10/12/2016 Click - Short Edition

Now on BBC News, it's time for Click.

This week, mapping the poorest parts of the world.

Inside a shopper's mind.

And, hands up who is not flying the plane?

Thanks for the information.

Pay attention.

We are living in interesting times.

There are things happening in artificial intelligence right

now that will fundamentally change our world.

Soon, machines will learn to do our jobs.

They will learn to walk.

They will learn to drive.

And at that point, things get very interesting.

We'll talk more about the consequences of an automated

society in a few minutes but, after thinking, walking and driving,

have you ever wondered how hard it would be for a computer to fly?

I'm not talking about drones that can fly between points, follow

a target or perform circus tricks.

I'm talking about aircraft that could intelligently

decide on a flight plan, just as a human would.

And that is what Mark has been hunting down.

Here at BAE Systems in Woolton in Lancashire, they are used

to the roar of jet engines.

They built and have tested Eurofighter Typhoons here.

Today, however, I am going to take a flight in an aircraft that is much

slower than a Typhoon.

This is a Jetstream 31, a small passenger aircraft.

It's a design from the 1980s but it's currently used by BAE

Systems as a flying test-bed for technology which could lead

to fully autonomous aircraft.

Maureen McCue is head of research here.

Tried and tested.

Very well flown and understood aircraft from the outside,

but on the inside, it's filled with the latest technology.

That technology will eventually allow this aircraft to fly

itself autonomously.

Today, they are testing the plane's ability to detect and avoid

clouds as well as testing its satellite communication systems.

But take-off and landing will still be handled by human

pilots and the plane will be remotely controlled

at some points, too.

How does this fit into the autonomous equation?

At the moment, it's effectively a remotely controlled aircraft.

It is, and really, with autonomous operations,

you need to progressively expand the boundary.

You can't start with such a big bang right out at the full range

of capability of operations.

This humble looking outbuilding houses the grand station.

Here, a pilot will remotely fly the plane and he can ensure it

will react to instructions from air traffic control.

I would expect to see a joystick and images coming

through from the cockpit but you're not going to be flying

like that at all, are you?

No, everything is through the numbers that you can see there.

Altitude, speed.

Time for take-off.

These flights are taking place in uncongested airspace.

Today, we will be flying over the Irish Sea.

One minute to handover point.

To help fly itself, this aircraft uses data from satellites

as well as identifying radio signals broadcast by other aeroplanes, so it

knows who it's in the sky with.

It is also fitted with a camera that can see other air

users, even if there are not warning signals.

So right now, the pilots aren't actually flying the aircraft.

This gentleman right here is.

From that 15-year-old Dell laptop that's running Windows XP.

At this point, the aircraft is flying autonomously with a human

watching what it's doing.

Handing control of the computing over to the autopilot in the back

and once established on the route, I can hand the computing

over to Clive on the ground by the satellite.

So that little shed-like building we were in earlier with Clive

in front of the computer, he is now flying the aircraft.

Over the course of the testing of this aircraft, it's going to have

to perform a variety of different, complex tasks.

For instance, it's going to have to recognise and avoid bad weather.

Not just weather, but other aeroplanes too.

It will eventually be able to select a safe landing spot

and touchdown by itself.

Today, we can't really test its weather detection

abilities, though, as, unusually for the UK,

we have a cloudless sky.

BAE suggests that autonomous aircraft could be used

to perform dirty, dangerous or repetitive tasks.

But could this technology be introduced into

commercial air travel?

At the moment all commercial aircraft have a set number of crew.

There are programmes in existence looking at how you reduce crew

either planned from the outset or, in the case of an emergency,

you've got the autonomous system as a fallback so that you can

still have perhaps a two-crewed aircraft on a certain length

of flight but one of the crew happens to be an autonomous helper

as opposed to a real person.

We're going to end up in the Hudson.

I'm sorry, say that again.

But what happens when things go wrong?

This is the captain.

Brace for impact.

While aerospace manufacturers are exploring the possibilities

of fewer cockpit crew, what do commercial pilots think?

To find out, we paid the British Airline Pilots Association a visit.

Many decades of looking at aviation has brought us to the position

where we have come to the conclusion it's best to have two

pilots in the cockpit, because if you reduce that to one,

the problem you've got then is you've got no one to cross check

each other's decisions.

Take for example, the miracle on the Hudson.

Captain Salinger.

When the aircraft lost both its engines, the pilots had

to have a discussion and they decided their only course

of action was to land on the river.

No computer can be programmed to do that.

The flight testing of autonomous aircraft continues but the debate

about regulating them and how they are going to be used has

really only just begun.

That was Mark and this is Tim Harford, columnist

for the Financial Times and Tim, you've written a book about how

the systems that we now rely on can sometimes backfire.

What do you make of the idea of planes that might only need one

pilot or even no human pilots?

It's exciting, isn't it?

And of course, auto pilots have made planes safer

but what worries me is, what happens when

something goes wrong?

No system is perfect, including a system where

the humans fly the plane.

I guess when it goes wrong, it has two hand back

to the human in the cockpit.

Autopilot hands back to the human in the cockpit but then what?

The human is out of practice, the human is not used to flying

the plane and because the autopilot has failed, it's probably

in an unusual situation.

There was a worrying example of this a few years ago.

An Air France crash over the Atlantic Ocean.

The plane was flying quite high above a storm.

The autopilot disconnected and the pilots just weren't used

to flying at high altitude.

They were only used to operating the plane on take-off and landing

and they flew a perfectly good plane into the Atlantic Ocean

because they were confused about what was happening.

They killed everybody on board, an absolute tragedy, and this

is the paradox of automation.

The autopilots are normally so safe, so reliable,

that when they fail, the pilots find

themselves so confused.

I guess the next question is, what about autonomous cars?

We have been talking about how they will blissfully drive us around

and we won't have to take control.

I suppose for the foreseeable future, they won't be good enough

to be 100% reliable?

I guess they'll never be 100% reliable but the model

where if it's confused, it hands back to the human,

it's not going to work.

You are there with your bagel, your coffee, your newspaper.

You look up, there's a bus coming towards you, and the car goes,

autopilot disengaged, human take control, it clearly not

going to work in that way.

What makes more sense is for the human to be driving

and for the computer to be watching out for a dangerous situation,

for the computer to take over if there's a problem.

Humans get bored, get distracted, lose their skills.

None of these things happened to computers.

But I guess we are in an extended period of time before the far future

happens and computers drive and fly us.

We've got possibly decades if not a century of being in this

interim period where, if there is a problem,

we are going to end up blaming them for this really unusual,

weird crash that a human wouldn't have made.

Yes, and I think a glimpse of that is where we are asking

the computers to make a decision not about planes or cars but about,

for example, who'll get a promotion or who gets a special deal in a shop

or who gets arrested for shoplifting because the computer

recognises your face.

We are already asking computers to make this sort of decision

and the lesson of the paradox of automation is that we need to be

much more savvy about the fact that computers do make mistakes

and we need to take this seriously.

Thank you very much, Tim.

Thank you.

Don't trust the machines.

Hello and welcome to The Week In Tech.

It was the week that inventor Haiyan Zhang developed a smart

wristband to help people with Parkinson's disease

draw in a straight line.

The device's in-built motors vibrate to distract

the user and reduce tremors.

It was also the week that we discovered queueing

at the shops and using those beyond infuriating self-service

checkouts could soon be a thing of the past.

Amazon has unveiled a sci-fi store in Seattle that uses your smartphone

and advanced technologies like deep learning, computer vision and sensor

fusion to automatically detect when products are taken

off its shelves.

When you're done, you can simply trot off and then wait

for that gargantuan virtual receipt to follow.

And if you feel like you're forever stuck in traffic,

Audi has rolled out an update to make every second count.

Its new traffic light information feature tells drivers exactly how

long they'll have to wait behind a red signal before it turns

green and the length of time it will stay green.

It works by connecting directly to the city's

traffic management centre.

And finally, robotic research has reached new heights, literally.

This hopping mad bot developed at UC Berkley cannot only jump a meter off

the ground but can then again jump off objects to reach

even greater heights.

Inspired by the agility of bushbabies, researchers hope

it can one day identify jumping spots for itself.

Also in the news, the news.

Which these days isn't guaranteed to be true.

In the run-up to the US election, for example, the Speaker

of the House of Representatives did not get naked, the Pope did not

endorse Donald Trump and he did not win the popular vote,

but these stories, from websites posing as real news sites,

were heavily shared on Facebook.

Of course, it doesn't help that in 2016, the real news sounds

sounds just as made up.

But anyway, it's made events like the Trust Hack here in London

even more timely and urgent.

Here, journalists and technologists from large news organisations

are workshopping ways to help readers tell the difference

from well-researched journalism, propaganda, advertising,

satire and complete hogwash.

The thought is to provide images like icons back-up materials

that the public could see connected to a piece of news and then it

would send a signal back to the news distribution platform like Google

or Twitter so that they can identify quality news out of the fake news

and elevate it in search and social.

There are already projects afoot to try to flag up stories on sites

known to generate fake news like this plug-in

but the ideas here are not about blacklisting sites or producing

automated fact-checking systems.

Both would be massive undertakings and would themselves provoke cries

of bias and censorship.

This is more about letting news organisations prove to their readers

that they are trustworthy.

The journals of the Washington Post, I work with some amazing people.

We produce really great stuff and they remain really

committed to continuing that.

The tools that we are building here are just a way for us

to communicate that we are putting in the effort,

where our stuff is coming from, who we are talking to.

We have access to great sources.

We are trying to create something that would easily allow audiences

to verify for themselves what sources we have used.

You'd be able to click and see, who did we talk to,

and you'd click on that.

Readers want to feel like journals are being held accountable to them

and this is a way of doing that.

We are showing our work.

Other ideas here include ways to fight information bias

by surfacing articles that support the opposite side of an argument

or to check the likelihood of truth by finding similar articles

by other outlets.

Many think the reputation of the journalists themselves plays

a big part in the trustworthiness of reporting.

Italian newspaper La Stampa is suggesting a system

where an author is assigned a unique identifier that shows

their piece can be trusted.

Come on.

Best platform goes to...

La Stampa.

The best ideas won a small prize at the end of the day but it's

the taking part that counts here.

The hope is that this part-Google funded initiative may lead

to a system that helps news outlet stories rank high up on search

engines and social media.

That said, I can't help wondering if that is actually something that

platforms like Google and Facebook really want.

Do you think they care what it is they serve to us or do

you think, really, as long as you click on it, that's

what they're interested in?

How do we know the motivations of any company?

Money.

These companies make their money through people clicking

on the adverts, so do you think any of the large companies care

what it is they present to you?

Based on my conversations with them, I think they do.

The argument would be if they start being perceived as not caring

about the information that's sent out there,

that damages their reputation.

Yes, it may be that the truth will out, not because of a desire

for the facts, but because everyone, readers and news aggregators,

ultimately don't want to lose face.

More and more people are shopping online but still at this time

of year, the high street seems pretty chaotic and the retailers

want to make sure it stays that way.

So they are trying to create some more engaging experiences.

But do they help us or are they just a distraction?

Here in London's Covent Garden, 140 shops and restaurants are taking

part in creating one huge augmented reality experience.

With the help of AR app Blipper, things come to life.

It may not have created the personalised shopping experience

I dreamt up, but there were some promotional offers presented

as virtual Christmas presents almost around the tree.

A reindeer hunt and a giant reindeer you can take a selfie with,

if the mood takes you.

I have to say, it wasn't quite as cutting-edge as I'd hoped

but I suppose it's a bit of light-hearted fun.

Rather more purposefully, the signs in windows can be scanned

using image recognition, taking you to online content,

partly the sort of stuff you'd be able to look at from your sofa

but with a few extras to boot.

Then came our trip to London's Westfield where augmented

reality has had a makeover.

We've seen technology like this before but now it's actually

on the shop floor here at Charlotte Tilbury.

This is what's known as the Magic Mirror and this

is what it does.

You choose a lipstick and in real-time, you will see

your face transformed.

Bright red lips, although it doesn't seem to have any around the edges,

I think that looks all right.

I've tried the Rimmel app that does something similar on your phone

and you can buy things through it, but here you can do it in the store

with assistants all around and a whole shop of products that

you can test, smell, and after you see what your face

looks like on here, you might want to have a go to check that

you can do it yourself that well.

Meanwhile, here at this eBay event, they are taking things

a step further.

They want to get inside your mind.

The data on this screen represents what is apparently visitors

emotional responses to products.

Using what they call facial coding, the camera looks for reactions

which these guys reckon you have when you do online shopping.

Nice but don't know who I'd give it to.

Interesting.

That's ridiculous.

And I've been told to overact my reactions.

Oh, that's cute.

Whilst my results bore absolutely no correlation to what I'd liked,

maybe they were the ones I contorted my face to the most.

Maybe it would have worked better if I'd reacted more naturally,

although I struggle to imagine that my face would have shown anything.

Now, in 2015, members of the United Nations adopted a set

of sustainable development goals.

Number one on that list is to end poverty and to achieve that goal,

you first need to work out where poverty exists and how

to measure it.

We met up with some scientists at Stamford who have that

task in hand.

Marshall Burke is a professor of earth systems science

at Stanford University but he spends much of his time in Africa

gathering poverty data.

The way this is done is to elicit from the household a listing

of everything they've consumed in the last week,

the last month.

So literally everything they've consumed.

Every single thing and the value of that item and then you add up

all these items for every single person in the household.

This can take hours and hours just for one single household.

Then you have to do this for thousands of households to get

a representative sample of the area.

It's painstaking work but Burke has teamed with computer science

students to test a new method.

Using machine learning to predict poverty data

by analysing satellite images.

But to find out whether the people living in those areas are rich

or poor, the researchers used a process called transferred

learning and this image of the Earth at night.

The parts of the world that are lit up are typically the wealthier parts

of the world.

So basically we use the lower resolution night-time images to help

us figure out what in the really high resolution daytime images

we should should be using and then we use that to predict poverty

on the ground.

Between 300 and 400,000 images were used to train the algorithm.

The algorithm will figure out what's important,

what it should be looking for.

So some of the things it finds are things that

you or I would recognise, things like roads,

urban areas, farmland.

Based on those features, the algorithm can predict

what a family owns.

Things like refrigerators, cars, the sum of all those assets.

It can also be used to predict incomes.

These poverty maps show the team's findings.

In areas marked red, people spend as little

as $1.5 a day.

In green regions like Uganda's capital, Kampala, they spent

closer to $8.

We are providing a very cheap and scalable alternatives

to traditional means of data collection.

Traditionally, you have to send people out into the field

with clipboards, the surveys aren't always accurate,

there can be corruption.

Like for example lots of governments where they are underperforming,

they don't want true surveys.

All we need to make our predictions are satellite images.

But can you draw conclusions about the economic well-being

of communities in Africa when you're thousands of miles away,

sitting at a laptop in an office at Stamford?

We actually have really good survey information in a few locations.

We can use the satellite imagery to make a prediction about poverty

and then we can compare that to what the survey says was actually

going on on the ground there.

So we used a couple of the really good surveys we had to validate

our satellite input.

To be a truly useful tool though, the algorithm needs an upgrade.

We would also like to use historical imagery so maybe we can figure out

how poverty dynamics work overtime and even give us the chance

of predicting what's going to happen in the future.

If you can pinpoint poverty on a map, aid could be distributed

more evenly, policies could be more effective.

People would get help faster.

A picture may be worth a thousand words but combining that picture

with artificial intelligence could make a world of difference.

That was Sumi and that's it for this week.

You can follow us on Twitter @BBCClick for backstage fun

and photos and extra technology news throughout the week.