- há 13 minutos
With unique access in Ukraine, in Israel and with the US army, this documentary charts the new technological arms race, from AI to drones. Will warfare become less destructive or even more brutal?
Categoria
🗞
NotíciasTranscrição
00:00.
00:02Feb. 27, 2026.
00:05Donald Trump's war room.
00:08At 3.38 Eastern time,
00:10the president orders
00:11a strike on Iran.
00:14As dawn crept up
00:16across the Central Command AOR,
00:18the sky surged to life.
00:21More than
00:22100 aircraft launched from land,
00:24sea, fighters,
00:26tankers, airborne early warning,
00:28electronic attack.
00:30Bombers from the states and unmanned
00:31platforms, forming
00:33a single, synchronized
00:35wave.
00:38The attackers
00:39send back positions, images
00:41and radio signals.
00:47Which are
00:48instantly processed and analyzed
00:49by artificial intelligence.
00:54Within hours,
00:56Iran's supreme leader,
00:57Ayatollah Aliya Khamenei
00:59is killed by an Israeli strike.
01:04The United States and its partners
01:07have launched Operation Epic Fury,
01:10one of the largest, most complex,
01:13most overwhelming military offensives
01:16the world has ever seen.
01:18Nobody's seen anything like it.
01:21There's never been a military
01:23like we possess,
01:25and frankly,
01:26there's nobody even close,
01:28but we are now using that military
01:30for good.
01:32The strike against Khamenei
01:34was successful because the Israelis
01:36knew exactly where to find him.
01:38In the preceding months,
01:40artificial intelligence had also
01:42been deployed to analyze vast
01:44troves of satellite signals
01:46and human intelligence.
01:49the strikes against Iran
01:50and before that Venezuela
01:51are just the latest
01:53and most vivid demonstration
01:55of how we're in the midst
01:57of a once-in-a-generation transformation
02:00of the technology of war.
02:02It is an inflection point.
02:04It's going to change our processes.
02:05It's going to change the way we fight,
02:06our operating systems.
02:07So you're watching all that play out.
02:10And whoever figures it out first
02:12will have an advantage.
02:14This is about winning, right?
02:16Winning.
02:17We can't forget our job is to win.
02:20The transformation has been
02:22building for decades
02:23as computers became more powerful,
02:26giving rise to AI.
02:29And more compact,
02:30allowing for swarms of lethal drones.
02:35This is a film about the development
02:38of these new war machines
02:39and how they're already being used
02:42by Israel, Ukraine
02:44and by the United States.
02:47The most important thing
02:48is to ensure that we in the West,
02:51broadly defined,
02:52have the most powerful versions
02:54of this technology
02:55and continue to do so.
02:57Will this new technology
02:59make war more precise?
03:02reducing collateral damage
03:03and even preventing conflict
03:06altogether by deterrence?
03:08Or might it simply
03:10make wars cheaper,
03:11more frequent
03:12and ever more brutal?
03:15When you rely on AI,
03:16you need less people to ground you.
03:18And people can tell themselves
03:20a beautiful story
03:22on the most precise war ever.
03:25And they don't have to speak
03:26to anyone that will face them
03:27with the fact that
03:28they're just killing families
03:30and with no actual solution
03:32in the near future.
03:51Bavaria, southern Germany.
03:567,000 US and NATO troops
03:59are preparing to act
04:00in case Russia invades
04:02a NATO country.
04:06They're racing to learn
04:07how to use this new technology
04:09on the battlefield.
04:11One, two, nine, nine, zero.
04:13One, two, nine, nine, zero.
04:23This is the headquarters
04:24of an army formation
04:26of about 1,000 soldiers.
04:30Sir, there is no change
04:31to combat power, location
04:34or arraignment of our forces.
04:36There it is.
04:36All right, cool.
04:37Let's have a lock-type plan
04:38to get all our replacements
04:40and our vehicles
04:41out to the front.
04:4310 miles away,
04:45more NATO troops
04:46acting as an advancing enemy.
04:50The big picture
04:51of what we're doing here
04:52is we're exercising
04:53our brigades
04:54and our battalion's ability
04:55to do a combined arms fight
04:57against an enemy
04:58that has near or peer threats
05:00against us.
05:02We're starting to see
05:03the indicators and warnings
05:04that the enemy is about
05:05to try to breach,
05:06break through our lines
05:08using assault maneuver fires.
05:20You are experiencing gas
05:22non-persistent
05:23for four-hour duration.
05:27We have a chemical attack,
05:28so I mean, hey, uh, tape.
05:30I need tape.
05:30I need tape.
05:31If you look at any
05:32of the exercises,
05:34any of the experimentation
05:35that we're doing,
05:36clearly it's leading
05:37to make sure that we have
05:37the right capability,
05:39but in particular
05:40along the Eastern flag.
05:47This is the command center
05:48for NATO and the US Army.
05:51Any land war against Russia
05:53will be orchestrated from here.
05:56The commanding general
05:57is Chris Donoghue.
05:59He led much
06:00of the technological innovation
06:02in the US Army
06:03and is now in charge
06:04of all US soldiers in Europe.
06:07It's our job to make sure
06:09that Europe
06:09and everyone
06:11that's part of Europe
06:12remains safe,
06:14remains secure
06:15and that we live up
06:16to what NATO
06:17expects us all to do.
06:19Donoghue was an early adopter
06:21of the idea
06:22that wars can be won
06:23or lost
06:24not only because
06:25of the strength
06:25of the armies,
06:26but also because
06:27of how much information
06:29can be gathered
06:30and analyzed.
06:31If you can truly
06:32harness the right data,
06:34have the right processes,
06:35you can really come up
06:37with a distinct advantage.
06:39And what is that advantage?
06:40You have to out-think,
06:42out-decide,
06:43out-act,
06:44and then do that
06:45multiple times.
06:49What's innovative
06:50is that this command center
06:51has access to all information
06:53from the battlefield
06:54through a single network
06:56of computers
06:57called MAVEN
06:58that supports all U.S.
07:00and NATO operations.
07:03On MSS,
07:04put something on there,
07:05keep people out of
07:06Hog Hill.
07:08These screens deliver the picture
07:10through
07:11MAVEN Smart Systems,
07:13which takes in multiple
07:14different streams of data
07:15that operators
07:17are looking at
07:18and analyzing
07:19and giving me
07:20the recommendations
07:21or their understanding
07:23of what they've seen
07:24on the battlefield.
07:26I think the first thing
07:28with MAVEN Smart Systems
07:29is it gives us the ability
07:31to take classified
07:32of all type,
07:34unclassified,
07:35and then commercial data,
07:36and you can aggregate it
07:38all together
07:38to help you make
07:39all the decisions
07:40that you have to in warfare.
07:45Underpinning MAVEN
07:46is artificial intelligence.
07:49MAVEN Smart Systems
07:50is a battle command
07:51system tool.
07:53We're starting now
07:54to use a lot
07:54of different machine learning
07:55or computer vision models
07:57to help,
07:58especially with imagery
07:59or videos.
08:01For example,
08:03you can have
08:04a computer vision model
08:05that's looking
08:05specifically for tanks.
08:08To do that,
08:09you have to take
08:09thousands of images
08:10of tanks,
08:11and you train the model
08:12to look for those.
08:14So what you do
08:15is you take
08:15a picture of a tank
08:17and you draw a box
08:18around the tank
08:19and you tell the computer,
08:20this is a tank,
08:21and you do that
08:21a thousand times.
08:24Now, when you feed
08:26an image to that model,
08:27it's going to tell you,
08:29yes, this is a tank,
08:30or no, it's not.
08:31Once the computer
08:32has flagged an image
08:33where it suspects,
08:35it's detected a tank,
08:37for instance,
08:37we don't just
08:39action that immediately,
08:40right?
08:40That's when we would
08:40take it to our human analysts
08:42and say, hey, can you confirm
08:43or deny that this
08:44was properly classified?
08:45What the AI is doing,
08:47it's not replacing
08:48the human,
08:49but it's enabling them
08:50to do their job faster
08:51so we can process
08:52more and more data
08:53and identify targets
08:55quicker and solve
08:56problems faster.
08:57Five, five, nine, zero, seven.
09:00It's a BMP stationary over.
09:02AI is helping glean
09:04what's important,
09:05what's useful,
09:06what's not useful,
09:07so I can say friendly
09:09or foe,
09:10or target or not.
09:12The enemy has
09:13the same capability,
09:14so I have to be quicker,
09:16I have to be faster,
09:17otherwise,
09:17I'm at risk
09:18for becoming that target.
09:25The revolution
09:26in how wars are forward
09:27has come about
09:28not only with the processing
09:29of data,
09:31but also with small
09:32unmanned vehicles,
09:34drones.
09:41The drone
09:42descend up,
09:43this is your eyes
09:44in the sky,
09:44to detect everybody
09:45around you,
09:46early warning,
09:47to see what's going on
09:48on the battlefield.
09:50And in addition,
09:52FPV,
09:53first person view drones,
09:55can carry bombs.
09:57The FPV drones,
09:59that's your strike drone,
10:00that's what's going
10:00to take everything out.
10:01on the battlefield.
10:02Okay, I think they're
10:02about to launch.
10:10Two BMPs, like 200 meters
10:12for a month.
10:13You can hear them driving.
10:14The battlefield makes me
10:16move quicker,
10:17and understand quicker,
10:18and make decisions quicker.
10:22I hit the rear BMP.
10:25Add one to the scoreboard.
10:26The enemy is currently moving
10:28about a click
10:30every two minutes.
10:32Over.
10:33I just spotted a vehicle
10:34in the wood line.
10:36What's the grid?
10:37Zero one.
10:37What you see is now
10:39a completely digitally
10:40data-driven unit
10:42that had all these
10:44new forms of mass,
10:45drones,
10:46unmanned systems,
10:47in there,
10:48to make them as able
10:49to find the enemy,
10:51hide from the enemy,
10:52see the enemy,
10:53and kill the enemy.
10:54Artillery.
11:01This year, Sergeant Cole
11:03has noticed that the
11:04Ukrainians are also
11:06putting AI into drones,
11:08so they can fly
11:09themselves to targets.
11:12It's just absolutely
11:13insane what they're doing.
11:15The targeting software,
11:16it'll detect, like,
11:18that's a person,
11:18that's a truck.
11:19The hard part is to
11:21program it to actually
11:22fly on its own
11:23and then hit something.
11:24You know,
11:24that comes in,
11:25with, like,
11:27morals and ethics.
11:28Like, do we want
11:28something to decide
11:30on its own
11:31to kill something?
11:33What's your view of that?
11:35I mean,
11:39if it flies completely
11:41autonomously
11:41and kills stuff,
11:43I don't necessarily
11:44agree with it.
11:47Completely autonomous
11:48drones flying around
11:49killing people,
11:50that's just insanity.
11:51that stuff out of
11:53nightmares.
11:55I've got to come land.
11:58I've got to go ahead.
11:59It's anyone's guess
12:00how these technologies
12:02play out in future wars.
12:04But clues already exist
12:06in Israel and Ukraine
12:08where the technology
12:09is already in use.
12:17The trajectory of this
12:19new war machine
12:20begins in about 2016,
12:22when the technology of war
12:24was limited mostly
12:25to hardware,
12:27like aircraft carriers,
12:29artillery,
12:30and tanks.
12:33But already there were
12:35experiments in data
12:36processing.
12:38Especially in Israel,
12:40a nation of the
12:41that promoted itself
12:42as a tech superpower.
12:45Today,
12:46our biggest export
12:47is technology.
12:49Israeli technology
12:50which powers the world's
12:51computers,
12:52cell phones,
12:53cars,
12:53and so much more.
12:56The future belongs
12:57to those who innovate.
12:59And this is why
13:00the future belongs
13:01to countries like Israel.
13:06Nearly 50 years
13:08after Israel's occupation
13:09of the West Bank and Gaza.
13:12Some Palestinians,
13:13hemmed in by the
13:14newly built separation
13:15wall,
13:16turned to lone wolf attacks.
13:20A short while ago
13:21here at the central bus
13:22station in Jerusalem,
13:24one terrorist
13:24carried out an attack
13:25against a security guard
13:26that was stabbed.
13:28I've trapped outside
13:28the entrance.
13:29At the moment,
13:29the area is still closed off.
13:31We've confirmed it was
13:32a terrorist attack.
13:34These attacks were different
13:36to the previous waves
13:37as they weren't organised
13:39by a militia organisation.
13:42Instead,
13:43single Palestinians
13:44skirted round
13:45the security controls.
13:49You just had lone individuals,
13:52usually very young.
13:53These people usually
13:54just took a knife
13:54or even a screwdriver
13:56to attack soldiers
13:57in checkpoints
13:58or just suddenly
14:00batting people
14:01on the streets
14:02or driving cars
14:03into people.
14:05The challenge was
14:07to try to figure out
14:08who's the random person
14:09that will wake up today,
14:11pick up a knife
14:12and stab someone.
14:16These witnesses
14:17are intelligence analysts
14:19who were called up
14:20to work for the Israeli army.
14:23They don't want to reveal
14:24their identities
14:25and their voices
14:27have been digitally altered.
14:29Their job
14:30was to monitor
14:31security threats
14:32in the West Bank.
14:34I was recruited
14:35to the army
14:35in the late 2010s.
14:38In the theoretical framework
14:40that I worked under,
14:42every Palestinian
14:43is a suspect.
14:46When I joined the army,
14:47it was in the midst
14:49of a technological transformation
14:51of trying to adopt
14:54new technologies.
14:57Israel had a special advantage
14:59in tracking the threat.
15:01It controlled
15:02Palestinian communications.
15:05Israel has basically
15:07unlimited data
15:09because all stellar data
15:12of the West Bank
15:13goes through Israeli
15:15technological centers.
15:18The whole network
15:19is basically
15:20in Israel's hands.
15:22All phone calls
15:23that are being made
15:25through the regular network
15:26in the West Bank
15:27and Gaza,
15:28they would all go
15:29to a big database,
15:32sort of archive.
15:38These low north attacks,
15:40they naturally encourage
15:41the army to think
15:42in terms of big data.
15:44the traditional approach
15:46of just, you know,
15:47trying to infiltrate
15:48a terrorist organization.
15:50This wouldn't really work
15:51for people
15:52to just spontaneously decide
15:53to attack Israelis.
15:57At first,
15:58Israel tried simple word search
16:00to catch lone wolf militants
16:02before they attacked.
16:04We were really doing
16:06the first steps
16:07of, for example,
16:08working with telephone calls
16:10on how to identify
16:12a relevant telephone call
16:13according to keyboards.
16:17They kind of figured out
16:19that many people
16:20who go on these
16:20long wolf things
16:21send text message, basically.
16:24Beforehand,
16:25some kind of declaration
16:27of intent.
16:29Certain words,
16:30whenever there are
16:31reporters in the system,
16:33these words would
16:34jigger,
16:34some sort of alarm.
16:38But soon,
16:39they experimented
16:40with data patterns
16:41that would predict
16:42who might merely
16:44be tempted
16:44to become a radical.
16:46We want to create
16:47a usable data set
16:49of possible terrorists.
16:50So,
16:51the general method
16:53to do it
16:54is specify
16:55as many attributes
16:56as we can.
16:58So, age,
17:00gender,
17:01and try to find patterns
17:03in the data
17:04that can be used
17:06to squeeze it.
17:08Thousands of individual
17:10Palestinians were categorized
17:12by dozens of attributes,
17:14and every attribute
17:15was given a value.
17:17A computer then calculated
17:19how likely
17:20each person
17:21was to turn violent.
17:24The initial algorithms
17:26created grades
17:27as to predictability
17:30to be terrorists
17:31for one to ten.
17:35You're a 7.8 terrorist,
17:37and here's the group
17:38of potential terrorists.
17:40Now let's surveil them
17:42on a daily basis
17:43to see which of them
17:45is actually planning
17:46to do terror.
17:51At some point,
17:52these attacks
17:53are reduced
17:53significantly.
17:57It was pretty efficient.
17:59I remember that
18:01I was almost shocked
18:02to see how efficient
18:03it was.
18:04They did feeling
18:05quite emboldened
18:06by this.
18:08But in the last year
18:10we managed
18:11to reach
18:12from 400
18:13attempts.
18:14to be able
18:14to a very
18:14strong
18:14It is a military
18:17act.
18:18It's a military act
18:19that is made
18:20on more light
18:21and more light
18:22and more light
18:22than more light
18:22and more light
18:23than more light
18:24and more light
18:26things.
18:27Although Israel's
18:29military leadership
18:30was convinced
18:31that the decline
18:32in attacks
18:33was due
18:33to prediction technology,
18:35human rights groups
18:37reckoned
18:37the waves
18:38of violence
18:39had always ebbed
18:40and flowed.
18:42E eles acusam Israel
18:44de um invasão
18:46de privacidade
18:47de milhões de palestinos
18:51Para mim, é muito claro
18:53que isso total control
18:55é precisamente
18:56o que permite
18:58o processo político
18:59para ser frozen
19:00Clearly, se nós temos
19:02essa total control
19:03não há no motivo
19:05para mudar nada
19:06Nós realmente pensamos
19:08que a tecnologia
19:09resolve a maioria das nossas problemas
19:11Enquanto Israel
19:13concentrou mais
19:14de seus recursos
19:15de inteligência
19:15em tecnologia
19:17eles negaram
19:18a papel de raiva
19:19de suas raízes
19:20particularmente
19:21na Strip de Gaza
19:24Isso foi uma parte
19:26de uma parte
19:26de inteligência
19:28de no passado
19:30e seria um erro
19:33em que a causa
19:34de a Hamas
19:35de 7 de outubro
19:36de 7 de outubro
19:48O próximo capítulo da história da tecnologia de guerra
19:52começou quando os russos de trânsos se derrotaram para a Ukraine em fevereiro de 2022.
20:11Os russos de trânsos se derrotaram para a Ukraine, que tinha 250 mil soldados,
20:34mas se derrotou contra um armão de cerca de um milion.
21:04A guerra pode ser em diplomática, paralelo com os steps de nosso armão.
21:14O desafio que Zelensky teve em lutar de volta, foi trabalhar em que os russos de trânsos se derrotar.
21:24De um centro de comandos na Alemanha, NATO ofereceu assistência na forma de maven smart systems,
21:30embora os militares não falam sobre o que o apoio foi dado.
21:36Mas a tecnologia de maven smart systems foi criada por uma empresa controversa chamada Palantir.
21:44O seu CEO foi e é um maverick skia chamado Alex Karp.
21:50Aqui ele está, falando com os investidores na década da guerra da Ukraine.
21:57Palantir, por exemplo, é uma empresa que vive em bons momentos e nós vivemos em bons momentos.
22:07Nós vamos continuar a distribuir o mundo mais importante produtos
22:10para os mais interessantes, criatóis e infelizmente pessoas do mundo.
22:13Palantir's produtos são na absoluta fronteira.
22:16E você vê eles em todos os dias.
22:19Palantir's early investor foi a CIA.
22:21E eles têm contratação com organizações de segurança global,
22:26incluindo U.S. imigração.
22:28O seu próximo cliente foi Zelensky.
22:31O convite veio dos Ukrainãos.
22:34Então, Alex Karp, o vice-executivo e fundador, e eu viajamos para a Kiev em junio de 2022.
22:44Eu me lembro que nós nos acordamos muito early em Tchechev,
22:48na polêmica parte da fronteira.
22:50Você sabe, é um dia em 4 a.m.
22:52e sete-se em uma longa carreira,
22:55a viajada da fronteira de Polônia,
22:58a viajada da Ucrânia,
22:59e então 8 horas na rua para a Kiev.
23:05Nós nos acordamos em um lugar de encontro de Zelensky.
23:10O que foi extraordinário naquela reunião,
23:12foi, francamente, o que ele era.
23:14Ele era muito engraçado.
23:15Ele era muito engraçado.
23:17Eu sei que ele é um comédio,
23:19mas, ainda, isso foi uma coisa muito surpreendida,
23:21e uma coisa muito estranha.
23:22Você sabe, algumas vezes ele tinha uns em stitches.
23:26Eu acho que Zelensky entendia
23:28que o maior desafio que os Ukrainãos enfrentavam
23:30foi uma das massas e de números.
23:33Eles foram outnumbered
23:35por os russos,
23:36e também outgunned.
23:38Eles tinham mais armamento,
23:40mais industrial capacidade do que os Ukrainãos fizeram.
23:42E então, a única forma de corrigir essa
23:46imbalance
23:47foi com tecnologia.
23:49E Zelensky explicou
23:50que ele queria que nós
23:51vim e ajudá-nos com o trabalho
23:52na guerra da guerra.
23:53E que ele acreditava que
23:54que a Ukraine
23:54era o laboratório de pesquisa e desenvolvimento
23:57para o conflito
24:00durante os próximos anos.
24:02Então, Alex disse sim,
24:04e então,
24:04eu e outros
24:07na equipe de Palantir
24:08para transformar isso
24:09em uma realidade.
24:11Nós nos oferecemos
24:12com uma plataforma de software
24:13chamada Foundry,
24:15e, na sua base,
24:16é uma plataforma de dados
24:17e, na sua base,
24:46uma plataforma de dados
24:47e, na sua base,
24:48é um problema de dados
24:50e, na sua base,
25:01um problema de dados
25:03de dados
25:03e, na sua base,
25:05é importante.
25:08Então,
25:08você pode identificar um banco
25:11no campo de dados
25:12no campo de dados
25:12na Ukraine,
25:13mas você não vai ser capaz
25:14de dizer
25:14que esse banco
25:16pode ser o centro
25:18para uma parte importante
25:20dos russos
25:21dos russos.
25:21E para isso,
25:22você vai precisar
25:22de um outro tipo de dados,
25:24por exemplo,
25:26freqüências,
25:27sinales de inteligência.
25:28Então,
25:29se você combina
25:29sinales de imagem
25:30e sinales de inteligência,
25:32você pode dizer
25:33onde algo está
25:34e o qual é.
25:35e isso permite,
25:37com o município
25:39que existem hoje,
25:40destruir
25:40com a pin-point
25:41accuracy.
25:46Em novembro de 2022,
25:48nove meses
25:49depois da guerra
25:50começaram,
25:51chat GPT
25:52nasceu
25:52mostrando a cena
25:53mostrando o poder
25:54de uma nova tecnologia
25:56de linguagem
25:57model.
25:59Palantir
25:59e os Ukraines
26:00são rápidos
26:00com a grande
26:03tecnologia
26:04da guerra.
26:05a emergência
26:05de linguagem
26:06model de 2022
26:07tem um impacto
26:09na batalha.
26:10Se você imaginar,
26:12há milhares
26:13ou tens
26:14de milhares
26:15de dados
26:15que precisam ser monitorados
26:17no momento
26:18na batalha.
26:19Análise,
26:20tradicionalmente,
26:21teria que
26:22revisar
26:23todos os dados
26:24dos dados
26:24individual
26:25e manualmente.
26:28Largagem
26:29model de linguagem
26:29como sabemos,
26:30podem synthesizar
26:31essas informações
26:32rapidamente
26:33e poderállar
26:34importantes
26:35inferências,
26:36links
26:37ou pontos
26:38dependendo
26:39que você está
26:39procurando.
26:40E eu acho que
26:41foi um papel
26:42crítico
26:42naquela fase
26:43da guerra.
26:45Não há algo
26:45difícil
26:46sobre a empresa
26:47que está
26:48empurrando
26:48o processo
26:49de matalha.
26:50É algo
26:51que éramos
26:53cada dia.
26:55Mas,
26:55finalmente,
26:57eu acho que
26:59a tecnologia
27:00é a melhor forma
27:02de preservar
27:03a paz
27:03e, portanto,
27:04salvar vidas.
27:11Mas,
27:11A.I.
27:12não foi a única
27:13tecnologia
27:13que levou
27:14a grande
27:14avançada
27:15na Ucrânia.
27:17Também
27:17foi o desenvolvimento
27:19de avanços
27:20dos soldados
27:21dos soldados
27:23describiam
27:24como os drones
27:25transformam
27:25a batalha.
27:26dos soldados
27:30dos soldados
27:31dos soldados
27:31dos soldados
27:33dos soldados
27:36Eles têm
27:37começam
27:37como ferramentas
27:38das
27:39agilidades.
27:41Mas,
27:42logo,
27:42tem sido,
27:42foram
27:42mortos.
27:49Um
27:50Esse é um drone FPV, mas com a SEID. Ele trabalha como um bombeiro.
27:58Ele tem uma camada de volta e SEID.
28:03Ele SEIDs com essas armas.
28:11Por 2024, cada lado tinha mais de um milho drones.
28:16Tudo bem, acíratamente.
28:21É para não o secretário, que de cerca de 70% de brinquedos são drons.
28:33Quando você tem, digamos, uma arma de PZRK,
28:39que custa 60 mil dólares,
28:43tem um drone, que custa 2 mil dólares.
28:46E a impressionante, em princípio, está no nível.
28:51Com uma camada de cerca de 30 quilômetros,
28:53eles transformaram a front line.
28:57Então, a guerra muito muito mudou por causa de brinquedos.
29:01Agora a técnica quase não pode chegar ao primeiro lugar.
29:07A que pode chegar, às vezes, muito rápido.
29:10e aumenta com as línias, com as línias e com os inimigos,
29:14são muito mais fortes.
29:16Por isso que quase qualquer drone
29:19desistiu até o primeiro lugar.
29:23Automata! Automata!
29:25A Ucrânia demonstrou, que a fundamental character da guerra
29:30mudou, com drones que matam hundreds de milhares de soldados.
29:38Por isso que agora, em meio que cada 3 ou 5 minutos,
29:46na área de fronteira,
29:48tem como mínimo um drone.
29:50É a sensação, quando eles constantemente escutem.
29:54O que você entende.
29:56Porque você entende,
29:57que quando você não pôs,
29:58e que você não fez,
29:59a hora ou tarde você vai te encontrar.
30:06A Ucrânia demonstrou.
30:08A Ucrânia.
30:11A Ucrânia.
30:12Estas novas tecnologias de AI e drones
30:15permitiam uma armada pequena para manter a terra,
30:18embora em uma bruta e intratável.
30:22Mas, logo, essas tecnologias seriam usadas em Gaza,
30:26com consequências devastadoras.
30:35Apenas bu estação!
30:36Aláhuia!
30:37Aláhuia!
30:39kuin as marculas!
30:40Aláhuia!
30:43Oaxaca, um bom命!
30:46Aláhuia!
30:47Oaxaca, um bomlandês mal hatá!
30:47Aláhuia!
30:47Quando o anche muda fighter en hamas operatifs
30:49abriam Israelbe a Maar
30:51a bajonet área de empresas altasстаточноporte livre
30:57Social media quickly revealed the brutality of the attack.
31:21I was in my apartment, Andrew Zoom, just me and one of my flatmates.
31:27And I got up in the morning from all the sirens.
31:36Like almost every Israeli, I have family closer to the border with Gaza.
31:41And these moments, there were actual fear for their lives, I didn't know what's going
31:45on with them.
31:50For me and I guess for many other people, it was more like from the collected level of,
31:55you know, if people like me that were attacked, it could have been me.
32:06In a matter of hours, 1,200 Israelis were killed.
32:16251 hostages were taken to Gaza.
32:23The Israeli military and political response was immediate.
32:53The three intelligence analysts who were witness to the technical attack of the military and
32:57e o desenvolvimento de Lone Wolves, agora de volta à duty.
33:02Quando você chegar lá, você está em frente à um computador,
33:06não é como se você pode parar o ataque, mas você pode se sentir assim,
33:10ok, ok, eu vou fazer minha parte.
33:13A whole atmosfera foi, tipo, muito insano, e, eu não sei,
33:17ninguém realmente sabia o que eles estavam fazendo, eu suppose.
33:21Like, hoje em dia, eles falaram que eles foram, tipo,
33:25eles falaram que eles falaram, porque eles falaram que eles falaram,
33:29eles falaram que eles falaram, e eles falaram que eles falaram.
33:32Em primeiros dias da guerra, a maior tarefa foi criar o que é chamado de TARGETS.
33:40A TARGETS foi um desafio, porque o governo Israel disse que
33:45seguiu a lei internacional, em que a strike seria prohibido
33:49quando a falta de perdão de vida civil é excessiva
33:54em relação a a direção militar militar.
33:59There were complaints that there aren't enough targets
34:01in the Gaza Strip, even before the war.
34:04People high up were very content with the amount of targets
34:07that were being produced.
34:08They set up a thing called the TARGETS factory.
34:13Then they were like, ok, we just need to make TARGETS.
34:17Just finding locations, and finding people, basically, to assassinate.
34:34The Israeli Air Force posted multiple videos of their strikes
34:38and the numbers of their targets on their Telegram channel.
34:43There was an urge for extreme measures.
34:47So the general belief was that human work will not do
34:52to create the number of targets required.
34:55So the convenient solution was to use non-human ways
35:00of creating targets, which is namely AI.
35:13It was already clear that we have to bomb as much as we can.
35:19No realistic amount of humans could do this job
35:22in this amount of time.
35:25Then, naturally, they turned to machines.
35:29Computers created lists of locations and people
35:32that were passed to target rooms.
35:35There, other soldiers checked the targets were correct,
35:39assessed the collateral damage.
35:41And it's reported that sometimes
35:43they had less than a minute to make this decision.
35:47Early on, it was very clear that anything
35:50that has any connection with Hamas,
35:53and that includes also people who are not part of the military wing,
35:56but also part of the political wings,
35:59anyone who has a connection to Hamas
36:01is a legitimate target.
36:30How did the Israeli army generate
36:33so many targets?
36:37In the preceding years,
36:39the Israelis had built targeting systems
36:41that had been derived from the software
36:44used to prevent lone wolf attacks.
36:50At some point, we were using an AI algorithm
36:54that gets a data set of people
36:56who are approved Hamas members
36:58and looks for people with similar attributes
37:01in general data sets of the entire Palestinian population.
37:06There were a number of different systems.
37:09One ascribed a threat probability
37:10to any individual between 1 and 100.
37:14Anyone over a threshold, say 90, could be targeted.
37:28If your requirement is to kill in numbers,
37:32you want to kill an impressive number of people.
37:35AI can give you theoretically an endless number
37:40as long as you're willing to give up on the precision.
37:45Within three weeks, Israel announced
37:48it had attacked 11,000 targets.
37:52And the Gaza Health Ministry,
37:55controlled by Hamas,
37:57had reported 8,800 dead
37:59and 22,000 injured.
38:03It was perfectly clear that we want this space of targeting
38:08with this level of crasseral damage.
38:18But in the end,
38:20what drove the high number of casualties
38:22was not technology but political and military choices.
38:27In the previous rounds of violence,
38:29the number of civilian casualties deemed acceptable
38:32by the Israeli military had been much lower.
38:36A different decision was made for Gaza in 2023.
38:44They have this decision at the source of the war
38:46that you can bomb any commerce operative anywhere he is
38:49and take 20 people along with him.
38:53Basically assassinating people in the homes
38:55just means that you know
38:57that people of the world only have killed
38:59wives, children.
39:05Israel basically took the license
39:07to kill a significant number of civilian casualties
39:11along with a target.
39:14Basically, you're right.
39:15Yeah, he's there.
39:16Boom.
39:18These intelligence analysts
39:21accuse Israel of allowing collateral damage
39:23of 20 civilians for any approved target.
39:27But they go further.
39:29There are actual numbers
39:31in Israeli military doctrine
39:33for each level of collateral damage
39:36which is allowed.
39:38I think they raised the base level
39:40and then for especially important targets,
39:43it could be even higher.
39:44So there was permission
39:46to kill 300 people
39:51with collateral damage.
39:54300 innocent victims
39:56for an approved target.
39:58Come on.
40:07On the 2nd of December, 2023,
40:10the Israelis appeared
40:12to reinforce the claim.
40:14They launched more than 400 strikes
40:17and one on a housing block
40:20in Gaza City
40:21where they said
40:23they'd eliminated a Hamas commander
40:25who'd helped plan
40:26the October 7th massacre.
40:28But in doing so,
40:30the strike had killed
40:30an estimated 300 neighbours.
40:32by 2nd of the soldiers.
40:35There were more than 50 hospitals
40:36in the nearby
40:37and then they were
40:39administered
40:39with more than 10
40:41high-level deaths.
40:42The number of the survivors
40:44were still under the land.
40:46You're still in residence.
40:47You're still in residence.
40:48You're still in residence.
40:50You're still in residence.
40:50You're still in residence.
40:50You're still in residence.
40:54I remember there was this operation
40:56where other innocent people
40:57were killed.
40:58I listened to the call
40:59beforehand of the guy
41:00saying,
41:01I'm in my family home
41:03and my whole family are here
41:04and I'm worried that we will get bombed
41:06and I finally will die
41:09and then I'm just sitting there
41:11at like 4 in the morning
41:12waiting for like the fucking planes to bomb
41:16this kind of bureaucratical
41:17apparatus takes
41:19like the personal responsibility
41:21of people
41:23and you don't see the face of people
41:25of you like killing you know
41:26but the weird thing is that I did hear these people's voices
41:29you know
41:30and I was like
41:33it's horrible to say this
41:34but you know
41:35I heard them crying
41:36when their relatives were killed
41:42I think a lot of what technology gives us
41:45is that it blurs the reality
41:48for us to be able
41:49to not be completely responsible
41:52for what's going on
41:55the decision was just
41:57to bomb and bomb and bomb
41:58if you want this space of bombings
42:01and still make them look
42:03legal
42:04let's call it that
42:06then you need at least this tool
42:08that does the combination of data
42:12to reach the legal threshold
42:14for what is a legitimate target
42:18in this sense
42:20this kind of technology is more
42:21of an excuse
42:25all three intelligence analysts
42:28are no longer on duty
42:29in the Israeli army
42:34by January 2026
42:37an Israeli official
42:39brief newspapers
42:40that they agreed
42:41there had been more than 70,000 deaths
42:43in strikes
42:43in the two years of the war
42:48organizations including
42:50Amnesty International
42:51Bet-selem
42:51and parts of the UN
42:52now accuse Israel
42:54of committing war crimes
42:55by amongst other things
42:57the disproportionate attacks
42:58on civilians
42:59in densely populated areas
43:01whilst Israel's strategy
43:03was driven by human choices
43:05the technology enabled the onslaught
43:09the Israel Defense Force says
43:12the IDF does not use
43:13an AI system
43:14that identifies terrorist operatives
43:16or tries to predict
43:17whether a person is a terrorist
43:19information systems
43:20are merely tools for analysts
43:22in the target identification process
43:24the IDF operates in accordance
43:27with international law
43:28each strike undergoes
43:30an individualized
43:31case-by-case assessment
43:32evaluating anticipated
43:34military advantage
43:35against expected
43:36incidental civilian harm
43:38proportionality decisions
43:40are made based
43:41on the information available
43:42at the time of decision
43:43and not in hindsight
43:44the IDF has also said
43:47that the 70,000 deaths figure
43:49does not reflect
43:50official IDF data
43:55whatever the lessons
43:56of the tactics
43:57in Gaza and Ukraine
43:59the race for warfare technology
44:01shows no sign
44:02of slowing down
44:03in January 2026
44:06the US Secretary of War
44:08Pete Hegseth
44:08was fully committed
44:11simply put
44:12the United States
44:13must win
44:14the strategic competition
44:15for 21st century
44:16technological supremacy
44:18we must ensure
44:20that America's military AI dominance
44:22so that no adversary
44:24can exploit
44:25that same technology
44:27to hold
44:27our national security interests
44:29or our citizens at risk
44:31America first
44:33in every domain
44:35in short
44:37we will win this race
44:38by becoming
44:39an AI first
44:40war fighting force
44:42across all domains
44:43the rest of the world
44:45is following suit
44:46Western governments
44:48have entered into
44:49hefty military contracts
44:50with American-based
44:52technology companies
44:53like Palantir
44:54Google
44:54Microsoft
44:55and Amazon
44:56making them ever more
44:57dependent
44:58on a few corporations
45:00for the next generation
45:01of military technology
45:02I think it's incredibly important
45:05if we want to maintain
45:06our way of life
45:07if we want to remain
45:09advanced first world economies
45:11if we want to keep
45:12our value system
45:13that we in the West
45:15broadly defined
45:15have the dominant militaries
45:18we have to maintain
45:19that technological advantage
45:20and if we are in an arms race
45:22that means we have to win it
45:26as drones replace missiles
45:28and computers replace men
45:29it seems that wars
45:31have not become cleaner
45:32more surgical
45:33and quicker
45:34if anything
45:35the costs
45:36in money
45:36and political capital
45:38of entering wars
45:38has declined
45:40and they're likely
45:41to become more frequent
45:44so now is the moment
45:46for citizens
45:46and their governments
45:47to decide whether
45:48just like for nuclear
45:50and biological weapons
45:51we need international agreements
45:53to control
45:54the new warfare
45:56I got to see like
45:58the advantages
45:59that AI gives
46:01a few years before
46:02the rest of the public
46:03and there's
46:04zero
46:05good value
46:07given by that
46:07it only
46:08creates more dust
46:10it only gives opportunity
46:11for deadly wars
46:13to be fair
46:14I really think
46:14it's a mistake
46:16it just creates
46:17more dust
46:17and more wars
46:26that any war
46:27is causing
46:29a growth of technologies
46:31and a growth of innovations
46:34and who will be the first
46:40to use these innovations
46:42are the same
46:43and who will be the first
46:45ready to work
46:58Unreported World returns with a new series beginning with the women risking their livelihoods
47:03and reputations, pushing boundaries on screen in Nigeria's prolific but highly conservative
47:08male-dominated film industry.
47:11That's next Friday at 7.30 here on Channel 4.
47:16Channel 4.
47:20A CIDADE NO BRASIL
Comentários