ED-209 and the Fear over Programmed Robotics
The ED-209 is a high-performance law-enforcement robot featured in the first Robocop film. It first appears before Robocop is made, less than nine minutes into the first film at a product demonstration in front of the Omni-Consumer Products board of directors.
The display in concept, is simple. A volunteer takes a loaded gun and points it at the ED-209. The 209 reacts automatically, identifying the gun, the posture of the person holding it, and referencing the local laws, to conclude that where the gun is being pointed is not lawful. A master stroke of law enforcement, this entirely programmed machine can augment officers on the beat, whilst being much faster, much tougher, and much more heavily armed. At least that is the theory.
So, enter the 209. It bears a strong resemblance to the real-world robot, Honda's ASIMO in that it did not learn how to do anything. Instead every action, every process is preprogrammed, and there is not a lick of sentient thought anywhere in its body. It is the archetype for what society commonly thinks of robots as: If its not programmed explicitly on how to do it, it cannot do it.
This lack of ability to reason is what in theory makes such robots 'safe', but in practice, it is what dooms them. They cannot react outside their programming, so if they are put in a situation outside of their programmed parameters, or an error or malfunction scrambles some data, they do not know what to do, and just continue on their programmed path as best they can.
As its creator puts it, The enforcer droid series 209, is a self-sufficient law-enforcement robot. 209 is currently programmed for urban pacification, but that is only the beginning. After a successful tour of duty in Old Detroit, we can expect 209 to become the hot military product for the next decade.
This also jives well with real-world applications of such machines. Mindless robots, doing only what they are programmed to do, would be the ultimate tool in the toolbox as far as utter control over one's own forces are concerned. Of course, without any AI or AGI of their own, these machines can <i>only</i> do what they are programmed to do and nothing else which is the whole point. Unfortunately this means again that they cannot respond to change, just as much as they cannot disobey orders.
Back with the demonstration, Mr. Kenny, a young department head brought to the meeting with his superiors is invited to take the stand, which he eagerly does. He is going to help simulate a typical arrest and disarming procedure, and is supplied with a high capacity firearm for the purpose. He takes it and points it at the speaker. The 209 does not react this is outside of its programming.
Sighing, the speaker says not to point at him but at ED-209. Mr Kenny does so, and the response is immediate. ED-209 lifts up onto its haunches, arms its own guns, and speaks in a loud stern voice through the speaker grill. Please put down your weapon, you have 20 seconds to comply.
The speaker advises that Mr Kenny had better do as he says, and Mr Kenny nervously nods, and throws the weapon down onto the carpet in front of him. It lands with a loud whumph, echoing throughout the room. The young department head is not going to argue with that much fire power.
The 209 leans forwards and emits a primal growl, both its weapons batteries pointed squarely at Mr Kenny. You now have 15 seconds to comply.
The technical team look at one another in alarm. This is not going as planned. They begin to fiddle with a control console hurriedly, as the 209 reads off which laws Mr Kenny is in violation of believing he still has the gun, and did not drop it.
As the 209 counts down, Kenny panics, and rushes between board members trying to hide, as the 209 follows its programming to the unreasoning letter, tracking every move he makes, with its gun turrets.
209 announces it is now authorised to use deadly force, as Kenny is pushed into the clear by two board members frantically trying to get the guns pointed away from them. He stands, horror-stricken in empty space as the 209 opens fire with twin high-calibre launchers, firing over 160 rounds straight into him, before the lead tech manages to disconnect ED-209's power supply. The force literally picks Mr Kenny up and smashes him bodily into a project display table.
Not exactly the 'typical arrest situation' planned.
In the panicked situation that follows, Mr Kenny is very, very dead, and the executives are left floundering, unsure what to do and what not to do. One of the techs cries out He didn't hear it.
The 209 had not properly processed the sound of the gun hitting the floor, so to its mind, the gun had never hit the floor. It was not programmed to follow the path of the gun from the user's hand, to lying on the ground. So although its visual sensors saw the gun follow that path perfectly, there was no way for it to equate that movement with the gun not being a threat any more.
This is the critical failing of a programmed mind, and is why an adaptive AI or an AGI mind is essential for such robots to work alongside humans, or in any real environment. They must be able to think outside of any programming, and come to their own conclusions based on what their senses tell them, just as a human does. High-level, strong, adaptive AI can do this. AGI is even stronger, and can also do this. The robot is no-longer programmed what to think, but becomes capable of making its own choices and reacting to stimuli.
Much as the programmed mind has advantages in that it will never rebel, or think an unauthorised thought, it is also only capable of the thoughts and actions, some programmer thought of beforehand. An increasing number of modern robotic systems are essentially dumping the programmed mind paradigm for this reason, even though a self-aware mind is capable of not following orders in some circumstances. We have to allow robotic minds to become more human, in order to fully trust their ability to react to circumstance. However, that ability to react to circumstance and to make judgement calls itself means they may find their orders don't necessarily apply.
That little nugget is central to all the Robocop films, as it shows what is effectively a human mind in a robot's body and how judgement calls conflict with duty.
There were two further incidents with ED-209 robots in the film. In the first it was defeated by a too narrow flight of stairs, which was more a severe body design flaw than a programmed mind flaw.
In the second incident however, it was defeated by an anti-tank cannon. A self-aware mind would know that when you face a weapon that can penetrate your armour, you do not stand still and give your opponent twenty seconds to put the gun down before you fire. A programmed mind knows no such thing, and dutifully gives its opponent all the time they need to line up and send two shots straight into it. Yet another glaring flaw in the 'programmed mind' approach.