Will AI robots on the frontline mark the end of human soldiers? - BBC World Service

Will AI robots on the frontline mark the end of human soldiers? - BBC World Service

BBC World Service

0:02 In February, Ukraine's military made an incredible claim about this robot.

0:07 For 45 days the drone rotated into combat duty and repelled enemy assaults.

0:12 According to soldiers, the bot was able to hold off attacking Russians

0:15 on the front line at some point last year.

0:18 Then, in March, an even more incredible claim from Ukraine's president.

0:22 For the first time in the history of this war,

0:25 the enemy position was taken exclusively

0:27 by unmanned platforms ground robots and drones.

0:31 The new deadly robots designed and deployed in Ukraine were lined up on display.

0:36 The message from the president was...

0:38 The future is here.

0:40 Zelensky's military also claims to have

0:42 forced Russians to surrender using robots.

0:45 Ukraine has become a testing ground for robotic warfare.

0:49 This conflict, and others like Iran are propelling us towards a world

0:52 where increasingly sophisticated robots designed to kill

0:56 will become common on front lines.

0:59 As the company commander says robots don't bleed.

1:02 Is this really the future of warfare?

1:04 And if so, what does it mean

1:07 for human soldiers and the civilians impacted by conflicts?

1:12 Uforce is a new Ukrainian weapons maker that's grown fast.

1:16 It recently became a unicorn a company worth a billion dollars.

1:19 Some of its land and air drones were used in the operations Zelensky revealed.

1:24 Although they're not saying much about it.

1:26 We met the firm at its London location,

1:28 which is secretive and unbranded to protect it from potential Russian sabotage.

1:34 So I can't go into specifics on the operation

1:37 or how Uforce were involved but we've

1:41 done over 150,000 successful combat missions since

1:45 the full-scale Russian invasion of Ukraine in 2022.

1:49 Do you foresee a future where you've got

1:51 more robots in a fighting squadron than humans?

1:55 That could potentially happen so with the system you see here

1:59 a control centre could have up to ten robots going into it

2:03 at one time so it is potential where you could have

2:08 more of these systems working together with one person overseeing them.

2:13 The company says this drone has been

2:15 used to destroy more than 150 Russian tanks.

2:19 Russia too is deploying its own robots,

2:21 for example to carry explosives into Ukrainian positions.

2:25 And for years now it's matched Ukraine in the air with drones.

2:29 I really consider Ukraine to be a major

2:31 teacher in the future of national defence and armaments.

2:36 It's a really impressive case study of how necessity breeds invention.

2:45 In the US, another weapons startup Anduril caused a stir in February when it

2:49 completed a test flight of its new fighter jet that has no fighter pilot.

2:54 Like Uforce, Andurill Industries is part of a new

2:57 wave of disruptive startups looking to reshape modern warfare.

3:01 It's raised billions of dollars in funding and US military contracts.

3:05 More established and larger defence giants like BAE Systems,

3:08 Boeing and Qinetiq are also working

3:10 on increasingly powerful unmanned robotic weapons.

3:14 You've seen Ukrainian and Russian drones fight each other.

3:18 So there is already a kind of robot-on-robot aspect of the battlefield.

3:24 Seeing that play out increasingly on the land and maritime domain,

3:27 I think, is extremely likely if not inevitable.

3:31 At the moment, most military robots across

3:33 all domains are being controlled by humans, but autonomy is accelerating.

3:38 Uforce's land drones use aim-assisting software for shooting.

3:41 And Anduril's drones can carry out the final stage of strikes on their own.

3:49 I saw that it has a function where

3:51 it can sort of kamikaze attack a target autonomously.

3:56 Is that right?

3:57 Yeah, that's correct.

3:57 I was wondering about this though because that kind of the last 200,

4:01 300 feet, what if a kid runs out or something?

4:03 These things have a wave-off feature.

4:06 Now, of course, at one point or other it's all too late,

4:09 but the speed they go, they're going pretty fast when they get down there.

4:14 So, you know, unless you're Linford Christie running out there,

4:15 you're not gonna be in the way.

4:17 You're not concerned about something like this, you know,

4:19 accidentally killing a civilian because it takes

4:21 a decision and can't get out of that decision.

4:25 This is where we keep the human in the loop.

4:29 So that last thing they do will make the last check and press.

4:32 The use of AI and robotics I believe,

4:35 will reduce the amount of mistakes made in the battlefield.

4:38 Humans need to eat.

4:38 They need to sleep.

4:39 And progressively, in a wartime situation,

4:41 they get less sleep and they eat less, and they tend to make more mistakes.

4:46 By using computing we avoid the ability to make mistakes

4:51 through more of that, what we call the kill chain.

4:54 The US administration is openly urging its military to embrace AI in all areas.

5:00 We will win this race by becoming

5:02 an AI-first warfighting force across all domains.

5:06 In the Iran strikes, AI tools from companies like Palantir

5:10 and Anthropic have reportedly played a major role.

5:14 Aside from occasional military parades,

5:16 Chinese weaponry developments are less well known.

5:19 But a US Department of Defense

5:21 assessment last year claimed that China's military

5:23 is rapidly developing in emerging tech

5:26 like drones and increasingly relying on AI.

5:30 The whole reasoning behind militaries using AI

5:34 systems is to increase the pace and accelerate,

5:38 for instance, identification of It poses huge

5:42 ethical and human rights risks and issues.

5:47 We should never really be wanting

5:49 to delegate life or death decisions to machines.

5:53 There's another concern too.

5:55 With killer robots able to think for themselves,

5:57 what happens to safety if their handlers lose control

6:00 of these deadly devices in, for example, a cyber attack?

6:05 Well, anything can happen when you're hacked.

6:07 What's important is how well you protect yourself from hacking.

6:11 The cybersecurity around what we do is hugely important, as you might imagine.

6:16 And to do that, we adhere to whatever government we work with.

6:20 I work with the MOD.

6:21 The MOD have a series of cybersecurity policies

6:24 and rules that we must follow, and we follow them.

6:27 Most experts say we are many years

6:29 away from these autonomous robots becoming commonplace.

6:32 Humans will always be needed on the battlefield, in some form,

6:35 due to a key limitation that seems hard to change.

6:39 You need a command link between the operator and the robot itself.

6:44 That command link can be severed either by electronic warfare,

6:47 by jamming, or if the robot has a physical connector,

6:50 we're starting to see fibre optic cable being used to connect ground robots,

6:55 as well as uncrewed aerial vehicles to their control stations,

6:59 which is unjammable but can be severed by explosions by shrapnel,

7:03 or even just by a pointy rock.

7:05 There's another area of war that robots and AI will likely never replace us.

7:11 While I really do expect these systems

7:12 to become more and more integrated into combat,

7:14 we have to keep in mind that robots destroying robots won't end wars.

7:20 All wars ultimately end through some kind of political

7:23 settlement when one side has suffered more cost and loss

7:26 than it can tolerate and that will remain true whether

7:30 the fighting ultimately is done by humans or by machines.

Study with Looplines Download Captions Watch on YouTube