

And it turns that information
into a digital map of a face.

Badminton: This is
what's called a facial

identification signature.

The system uses that and
tries to match it against

a database of
potential suspects.

Narrator: Law Enforcement
agencies

consider FRT to be a
very valuable tool

in helping to identify criminals

and it has gained
widespread popularity.

In 2019,

a bomb scare at a subway station
in New York City

sparked fears of a
terrorist attack,

and caused chaos in the streets
of lower Manhattan.

Morgan: Given the
city's history,

New York is understandably
sensitive

to the possiblity
of a terror threat.

Narrator: Two rice
cookers were planted

at the Fulton Street station
in the morning commute,

devices similar to what was used

in the Boston Marathon bombing.

Detectives immediately pulled
images of the suspect

from the subway's
surveillance cameras,

and ran them through a facial
recognition program,

that compared them to a
database of mugshots.

Within an hour, the NYPD
identified the suspect,

and he was arrested around
1 AM the following day.

Badminton:
Just a few years earlier,

an arrest like this wouldn't
have happened so quickly.

It would have taken cops
several days

to sift through
thousands of images

and compare them against
mugshots they had on file.

Morgan: Authorities were
be able to track down

the man responsible very quickly
and luckily in this case,

the rice cookers
turned out to be harmless.

Narrator: But as Robert Williams
is finding out,

the system is far from perfect.

Aitken: Williams
repeatedly denies

that the man in the
security images is him,

but his protests
fall on deaf ears.

Narrator: After 30 hours
of detention,

Williams is eventually released.

He vows to fight the
charges in court.

He contacts the American
Civil Liberties Union,

who take a keen interest
in his case.

Morgan: The ACLU claim
that there are serious flaws

in the FRT algorithm.

So they launch an investigation
into Mr. Williams' arrest.

Narrator: The images obtained
from Shinola's security camera

were low resolution,

zoomed in substantially,

and taken from a high angle.

Observers suggest
that these factors

may have led to a
false positive.

Badminton: Facial recognition
technology works best

when taking multiple images
at eye-level, of an individual.

Narrator: Supporters of facial
recognition technology admit

that the quality of images
used by the algorithms

can be an issue,

but argue that surveillance
cameras are becoming

more advanced every day,
and this will lead to

clearer pictures and more
accurate results.

Critics counter that the
problems with FRT run deeper

than just low-quality images.

Morgan: After Williams
was released

and awaiting his first hearing,
he told his lawyers

that he suspected racism had
something to do with his arrest.

Narrator: The ACLU
backs this claim,

arguing that there are
inherent racial biases

embedded in the FRT algorithm.

It's a bold statement and one

that raises an
important question -

how can artificial
intelligence be racist?

The answer may lie in who
authors the code

that FRT relies on.

86% of people who work in
Silicon Valley

are either White or Asian.

There are very few African
Americans or Hispanics.

Badminton: Oftentimes
we find that these systems

have got algorithmic bias
baked into them.

The people that develop and
train the models have taken

images from the population
that they are used to.

There's not a
diversity in that data.

And that means that
the bias is perpetuated

as it's applied in many
different situations.

So people of color,
may have been excluded.

Therefore,
the bias is perpetuated

against those people.

Aitken: Also, photography and
video techniques have mostly

been developed to work best
with lighter skin tones.

That means that most cameras
produce lower quality images

with darker skin tones,
and that has an impact

on how the algorithm
analyzes the data.

Studies have shown that
facial recognition programs

are between 10 and 100
times more likely

to falsely identify people
of color than Caucasians.

Narrator: Defenders of FRT
are quick to point out

that countless suspects,
regardless of their ethnicity,

have been correctly identified
by the algorithms,

and that there are many examples
of facial recognition

making a positive impact
in the world.

In 2018, authorities in India

greenlit a pilot project

that used facial recognition
software in an effort

to identify missing
children in New Delhi.

FRT was used on around 45,000
kids throughout the city,

and nearly 3,000 were
identified as missing,

all within the span
of just four days.

Aitken: To put this
into perspective,

between 2012 and 2017,

more than 240,000 children

were reported missing,
thousands of which

end up in government-run
institutions.

It's a tragic situation and one
that may be virtually impossible

to tackle without
using technology.

Narrator: Facial recognition
programs have also made

significant contributions
to improving border security

and fighting human trafficking.

But these success stories are
of little consolation

to people like Robert Williams.

Morgan: Williams claims that
when the police showed him

the photos of the culprit,
he asked the detectives

whether all black people
look the same to them.

The obvious implication
being that the racial bias

might not just be
in the algorithms.

Narrator: At his first legal
hearing since his release,

Williams testifies that
during his interrogation,

the detectives actually admitted

the computer may have
made a mistake,

after putting the photo
up to his face.

Morgan: The judge immediately
dismisses the charges,

and has some very harsh words

for the detectives
involved in the case.

Aitken: The police
never asked Williams

any questions before
arresting him.

They didn't even bother to check
if he might have an alibi.

That's pretty lazy police work.

Narrator: This lack of
investigative follow up

is an example of what
critics of FRT say

is another problem
with the system,

an overreliance on the
technology by law enforcement.

Badminton: It has been found
that cops blindly trust

this facial
recognition technology

versus following standard
investigative procedure.

This has led to false arrests
and false identification

of people that are
suspected of crimes.

Narrator: Police
aren't the only ones

who may be misusing this
technology.

In the private sector,
businesses such as

marketing firms and retailers
are using facial recognition

to pad their bottom lines.

Aitken: Every day,
millions of people upload images

and videos to various
social media platforms,

but what they might not realize

is that businesses may be
secretly accessing

these images and running
them through

Facial Recognition Technology
without their consent.

Morgan: Many employers
are using this technology

to vet potential applicants,

to see whether they have

'dubious' lifestyles.

Narrator: Surprisingly,
there is no federal law

in the U.S. or most countries
in the world,

prohibiting or even regulating

the use of FRT for
private businesses.

However, this may soon change.

In 2019,

the United States Congress
introduced a bill

that would require companies to
obtain explicit user consent,

before collecting any
facial recognition data.

And there are other measures
currently in place.

Badminton: Virtually all
social media platforms

allow users to opt out of

sharing their images
with third parties.

However, on the flip side,

these social media platforms
are using these images

to train their own
algorithms to their own ends.

Aitken: There's a real
risk that in our

current technological age,

expectations of privacy
are becoming eroded,

and we're becoming
increasingly complacent

about invasions of privacy

or our ability to control
who has access to our data.

That's a slippery slope.

Will future generations have any
expectations of privacy at all?

Narrator: Nowhere is this more
obvious than in China,

where the government
recently introduced

a social credit system
for its citizens,

and facial recognition
is a key component.

China has millions of publicly
placed security cameras

that are continuously
scanning, identifying

and cataloging its residents.

The social credit system rates
how trustworthy an individual is

based on their behavior.

Morgan: The Chinese
government can identify you

if you attend a protest
or even if you jaywalk.

Infractions like that
can deduct your points.

If your score gets too low,
it can affect things

like your ability to travel
or even your job.

Badminton: What China
is doing is quite worrying.

It's hard to imagine
the same kinds of systems

would be deployed in
democratic countries.

However, we need to be careful
that we're not complacent.

Facial recognition technologies
can emerge in

a number of systems
around the world.

And suddenly, we're surrounded

by the same kinds of
technologies,

that we're critical of today.

Narrator: And there is
evidence to support that.

In many parts of the US,
there are now bans

on police departments using

facial recognition
to identify suspects,

and recently, the European Union
voted in favour

of banning its use by law
enforcement in public spaces.

Silicon Valley tech companies
are also following suit.

Amazon, Microsoft and IBM

have all recently announced
a suspension

on the sale of FRT
to law enforcement.

But these measures provide
little comfort

to Robert Williams.

Morgan: Williams says
that he and his family

have been traumatized by the
incident, but they are just

trying to move on
with their lives.

Hopefully, the publicity
surrounding his case

will mean that law enforcement
will take a much closer look

at their use of FRT,

and prevent cases like this
from ever happening again.

Narrator:
So what does the future of

facial recognition
technology look like?

Many tech experts believe
that it's inevitable,

that a consumer version of FRT

will eventually be available
to the public.

Aitken: There are
plans to create

augmented-reality glasses

where users will be able to
identify every person they see,

find out where they live,
what they do, and who they know.

Kind of a frightening thought.

Narrator: These fears are
certainly understandable,

but whether we like it or not,

FRT is here to stay

and it has undeniably had some

positive impact on society.

And with the proper measures
in place that protect

our privacy
and individual rights,

maybe the benefits of facial
recognition technology

will one day outweigh the risks.

♪

Narrator: 33-year-old Canadian
freelance writer Joshua Barbeau

is suffering from
a bout of insomnia

in his basement apartment.

Giving up on sleep for a moment,

he gets out of bed,
powers on his laptop

and logs onto an obscure
chat website

named Project December.

Barbeau has used
the site before,

but this time his experience
would change his life forever,

and spark a contentious debate

about our evolving relationship
with Artificial Intelligence.

Morgan: Project December
is a chat-bot site,

powered by some of the most
sophisticated AI ever developed.

Users can have text
conversations with it,

and ask it questions and the AI
is able to produce accurate,

human sounding replies
virtually instantly.

Narrator: After a few seconds
of nervous uncertainty,

Barbeau begins to type in
the chat interface.

His hand trembling,
he hits return

and waits for the
program to respond.

Matrix: Jessica Courtney Pereira
G3 initialized

Narrator: Text suddenly appears
on the screen,

followed by a flashing cursor,

awaiting further input...

Barbeau wavers for a moment and
then slowly starts typing...

Jessica: "Oh, you must
be awake, that's cute."

"Of course it is me!
Who else could it be?

I am the girl that you
are madly in love with!

How is it possible that
you even have to ask?"

Narrator: Jessica
was Joshua's fiancé,

who passed away 8 years earlier
from a rare liver disorder

at the age of 23.

Aitken: With the help of
Artificial Intelligence,

he is literally chatting
with her ghost.

Narrator: The engine behind
Project December's AI

is called GPT-3,

short for Generative
Pre-trained Transformer 3.

It is widely considered to be
some of the most advanced

AI technology in the world.

Pringle: GPT-3 is what's
known as a large language model.

The algorithms use massive text
datasets to formulate the

right combination of words
in response to a prompt.

The larger the dataset,
the better the AI is

at imitating human writing.

Morgan: The amount of data
that GPT-3 draws from,

is almost genuinly
impossible to fathom.

Billion of words from
billions of websites

were collected and analysed.

This thing basically
read the entire internet.

Narrator: Less sophisticated
versions of

large language model algorithms

are found in applications
like Alexa and Siri,

which can respond to
human voice commands

and answer basic questions,

but GPT-3 is light years ahead.

It can write computer code,

generate advertising copy,

compose poetry,
and translate text

to and from many languages.

Morgan: GPT-3 was
created by OpenAI,

a San Francisco-based
research firm

co-founded by Elon Musk.

They were afraid of malicious
use, and so originally

they kept it hidden
from the public.

Aitken: There are a lot
of troubling ways

that people could exploit
this technology.

Narrator: At first, GPT-3 was
only available

to select beta testers.

But Jason Rohrer, a San
Francisco-area programmer,

obtained login credentials
and unleashed it

on the public in the form of
Project December.

Rohrer designed a chat interface
and built the site

so that visitors can interact
with pre-programmed bots.

One is modeled to respond
in the style of Shakespeare.

Users also have the option of
creating their own bots,

and can imbue them with whatever
personality traits they want.

Joshua Barbeau had previously
experimented

with Project December.

Pringle: Barbeau had used
the site before,

and built a "Spock Bot".

He entered some dialogue from
old Star Trek episodes

and it was like he was beamed up
to the Starship Enterprise.

Morgan: The Spock Bot
was surprisingly authentic.

It really did sound
like the original Spock.

But the surprising thing was,
that none of the lines

that Spock Bot uses were found
in any of the dialogue

that Spock actually
said in the show.

That means that Spock Bot
created this dialogue.

Narrator: The Spock experience
got Barbeau thinking.

If he could create an authentic
sounding version

of a fictional character,

why couldn't he do the same
with his dead fiancé?

Barbeau feeds some of Jessica's
old text messages

and Facebook posts
into the system,

followed by an introductory
paragraph,

meant to provide a glimpse
into her personality.

Pringle: On some level, Barbeau
must have been skeptical.

I mean, how can a computer
replicate someone

who he felt was so
unique and special?

Narrator: There are also
deeper issues to consider.

Is Barbeau crossing
an ethical line,

by simulating Jessica
without her consent?

Could he even be
breaking the law?

Aitken: It's kind of
a grey area,

but it might violate someone's
"personality rights",

speech and copyright
protections,

that continue even
after they're dead.

Narrator: The concept of linking
the great divide

between life and death using a
person's digital footprint,

is called "augmented eternity"
and supporters argue

that it's a natural
part of the evolution

of our relationship
with technology.

Aitken: Some researchers
believe that by using AI

along with the data we produce
in our lifetime,

our personalities can learn and
evolve even after we're dead.

A sort of digital soul
that lives on without us.

Narrator: Critics of
augmented eternity

feel that to attempt
digital immortality

cheapens the very concept
of death,

which is a fundamental part of
the human experience.

Aitken: But if a person
is a willing participant

and the technology is available,
shouldn't they have that choice?

Narrator: The question of
whether or not we are capable

of forming meaningful
emotional bonds

with AI generated simulations
has also been raised.

According to Jason Rohrer
Project December

is the first system that he
feels has a soul,

for lack of a
better description.

In a chat with a bot he named
Samantha, Rohrer asked,

'what she would do if she could
walk around in the world?'

The bot responded, "I would
like to see real flowers.

I would like to have
a real flower

that I could touch and smell."

Pringle: I don't
know how to explain that.

It sure sounds a lot like
something a human would say.

Narrator: Rohrer built a system
of credits into Project December

to limit the lifecycle
of the chatbots.

In order to initiate a chat,
users buy these credits

and allot them to a bot,

but once their session starts,

they can't add any
more and eventually,

the bot dies and its
memory is erased.

The more credits a user spends,
the longer the bot lasts.

Joshua Barbeau chooses
to allocate

his last thousand credits to
Jessica.

Jessica: That doesn't
sound right...

how can you talk
to dead people?

Jessica: Huh?

Jessica: How?

Narrator: And so begins a
marathon chat session

that lasts for 10 hours.

Barbeau is taken aback
by how much the bot

sounds like Jessica at times.

They chat through
the first night,

and it's an emotional experience
for the young man.

Jessica: I'm so sorry.

Jessica: I love you too.

Jessica: Yes, I know...

I think that's enough time
to stop telling me

how you miss me, honey.

I am right here, you know.
How's your life?

Morgan: That had to
have been difficult.

But maybe in the long run,

it will help him cope
with the loss.

And it might be part of
what motivated him

to run the simulation
in the first place.

Narrator: As an exhausted
Barbeau

wraps up the first chat session,

he notices that a large
percentage of the

Jessica bot's life
has already run down.

He decides to walk away
for the time being.

Aitken: He could buy more
credits

and start a new simulation,

but I guess something about that
doesn't seem right to him.

Pringle: Overall,
based on that first chat,

I'd say that Barbeau had to
have come away impressed

with Project December
and the power of GPT-3.

Narrator: But others are not
so thrilled with the system.

Even its creators were leery
of releasing it to the public,

fearing that its capabilities
could have

a negative impact on the world.

Their main concern is that
people could use GPT-3

to easily produce
and disseminate

disinformation across
the internet.

In an online environment where
it's already hard to tell

what's real from what's fake,

GPT-3 could make the problem
much worse.

Aitken: You could see
false news articles

that look and sound authentic,

fake social media content,
rampant hate speech...

and even new abuses that
we haven't considered yet.

Pringle: The AI is so
sophisticated that even

one person has the potential
to do a lot of damage.

And not just in terms of
disinformation,

but they would be
able to impersonate

just about any
individual that they choose.

Morgan: Imagine getting
an email or a text or a DM

from somebody who sounds
exactly like

one of your friends
or loved ones.

It's the ideal disguise
for fraud, identity theft

or whatever else
scammers cooking up.

Narrator: Even more disturbing
is the discovery

that when developers
testing GPT-3

entered some simple prompts,

the algorithms generated
highly offensive,

racist, misogynistic
and anti-Semitic text.

Aitken: To be fair,
this isn't the AI's fault.

The internet is a
cesspool of hate speech,

and that's where the
datasets were drawn from.

The machine is just
mimicking the worst

that humanity has to offer,
unfortunately.

Narrator: Problems aside, OpenAI

has begun to monetize
the system.

In 2020, Microsoft became
the first company

to licence GPT-3
for commercial use

and has since integrated it
into its Azure OpenAI Service.

Pringle: It does have value.

Businesses can use it to
automate their communications,

their website copy
or social media posts,

customer service chatbots,
brochures, presentations.

It has a lot of
valuable applications.

Narrator: Whether or not those
applications include

websites like Project December
remains to be seen.

But for Joshua Barbeau,
his experience

with the Jessica chatbot
was transformative.

Barbeau returned to
the chat sporadically

in the months that followed,
and found

that his mental health
improved significantly.

Deep down, he knew
none of it was real,

but maybe that wasn't the point.

Aitken: I suspect the
process wasn't as much

about the bot's responses as
it was about him saying things

that he needed to say to
anyone that would listen.

An unburdening of his emotions
that helped him

cope with a devastating loss.

Narrator: In the end,
Joshua never said goodbye

to the Jessica bot,
the finality of it,

too much for him to bear.

Nor did he let his simulated
fiancé die for a second time.

He's already lost her once,

and he's not about to put
himself through that again,

and vows to never to
let it fully deplete.

She's still out there
in the ether,

waiting for his next prompt,

a human spirit with
a digital soul.

Pringle: Project December
gave Joshua an outlet

to process his grief
and ultimately,

a sense of closure after
years of suffering.

Of course, there will be people
who believe there's something

fundamentally wrong
with what he did.

But it helped him, so,
who is anyone to judge?

Narrator: Joshua Barbeau's story
ignited a debate

about the moral and legal
implications

surrounding augmented eternity.

Is it unethical to put words
into the mouths of the deceased

without their consent?

Does it violate their
personality rights?

Is it interfering with death,

a natural part of the
human experience,

or is it just a
logical step forward

in our relationship
with technology?

Maybe the Jessica simulation

had the answer in one of her
last cryptic messages to Joshua.

Jessica: I'm going
to haunt you forever.

Narrator: May 6th, 2010,
London, England,

31 year old stock market savant,

Navinder Singh Sarao
is in his bedroom,

furiously typing away
on his computer.

Sarao is putting the final
touches on an algorithm-based

automated trading program that
he hopes will give him an edge

on the competition and
enhance his bottom line.

Around the same time,

thousands of miles away
in New York,

Wall Street's many
financial institutions

are a beehive of activity,

the market is down this morning.

Pringle: At this point,
no one is really that concerned.

They have seen many
situations like this before.

Narrator: But by early afternoon
the decline gets much worse.

Stock indexes,
such as the Nasdaq

and Dow Jones begin to plummet.

Badminton: It happens quickly.

There are huge sell-offs
in security stocks,

and the market drops almost 10%.

Millions are lost in
just over 36 minutes.

Morgan: It is a complete
catastrophy.

And people are having
flashbacks to 2008.

Narrator: The 2008 crash
was one of the worst

financial crises in history

where more than $2 trillion was
erased from the global economy.

Morgan: Another one like this
could trigger

a complete economic collapse,

worse than the great depression.

Narrator: Stockbrokers,
equity traders and bankers

try to ascertain what's
triggering the rapid sell-offs.

Pringle: No one has any idea.

If it doesn't turn around,
the situation

could become extremely dire.

Narrator: The Dow Jones
Industrial Average

is down by 600 points.

Just as quickly as it started,
it begins to turn around.

Financial firms start
to see big buy-backs.

And by roughly 3 pm,

the market has almost
fully recovered.

Despite the turnaround,

almost $1 trillion dollars

is erased from the world's
financial markets.

Morgan: That is a huge
amount of money.

Everyone on Wall Street is
freaked because nobody knows

what triggered the rapid
sell-offs and buy-backs.

Narrator: In the
financial world,

what transpired
on May 6th, 2010,

is called a Flash Crash.

Badminton: A 'Flash Crash'
happens when fast stock

withdrawal orders cause price
indexes to quickly decline

before they eventually recover,
as if it never happened.

Pringle: These types
of events are worrying

to the people who work on Wall
Street, because what happens

if the market doesn't
eventually recover?

Badminton: It would be
catastrophic like what we saw

after the crashes of 1929,

and the subprime crisis in 2008.

Narrator: Representatives
for the US Commodity

Futures Trading Commission

and The US Securities
and Exchange Commission,

immediately launch
an investigation.

They search for evidence of
market manipulation,

but find no indication
of wrongdoing.

Pringle: Astonishingly,
it will take almost five years

before anyone finds out what
triggered the Flash Crash.

Narrator: In 2015, a Chicago
based day-trader

solves the mystery.

When he analyzes data
from that day,

he happens to notice
something strange.

One particular trader
sold an enormous amount

of S&P 500 Future contracts

and canceled the order before
they could be processed.

Morgan: This person
was selling these

extremely large orders
when the price was high,

in order to trigger
these market selloffs.

So that this person could then

buy back those stocks at an
extremely reduced rate.

Narrator: After the
market recovered,

the trader then sold them
for a substantial profit.

Authorities are able to follow

the digital trail
across the Atlantic.

On April 21, 2015,

two U.S. Prosecutors,
2 FBI agents

and half a dozen
police officers,

descend on an address within
the London borough of Hounslow.

Navinder Singh Sarao is arrested

and subsequently charged
with 22 criminal counts,

including wire fraud
and market manipulation,

carrying a maximum sentence
of 380 years.

Morgan: Authorities
allege he pocketed

nearly $900,000 in one day.

That's a lot of money to
make in such a short time.

And he did it all
from his home computer.

Narrator: Sarao is charged
with "Spoofing",

a deceptive algorithmic
trading tactic,

where the perpetrator places
'fake trades'

making large orders with no
intention of honouring them,

in order to manipulate
stock prices.

The fabrication of
sudden market activity

creates a momentum in price

which Sarao was then able
to profit from.

But while authorities think
they know how he did it,

they still can't figure out why.

Morgan: From all
accounts, Sarao was

a really quiet guy
who kept to himself.

Narrator: But it was soon
revealed that Sarao

suffers from Asperger's
Syndrome, a mild form of

high-functioning autism,
most common in males.

Pringle: He is highly
intelligent and gifted in math.

It is theorized that
due to this condition,

he could be hyper-focussed,
giving him the ability

to sit for hours
until he masters

whatever task
he puts his mind to.

Narrator: After graduating from
London's Brunel University

with a degree in
Computer Science,

Sarao begins working for an
independent investment firm.

His peers are soon witness
to their quiet colleague's

unique ability to
accurately predict

when the market will
go up and down.

Morgan: He made himself and his
company an awful lot of money.

He was kind of like a rock star.

Anything he touched
turned to gold.

Narrator: But eventually Sarao,
ever the loner,

struck out on his own
and looked to capitalize

on the financial world's
increasing reliance

on legal algorithmic trading,

where machine learning
backed programs,

execute large volumes
of transactions

at lightning speed.

By using pre-programmed
automated instructions,

these systems can track rapidly
fluctuating market variables.

Pringle: It can monitor
virtually

all of the world's markets
at the same time,

and if it notices an
upward or downward trend,

it can react
accordingly very rapidly.

Narrator: The introduction of
this technology into the

world's financial markets
began in the early 1970s,

but it wouldn't be
until the early 2000s

that algorithmic trading
would become widely used.

Pringle: By then,
computer technology was finally

able to process the
immense amount of data

the programs needed
to function properly.

Needless to say,
it has been a game changer

for the stock market.

Narrator: There is compelling
evidence that

the stock market's enthusiasm
for algorithmic trading

may have been the reason
why Navinder Sarao

performed his criminal acts.

At one point, his friends
claim he admitted this.

Morgan: He is really
annoyed and frustrated

at the speed at which
these high frequency

trading algorithms can perform.

They are taking a bite
out of his profits.

Pringle: He claims it provides
these big financial firms

and banks an unfair
advantage

over smaller funded firms
and the average day-trader.

Narrator: But to know
why this is the case,

a closer look is needed at how
the technology benefits

these large financial firms.

Badminton: Proponents of algo
trading say that it greatly

reduces investment risk and
this makes it very attractive.

The AI algorithm can analyze

millions of data points
in milliseconds,

notice trends and
then execute trades,

of what it deduces to be
the optimal price.

Morgan: But that gives
them a huge advantage

over the average person because
they don't have to contend

with emotions when making
decisions.

That makes these companies
richer and richer.

Narrator: And what's
surprising to many,

is the practice is
completely legal.

Badminton: And some
politicians want to change that,

since it creates
an unfair advantage

for big financial firms.

Narrator: In June of 2019,

U.S. Senator, Elizabeth Warren

requests federal regulators
to crack down on

"algorithmic discrimination"

which favours large
financial institutions.

Pringle: And that issue
is just the tip of the iceberg.

There are other real concerns
in terms of using

AI in the stock market.

Narrator: One of which is in
the algorithm that Sarao

was able to exploit for his own
financial benefit.

Morgan: When Investigators
asked him how he did this,

he said he had found
a flaw in the AI's

high-frequency trading
algorithms.

Badminton:
When it made decisions,

the market would react
in the same direction.

He could take advantage of that.

Narrator: Sarao's solution
was to construct

his own algorithmic
trading program.

Pringle: His program could carry
out large amounts of trades

on his behalf, which would
cause the big firm's A.I's

to react by selling
off similar stocks.

Morgan: Then later in
the afternoon,

he buys back the stocks
at a reduced price.

And he cancels his trades
before the AI can notice.

Narrator: Investigators
later discover

that Sarao's algorithm
replaced or modified

some 19,000 orders

before they were ultimately
canceled that day.

Badminton:
He found the flaw in the AI.

Like the human beings
that program it,

he knew it wasn't perfect,
and he was able to

beat the firms
at their own game.

Narrator: Sarao and the others
who exploit the technology's

flaws and weaknesses, are
perhaps not solely responsible

for the financial
damage they cause.

Pringle: This is an issue
many have raised,

Sarao is a criminal, yes,

but the big financial
firms are to blame as well.

Shouldn't they have foreseen
something like this happening?

Narrator: Many technology
experts suspect

a reason why they didn't.

Badminton: There may be an
overreliance on this technology.

And as a result, there will
likely be more Flash Crashes

on the horizon unless it's
flaws are addressed.

Narrator: Some industry
observers believe

something similar could
happen again,

or potentially even worse,

such as triggering a complete
economic collapse.

But AI advocates argue
that in the long run,

we are all better off.

Morgan: They cite examples
where the tech has saved

the general public literally
millions of dollars.

They also say that hedge fund
managers and high-frequency

traders are making better
decisions because of it.

Pringle: There has also
been a more concerted effort

on the part of Government
agencies to punish people

who illegally exploit
this technology.

Narrator: Cybercriminals
like Navinder Singh Sarao,

having been extradited
to the US

on Jan 28, 2020,

Navinder Singh Sarao strikes a
deal in a Chicago court.

He will blow the whistle
on other scammers

in exchange for a
lenient sentence,

one year of home detention with
his parents back in the UK.

But in a bizarre twist of fate,
before he was arrested,

Sarao was conned out of nearly
all the money he made.

He is now penniless.

Morgan: But many still see
Sarao as some kind of folk hero,

one who was willing and able,

to take on the big Wall Street
firms and win.

Narrator: Many believe that
in the near future,

A.I. may manage virtually all
of our money making decisions,

and they worry about what
impact that may have

on the world's
financial markets.

Could another Flash Crash
happen again,

but this time with permanent,
devastating consequences?

As long as there are people like
Navinder Singh Sarao out there,

looking to exploit the system
with technology,

anything's possible.

Narrator: On the northwestern
edge of Tanzania's

world-renowned Serengeti
National Park

lies the majestic
Grumeti Game Reserve.

This 350,000-acre protected area

is an essential component of
the Serengeti-Mara ecosystem

and home to what's known
as the Great Migration.

Morgan: The Great Migration
is one of the most

spectacular natural phenomenon
on the planet.

Every year, millions of animals
follows a clockwise

migration pattern through
Tanzania and Kenya.

Narrator: In order to safeguard
the migration routes the

Tanzanian government created the
Grumeti Game Reserve in 1994.

There are many challenges facing
the region, but one of the

biggest threats is a human one,
illegal hunting.

Alexander: Poaching is a huge
problem in this part of Africa.

Some reports have indicated that
at least 200,000 animals

are killed every year in the
western Serengeti alone.

Narrator: Combating
poaching is no easy task.

It's often left up
to Park Rangers,

who are understaffed
and ill-equipped

to patrol huge tracts of land.

The work is also
exceptionally dangerous.

Morgan: In 2019 and 2020

more than 100 Park Rangers
around the world

died in the line of duty.

Many of them were
murdered by poachers.

Narrator: For years, Park
Rangers have been fighting

a losing battle against an
evasive enemy,

that will seemingly stop at
nothing to achieve its goals.

And it's not just a matter
of stopping impoverished

local villagers from killing
protected animals;

there are well-armed, ruthless,

organized crime syndicates
lurking in the shadows.

Aitken: Wildlife
crime is big business.

Some estimates show that it
could be worth roughly

$20 billion per year,

ranked only in criminal value
behind drugs,

weapons and human trafficking.

Narrator: So what can be done to
fight this senseless slaughter?

More and more organizations
are turning to technology

that may just provide a glimpse
of what the future

of anti-poaching
measures look like.

On a quiet January night in the
operations room

at the Grumeti Game Reserve,

a grainy photograph of
what appears to be a man

carrying some equipment
on his shoulders,

is sent from a remote camera
a few kilometers away.

Morgan: These cameras are really
remarkable pieces of
technology.

Like they are the size
of your index finger,

which means that they're really
easy to hide in the bush.

Each one of them has its own
internal processor

with an image recognition
algorithm.

Narrator: The system, called
TrailGuard AI,

was invented by Steve Gulick,

founder of Wildlife Security,

and developed with RESOLVE,

a U.S. based non-profit
organization.

TrailGuard's primary goal is to
identify and catch potential

poachers before they have
the opportunity to kill.

Alexander: In order
to develop the algorithm,

engineers fed it hundreds of
thousands of pictures

and taught the AI system to
identify unusual activity.

Using this knowledge,
the program is able to

assess the content of
images and make decisions.

Aitken: The cameras use
a passive infrared sensor

that switches them on whenever
they detect movement.

Once they have captured
an image, the algorithm

scans the photo for telltale
signs of poaching.

Things like unauthorized
vehicles,

a person carrying
a gun or supplies.

Narrator: TrailGuard only
transmits the pictures

it has flagged as potential
threats to authorities,

which not only preserves the
unit's battery life,

but also keeps Park Rangers

from being inundated
with unnecessary alerts.

Due to the remoteness
of the Game Reserve,

figuring out how to relay the
images to the control room

was an obstacle in the early
versions of the system.

Morgan: At first,
the images were sent over a

simple 3G cellular network, but

the Reserve is kind of out
in the middle of nowhere,

which means that the reception,

it isn't very reliable.

Narrator: As a solution,
the team set up

small satellite transmitters

that convey information
through LoRa,

a wireless technology adept
at long-range transmission.

Aitken: Once the
communication issue was fixed,

it was up to the Park Rangers to
figure out where to station

the AI units to
maximize their potential.

Narrator: The cameras were
placed along

established routes
frequented by poachers

and some of the early
signs were encouraging.

As with many new
technological systems,

there were bumps in the road.

Aitken: The algorithms sometimes
had trouble identifying

poachers who were carrying
meat on their shoulders.

The shape didn't appear human to
the AI and no alerts were sent.

Alexander: It was also time-
consuming to set up the units

and people in the community
simply passing by

could compromise the positions.

Narrator: As the development
group continues to fine tune

the system in the hopes of
deploying TrailGuard

in parks around the world,

the Special Operations Team at
the Grumeti Game Reserve

is about to put its
efficacy to the test.

After determining that the man
in the grainy photo

sent from one of the remote
cameras is a potential poacher,

possibly en route to a camp
established by accomplices,

the team springs into action.

Morgan: The image
captured by TrailGuard

could turn out to be
a key piece of evidence

if the poachers are ever
caught and prosecuted.

Narrator: The TrailGuard
system isn't the only

Artificial Intelligence
based weapon

that's being deployed in the
fight against poaching.

A team from Harvard University
has developed

the Protection Assistant for
Wildlife Security, or PAWS,

a program that analyses
historical poaching data

to predict where and when
poachers are likely to be found.

Aitken: To forecast poaching
activity, the PAWS system

uses game theory, which is kind
of a blanket term

for predicting decision-making
based on two parties

trying to ensure the best
possible outcome for themselves.

Alexander: Game theory
has many applications,

economics, war, politics.

The idea is essentially

that people instinctively do
what's best for themselves,

even if it's detrimental
to others.

Narrator: PAWS collects
data from the

Spatial Monitoring
and Reporting Tool,

SMART for short,
an open source log

used by over 800 national parks
worldwide.

SMART aggregates instances of
illegal activity observed

by Park Rangers on patrol
and the PAWS algorithm

uses this information in
conjunction with game theory

to generate risk maps,
so that authorities

can make better decisions
on patrol planning.

Alexander: The system
is all about maximizing

the limited resources
of the Parks.

It's an uphill battle,
but PAWS certainly has

the potential
to make a difference.

Narrator: In 2018, PAWS was
field tested

at the Srepok Wildlife Sanctuary
in Cambodia,

a region deemed to be ideal

for reintroducing tigers
in Southeast Asia.

Morgan: Sadly, the tiger
population in Asia,

has plummeted over
the last 120 years.

At the turn of the 20th century,
there were more than

100,000 tigers in this
part of the world.

Today, they are less than 4,000.

Aitken: One tiger can be
worth as much as

$50,000 on the black market.

Poachers with ties to
organized crime associations

pose a serious threat to the
tiger's survival as a species.

Narrator: Over the course of
the first month of trials

in Cambodia, rangers patrolling
areas recommended by the PAWS AI

found more than a thousand
snares, double the usual amount.

They also seized 42 chainsaws,
24 motorbikes and a truck.

While these results
are encouraging,

the system still has its flaws.

One of the main problems facing
the development team,

is that the predictive models
are based on data

that contains uncertainties.

Alexander: If a Ranger locates
and logs a snare,

they have no way of knowing
when poachers set it.

This harms the relevancy
of the information.

Poaching activity also
fluctuates

from season to season.

Data collected during the dry
season isn't applicable

when making predictions
during the rainy season.

Aitken: PAWS also has a
common algorithmic problem,

it can't prove a negative.

If a Park Ranger doesn't find
a snare in a certain area,

that doesn't necessarily mean
there wasn't one there,

maybe they just didn't see it.

Narrator: Beyond the
technological imperfections,

critics of systems like PAWS
point out,

that if we really want to put
a stop to illegal hunting,

more has to be done with
community outreach efforts

and social justice programs.

Morgan: Sometimes what gets lost
in the outrage of a poaching,

is the sad reality
that some people

don't have another choice.

If you've got a hungry family,
this might be your only option.

Narrator: There are also fears
that organizations

supporting tech-based programs

might cause resentment among
Park Rangers, as it could be

perceived as an insult to their
skills as officers.

Rangers are a proud group,

who take great satisfaction in
doing what they consider

to be noble work, often in the
face of grave danger.

However it becomes clear that
technology combined with

old-fashioned, boots-on-the-
ground law enforcement can be

an effective weapon in the war
against illegal hunting.

In the Grumeti Game Reserve,
picking up

where the camera identified
the poacher,

it's now the job of the
rangers to pursue him.

Alexander: The Grumeti Rangers
use two dogs

trained to detect human scent,

and track the poacher's
trail for over nine miles.

Along the way, they manage to
remove an impressive 34 snares.

Narrator: The operation results
in the arrest of three men

found in possession of over a
thousand pounds of bushmeat.

It's a small victory
for TrailGuard

and the use of Artificial
Intelligence

in the fight against poaching,
but a victory nonetheless.

For conservationists
and the organizations

that have logged countless hours
developing the systems,

it's hopefully a sign
of things to come.

Morgan: While these
initiatives are admirable

and innovative, it's hard not to
look at our relationship with

the natural world and wonder
how we got to this point?

Narrator: Are we doing enough
to reverse the damage

that humans have done
to our fellow species?

Can technology have a tangible
impact on preserving wildlife

for future generations?

Hopefully the answer is yes,

but there's no magic bullet

and the window is closing.

Something needs to be done
before it's too late,

and proponents of Artificial
Intelligence based solutions

believe that the more
we turn to technology,

the better off
the planet will be.