US Special - Part Three

Download Subtitles

Transcript

:00:00. > :00:00.stuck in traffic queues trying to get to the Port of Dover because of

:00:00. > :00:11.extra French security in Calais. Drivers are being warned to expect

:00:12. > :00:15.delays. Now on BBC News, sentence -- Click.

:00:16. > :00:19.With the AI film director, emotional clothes, and robot

:00:20. > :00:46.This is what happens when you let an artificial

:00:47. > :00:54.In a future with mass unemployment, young people

:00:55. > :01:10.Sunspring is a film whose script was written entirely

:01:11. > :01:12.by an AI chatbot, one that's been designed to mimic

:01:13. > :01:27.It's a fun watch but only, I feel, if you're in on the joke.

:01:28. > :01:32.It seems that writing a gripping, real and coherent narrative

:01:33. > :01:36.is beyond AI at the moment, but what if you were to give it

:01:37. > :01:40.the words and let it be the director?

:01:41. > :01:43.What if you gave an AI complete creative control, over

:01:44. > :01:56.That was what Saatchi Saatchi commissioned LA-based

:01:57. > :02:06.It's a real out on a limb kind of experiment for film.

:02:07. > :02:09.They decided not to make a full length feature film,

:02:10. > :02:14.but instead to embark on a more manageable project, a pop video,

:02:15. > :02:18.which is something that many new directors cut their teeth on.

:02:19. > :02:20.A French dance group gave them permission to interpret

:02:21. > :02:25.The lyrics would form the script but the interpretation would be

:02:26. > :02:32.completely down to the machine, or more accurately, the machines.

:02:33. > :02:34.A director's job involves a lot of different disciplines

:02:35. > :02:40.So given that there is no one big artificial brain somewhere

:02:41. > :02:43.that can do all that, we assembled the best

:02:44. > :02:47.and most progressive pieces of artificial intelligence,

:02:48. > :02:52.kind of in a collective, to do that.

:02:53. > :02:58.First up, IBM's Watson was used to analyse the lyrics and make

:02:59. > :03:01.an educated guess at the emotions of each line.

:03:02. > :03:05.Based on these, the humans could ask the next AI questions about

:03:06. > :03:12.For that, take to the director's chair, Ms Rinna.

:03:13. > :03:15.We came up with a series of questions that you might ask

:03:16. > :03:17.somebody like me when you are prepping for a job,

:03:18. > :03:33.I could imagine a creative type coming up with that.

:03:34. > :03:36.Ms Rinna is a Japanese Twitter chatbot that appears to be

:03:37. > :03:42.And it was asked a whole load of questions on who the characters

:03:43. > :03:44.in the lyrics were, what actions they should perform and what outfits

:03:45. > :03:49.Do you have any kind of understanding about how it came

:03:50. > :03:53.up with those answers, day and night, night

:03:54. > :03:58.From what we understand, this particular AI has been

:03:59. > :04:03.programmed by having conversations with fans.

:04:04. > :04:05.It is a Twitter handle, it has millions of conversations

:04:06. > :04:11.It can relate but it is relating in a way that a 17-year-old

:04:12. > :04:13.girl on Twitter would, I guess.

:04:14. > :04:21.So how does a director choose the right actor for a role?

:04:22. > :04:23.Well, it is probably someone who feels emotionally connected

:04:24. > :04:27.to the work and then can transmit those emotions to the audience but

:04:28. > :04:34.And that is something that computers don't have, but what computers can

:04:35. > :04:42.do is measure the performer's brain activity using an EEG like this.

:04:43. > :04:46.In a very unusual casting session, ten unsuspecting actors

:04:47. > :04:49.were asked to perform the song while their facial expressions

:04:50. > :04:53.and brainwaves were compared to those of the song's lead singer

:04:54. > :05:00.The guy with the best emotional match got the gig.

:05:01. > :05:02.And when we asked them to do that, they said,

:05:03. > :05:06.And we said that the director needs that information

:05:07. > :05:09.in order to make a choice, so that whoever he picks has

:05:10. > :05:13.Did the actor know at this point that

:05:14. > :05:19.When did the actor find out the director was an AI?

:05:20. > :05:23.Does the actor know currently that the director is AI?

:05:24. > :05:28.In fact when we cast him, we brought him down to the set

:05:29. > :05:33.We said, it's going to be a little weird because we are going to be

:05:34. > :05:35.giving you direction based on the artificial intelligence's

:05:36. > :05:37.understanding of what this song is and what your actions

:05:38. > :05:42.And unfortunately, I'm just an assistant director in this

:05:43. > :05:45.scenario and I'm not really allowed to do anything other than to execute

:05:46. > :05:50.the vision of the artificial intelligence, or help it along.

:05:51. > :05:53.If it seems weird, it is weird because my nature is to tell

:05:54. > :05:57.I have to go, excuse me, what do you think?

:05:58. > :06:01.On the day of the shoot, even the camera drones followed

:06:02. > :06:04.flight paths that were based on Watson's emotional

:06:05. > :06:09.And when it came to the edit, choosing the shots and cutting it

:06:10. > :06:14.all together, Zoic built a brand new editor AI.

:06:15. > :06:17.This started by creating random assortments of shots and then

:06:18. > :06:20.learned from feedback from the humans which types

:06:21. > :06:26.After many iterations, the finished video was given

:06:27. > :06:30.a visual treatment that was itself a type of mural art.

:06:31. > :06:35.Adding a day and night style to the two characters' scenes.

:06:36. > :06:46.The result is a pop video that is not going to win any awards.

:06:47. > :06:50.In fact, the band, whose videos are usually very highly produced,

:06:51. > :06:53.have distanced themselves from the whole thing.

:06:54. > :06:56.And I'm not really convinced that an AI has actually

:06:57. > :07:03.I mean, what they've done is got a load of different AIs to do

:07:04. > :07:08.Including asking a chatbot what this fellow should wear.

:07:09. > :07:13.But then the people at Zoic aren't really saying otherwise.

:07:14. > :07:19.The exercise we went through provided a lot

:07:20. > :07:24.The output itself of the video, it could be better,

:07:25. > :07:31.But I think for a first attempt, a worthy effort.

:07:32. > :07:34.As the director behind the scenes, I'm like, wow, that is super

:07:35. > :07:37.intimidating because it won't be long before this technology actually

:07:38. > :07:44.evolves where that editor that we created becomes

:07:45. > :07:47.absolutely able to interpret, like, that shot is running and it

:07:48. > :07:49.has emotion and it sweeps, I should definitely use that.

:07:50. > :07:53.And if that cuts to the close-up of this emotional expression,

:07:54. > :07:56.you will feel and it will understand how one influences the other.

:07:57. > :08:02.And all of a sudden we are going to get really emotional cutting

:08:03. > :08:13.And if you might permit just a slight indulgence on my part,

:08:14. > :08:16.the eagle-eyed amongst you may have noticed that Loni was

:08:17. > :08:25.It's actually a door from the ship in Joss Weedon's really rather

:08:26. > :08:28.excellent and criminally cancelled before its time TV show,

:08:29. > :08:34.Loni worked as the series' visual effects supervisor

:08:35. > :08:38.and if you would like to find out why he has the door, as well as how

:08:39. > :08:40.he helped design the ship, head to our YouTube and Facebook

:08:41. > :08:43.pages for the geekiest chat that two grown men can possibly have

:08:44. > :08:57.Time to change gear here in California, a state which is not

:08:58. > :09:03.It also has a long history of innovative aviation

:09:04. > :09:09.and aeronautics companies testing aircraft over its wide open spaces.

:09:10. > :09:13.But there is one outfit here that has worked on projects so secret

:09:14. > :09:17.that the US government has often denied its existence.

:09:18. > :09:27.But Marc Cieslak found it and got exclusive access to the Skunk Works.

:09:28. > :09:33.A Cold War spy plane capable of flying

:09:34. > :09:49.It evolved from a plane commissioned by the CIA in the late '50s,

:09:50. > :09:51.designed at Lockheed Martin Skunk Works.

:09:52. > :09:53.More used to creating secret aircraft, Skunk Works has

:09:54. > :09:56.turned its talents to creating something much slower.

:09:57. > :10:02.We have to be careful about what we film here

:10:03. > :10:07.at Skunk Works because they perform lots of work on classified projects.

:10:08. > :10:09.However, Lockheed Martin have allowed us to film

:10:10. > :10:13.their one third scale model of the airship.

:10:14. > :10:16.It is the biggest scale model I have ever seen although technically this

:10:17. > :10:22.This is a hybrid airship designed for lifting cargo.

:10:23. > :10:27.It has three inflatable hulls, with helium providing 80%

:10:28. > :10:34.It has thrusters which can change direction allowing it to take off

:10:35. > :10:37.They also allow the airship to fly much more like

:10:38. > :10:43.It does away with landing gear, making use of hovercraft technology

:10:44. > :10:47.to get it on the ground or onto water.

:10:48. > :10:49.This demonstrator is actually three times smaller

:10:50. > :10:57.So airships have been around for a long time.

:10:58. > :10:59.Why is it taking so long for governments, for commercial

:11:00. > :11:03.operators to get on board with this sort of aircraft?

:11:04. > :11:08.At the end of the day, for us it is hard to change.

:11:09. > :11:12.And so when you see something like this, so different,

:11:13. > :11:15.and aircraft are expensive, so this is a fairly big risk,

:11:16. > :11:18.but the biggest aspect we bring in is remote cargo.

:11:19. > :11:23.Areas that do not have a road or a rail line or any

:11:24. > :11:28.We would like to do that in a sustainable way so we are not

:11:29. > :11:30.hurting the environment, or at least as little as possible,

:11:31. > :11:33.and without a road or rail line to get to those sites,

:11:34. > :11:36.With the airships, we can take that part away.

:11:37. > :11:38.Lockheed is not the only aerospace outfit working

:11:39. > :11:44.Rival cargo carriers are being developed in the US and UK.

:11:45. > :11:48.However, the boffins at Skunk Works have developed a novel way to keep

:11:49. > :11:54.In order to see how they do that, I've got to get inside

:11:55. > :12:03.So this is how you enter the inside of the airship.

:12:04. > :12:06.You've got to try to make sure you don't lose any of the air

:12:07. > :12:15.As airships are inflatable, they are susceptible to damage

:12:16. > :12:17.which could create holes in its skin.

:12:18. > :12:23.Finding and patching those holes is an important task.

:12:24. > :12:26.Ever wondered what it's like to be inside a bouncy castle?

:12:27. > :12:36.Somewhere in here there should be a spider.

:12:37. > :12:42.Or to be more precise, it is one half of the spider

:12:43. > :12:45.because the other half is on the other side

:12:46. > :12:53.On the other side, there is a light that is shining upwards.

:12:54. > :13:00.Because there will be light shining through the skin of the airship.

:13:01. > :13:08.If it finds a hole, its job is then to patch it up.

:13:09. > :13:11.In fact just under here we can see a patch that has

:13:12. > :13:18.We had the opportunity to go inside the airship.

:13:19. > :13:22.Normally you would not be able to go inside it, a human being would not

:13:23. > :13:26.That's right, it's full of helium so it's hard to breathe in there.

:13:27. > :13:29.They wear oxygen masks to do that kind of thing.

:13:30. > :13:32.Also, it's very hard to plug a hole on the top of the airship

:13:33. > :13:35.from the inside of the airship, there's no way to get scaffolding up

:13:36. > :13:39.Patching it up with a robot, that allows you to patch those holes

:13:40. > :13:43.Lockheed aims to have their hybrid airships operational

:13:44. > :13:57.At which point, swarms of spiders will perform robotic repairs.

:13:58. > :14:02.It was the week that Tinder announced its group function,

:14:03. > :14:05.meaning that the app could have a long-term relationship

:14:06. > :14:08.with you even after you have found love.

:14:09. > :14:11.Tinder Social hopes to help you plan a night out with a chance

:14:12. > :14:15.to make friends or meet up with groups nearby,

:14:16. > :14:19.making connections that are not just about hooking up.

:14:20. > :14:24.Mercedes put its self-driving bus to the test in Amsterdam.

:14:25. > :14:27.The Future Bus successfully took a 20 kilometre trip without the need

:14:28. > :14:33.for a driver, leaving him able to nod knowingly at passengers.

:14:34. > :14:37.As well as camera and radar systems, it recognises traffic lights,

:14:38. > :14:42.The Museum of London is using the video game, Minecraft,

:14:43. > :14:52.350 years on, Great Fire 1666 will allow players to walk down

:14:53. > :14:55.the streets of London, interact with people and combat

:14:56. > :15:01.the plague, whilst also searching for audio clips to tell the story.

:15:02. > :15:04.And finally, children at a Connecticut school have joined

:15:05. > :15:09.forces with a local aquarium to help an injured penguin.

:15:10. > :15:13.Five years after tearing her flexor tendon, she got lucky

:15:14. > :15:16.when the school purchased a 3D printer.

:15:17. > :15:19.They have created her a lightweight book to help her p-p-pick

:15:20. > :15:33.Now back in January, I visited Pier 39 in San Francisco,

:15:34. > :15:37.home to Autodesk's creative workshop, where artists

:15:38. > :15:42.and researchers are let loose on CAD software and 3D printers to come up

:15:43. > :15:46.Well, now it is time to meet one of those artists over

:15:47. > :15:52.at the University of Southern California in Los Angeles.

:15:53. > :15:55.I have trained as an architect and currently I am doing

:15:56. > :15:58.my PhD in media art, and exploring the relationship

:15:59. > :16:03.of human bodies with the space around.

:16:04. > :16:05.I am basically exploring this relationship by using interactive

:16:06. > :16:12.One of them is our near environment, which basically means our

:16:13. > :16:18.And the other, I am exploring that relationship with the space around,

:16:19. > :16:22.I am exploring how architecture can respond to humans,

:16:23. > :16:31.Curse Of The Gaze is an interactive 3D printed outfit.

:16:32. > :16:34.It is basically exploring how our fashion can be

:16:35. > :16:38.I really wanted to create something that becomes

:16:39. > :16:46.It is equipped with a camera, and the camera is located

:16:47. > :16:52.The camera can see your age, gender and where you were looking,

:16:53. > :16:55.and based on what you are looking at, it moves and

:16:56. > :16:59.So if someone is here, they are looking at the cape

:17:00. > :17:08.It is an extension of your brain to your garment, basically.

:17:09. > :17:13.The garment thinks it is real and perceives the world,

:17:14. > :17:16.almost to the most fundamental aspects of social interaction,

:17:17. > :17:23.And by looking at each other, we communicate meanings,

:17:24. > :17:25.and sometimes our gaze can be intrusive.

:17:26. > :17:28.Sometimes our gaze can be reassuring.

:17:29. > :17:30.In other words, if you can feel someone's gaze on your body,

:17:31. > :17:36.your perception of the world is going to be different.

:17:37. > :17:40.Synapse is a 3D-printed interactive helmet that moves

:17:41. > :17:48.When it goes higher, your attention goes higher,

:17:49. > :17:51.and the helmet opens up so you can actually pay attention

:17:52. > :17:53.to the world around you, and as your attention level gets

:17:54. > :17:57.less, basically it creates a cocoon around your head.

:17:58. > :18:04.The brain sensor is basically capturing information,

:18:05. > :18:08.gathering it on various different brain frequencies.

:18:09. > :18:12.And then it sends the information related to your attention level

:18:13. > :18:16.to a microcontroller, which is again embedded inside.

:18:17. > :18:22.And that information can control two small servos located

:18:23. > :18:30.Basically you map your attention level to control servos

:18:31. > :18:41.Aurora is an interactive ceiling application which can see body

:18:42. > :18:46.movement underneath it and respond accordingly.

:18:47. > :18:51.Particularly it uses the ideas on body interaction, in other words

:18:52. > :18:57.how our bodily movement, not just the tips of our fingers

:18:58. > :19:00.on the mobile phone, can inform interaction,

:19:01. > :19:03.but how our movements within the space can inform

:19:04. > :19:12.The ceiling has a computer vision which is basically using a Kinect

:19:13. > :19:20.It sends that information to a series of microcontrollers

:19:21. > :19:23.in which they are controlling 15 different motors.

:19:24. > :19:29.And those motors can inform the motion of each of the notes

:19:30. > :19:47.Like most American cities, LA is a driving city.

:19:48. > :19:51.It can take ages to get across town because of, well,

:19:52. > :19:57.So if you are a parent who has to take their kids somewhere for two

:19:58. > :20:00.or three hours before picking them up again,

:20:01. > :20:02.it might not actually be worth your while coming

:20:03. > :20:08.So what do you do if you are a parent who is too busy to ferry

:20:09. > :20:13.Kate Russell has taken a hop, skip and a drive

:20:14. > :20:25.Lots of homework, hectic after school activity schedules,

:20:26. > :20:29.being ferried around by your parents.

:20:30. > :20:32.I think it is time for you to get ready to go.

:20:33. > :20:34.Hang on a minute, that actually looks pretty good,

:20:35. > :20:40.Juggling work commitments with school runs and extracurricular

:20:41. > :20:46.For some parents, a taxi is the only option, but how do

:20:47. > :20:50.you know your children are safe getting into a car with a stranger?

:20:51. > :20:56.That is the question HopSkipDrive set out to answer.

:20:57. > :20:59.Three of us created the company and we are all working mums.

:21:00. > :21:02.We have eight children between us who go to six different schools

:21:03. > :21:05.and participate in every kid activity known to man.

:21:06. > :21:14.Carly loves to dance and she is pretty good at it, too.

:21:15. > :21:25.We originally lived somewhere different, we lived really close

:21:26. > :21:27.to the freeway and we had good access.

:21:28. > :21:30.And then we moved here and we could not get to class

:21:31. > :21:32.on time, coming to work and then back down.

:21:33. > :21:35.So we put her in a local dance studio which wasn't the same.

:21:36. > :21:37.HopSkipDrive says it runs background checks on all its drivers

:21:38. > :21:42.They also told us that applicants must have at least five

:21:43. > :21:46.Compared to other rideshare services,

:21:47. > :21:54.A minimum fare of around $15 depending on where you are based,

:21:55. > :21:56.and looking at fare calculation comparisons with Uber,

:21:57. > :21:58.it seems to be about 30-50% more expensive over the distance,

:21:59. > :22:02.although there are no surcharges during peak hours with HopSkipDrive.

:22:03. > :22:06.But the founders think what they do is different from a taxi.

:22:07. > :22:12.So for my son Jackson, his care driver will park,

:22:13. > :22:18.go into the school, sign him out, there is a booster seat for him

:22:19. > :22:20.because he's not eight, and I can track the ride

:22:21. > :22:24.Rides are scheduled ahead of time with clearly branded cars

:22:25. > :22:27.and drivers who have a secret password to give to the children

:22:28. > :22:32.on pick-up so everyone can be sure they are getting in the right car.

:22:33. > :22:34.What happens when they come and pick you up?

:22:35. > :22:37.First, I get a text message on my phone and then

:22:38. > :22:40.when they come up to the door, they tell me the password

:22:41. > :22:44.and they go in the car and we have a conversation.

:22:45. > :22:47.For the drivers, many of them mothers themselves,

:22:48. > :22:51.it is a much more rewarding job than just driving a taxi.

:22:52. > :22:55.How is it driving lots of different kids?

:22:56. > :23:05.My main objective is just to obviously be friendly,

:23:06. > :23:08.and some kids just get in and they have their headphones

:23:09. > :23:11.on and they have their music playing, you know, and they just

:23:12. > :23:17.need to chill out because they have just come from school or dance,

:23:18. > :23:21.and other times, you know, some of the kids are really chatty,

:23:22. > :23:25.so I just have to take the emotional temperatures.

:23:26. > :23:29.Now Carly can go to the dance studio she loves, and Jennifer can manage

:23:30. > :23:35.I cannot leave work early every single day, so that is good.

:23:36. > :23:41.So far limited to a few areas in California,

:23:42. > :23:43.HopSkipDrive plans a careful expansion into more

:23:44. > :23:49.They have also recently added carpooling to reduce the cost,

:23:50. > :23:52.and this should help to ease congestion around school

:23:53. > :24:02.That was Kate and that is it from LA for the moment.

:24:03. > :24:05.I have a feeling we will be popping back here every so often

:24:06. > :24:09.Meanwhile, on Twitter we live @bbcclick, online we are:

:24:10. > :24:45.The fortunes weather wise across the UK very from west to east as we look