Robotics Prof Fears Rise of Military Robots

An anonymous reader writes "Interesting video interview on silicon.com with Sheffield University's Noel Sharkey, professor of AI & robotics. The white-haired prof talks state-of-the-robot-nation — discussing the most impressive robots currently clanking about on two-legs (hello Asimo) and who's doing the most interesting things in UK robotics research (something involving crickets apparently). He also voices concerns about military use of robots — suggesting it won't be long before armies are sending out fully autonomous killing machines."

Comments

"Friendly AI"

[Baldrson]

This is one of the things that makes me think the concern about "friendly AI" is blown out of proportion. The problem isn't making sure teh AI's are "friendly" -- its making sure the NI (natural intelligence) owners of the AI's are "friendly".

If half the effort spent on "friendly AI" were spent on examining the ownership of AI's, there might be some hope.

Re:"Friendly AI"

[Shihar]

As dark as the potential for drones can be, I think it actually has the chance to make war a far less indiscriminate and bloody thing.

Right now, if a square of Marines gets fired on, they can return fire. A square of marines has the firepower to flatten a village. Give them access to artillery or air support, and they can literally level a city. In other words, whenever you have a squad of supported marines fight, you are having a group of kids (and they are just kids) holding their finger over enough firepower to take out a small army. Their job is to use as little as that firepower as humanly possible. You might be able to level every building in a half mile radius, but you are not supposed to. When it comes to a firefight though, especially a desperate firefight where soldiers have their lives on the line, they, like most humans, choose life over death, and if that means flattening an entire apartment building to get at one sniper, they do it and hope that no one else was inside. Generally speaking, unless a soldier walks up to a civilian and splatters their brains on the floor, they are let off free. It is war, your life is on the line, you take your risks and respond in the best way possible. If a civilian gets accidentally whacked, that is sad but acceptable. Most soldiers develop a pretty thick "us vs them" mentality that see civilians if not the enemy, as hostile terrain, especially in a guerrilla war.

Drones offer up another possibility. It is true, you can order a drone army to go out and kill civilians and it is probably easier to get a soldier to do it. That said, if you policy is civilian murdering, a nation like the US doesn't need to use drones. You can handily exterminate all life through impersonally aerial bombing. What drones offer is more control over the rules of war. Rules mean little when you are surrounded by gunfire. You do what you have to do to survive. On the other hand, when you are sitting in the US with a military lawyer over one shoulder, a commander over the other, and and every single second and action you take is getting recorded, rules are a lot more enforceable. If the rules call on you to die before you level an apartment complex just to get at one sniper, a drone can simply die. A soldier generally wont.

With drones, you have complete accountability for your actions. You can always go to command before doing something. You never need to make snap judgments. Hell, you can call a damned military lawyer over and get his take on the rules of engagement. Further, every bloody thing you do is being recorded, so if you decide to start murdering civilians you will be caught and tried.

On the balance, I think drones are going to lessen the lives lost. The few potential abuses are pointless to worry about. If someone wants to exterminate another people indiscriminately, you can do it the cheap old fashion way of aerial bombardment. On the other hand, if you are an army that wants to enforce ironclad rules of engagement, drones ensure there is never an excuse for fucking up, and that fuckups get caught.

Re:"Friendly AI"

[tsm_sf]

I'm not really worried. I'm sure we'll hear about some 'bot wiping out it's own platoon in the next decade, and that will be the end of semi-autonomous killbots.

In fact, I'd be very surprised if this didn't happen in the next ten years. Armed robots are a great idea in that they'd cost less than a fully trained human and are more easily repairable. It's a natural way to go for the military. I also know enough about software development to see that a catastrophic failure is fairly likely, and that the idiot-proof failsafe they'll set up will turn out not to be and won't, respectively.