The Human Resistance Report

Will robots adhere to the Three Laws?

By now, most of us are familiar with Asimov’s three laws of robotics.  Isaac Asimov was a science fiction author, with many notable works, including the famous  “I, Robot” .  Having written some of the first stories about robots, and so many of them, he also went so far as to declare a short constitution of laws that they should abide by – primarily known as the  “Asimov’s Three laws of Robotics.”

The idea is that if robots are programmed with these three laws in mind, then no harm should ever come to a human at the hands of a robot. Interesting to note that while functional robots seem relatively new today, these laws have been around for over 70 years now.  For those who need a reminder of what these three laws are, here they are as they first appeared in his book “Liar!”(1941) :

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey any orders given to it by human beings, except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

So the idea sounds good on the surface and we don’t have killer androids running around decapitating people and each other – following law as humans do. And at face value, the laws seems to work with each other very well , avoiding loopholes that the robot might use to get out of one the laws to hurt someone.  Of course any lawyer worth his salt  would be quick to point out that his client robot did not cause harm to anyone when he stole something form them….but that is a whole other discussion.

The main problem here is that robots are not even in full domestic production, and they are already going around breaking their own laws before they can even establish a presence in society.  Already, unwitting robots are being sent into other countries where they not only hunt down humans.. but also kill them. There have already been news reports going back as far as two years ago indicating that not only are bad guys in the crosshairs, but that sometimes civilian families with children get caught up in the mix and are killed right alongside intended targets.  Clearly this is a violation of the first law of robotics.

But a little research into the subject will find that this law applies specifically to robots acting on their own will. So far as we tell, the drones that are bombing targets are doing so under explicit command of military leaders.  But that doesn’t change the fact that the very nature of automated machines are being used as killing tools and it is no big stretch to imagine that someday they may allow them to make tactical decisions on their own and carry them out at will.

This is not very far off from what has already been happening for over a decade now with the Apache helicopter and other military weapon command centers  designed with a target acquisition systems able to identify enemy vehicles and people on the ground. It uses object recognition to determine whether an object is a jeep, a tank, a chopper or a human (among many other objects it can identify). It then uses this data and a computer algorithm to determine the order of importance and can then cue up the targets in the right order to eliminate.  When the ground commanders give the ok .. they press a button and all of the targets are taken out in short order using something they refer to as  “Fire and Forget”, pulling the trigger and knowing that there is about 100% chance that their target will be destroyed.

Such confidence in systems can only increase as artificial intelligence improves. The military has shown that they are indeed excited about the possibilities of sending battle bots into war instead of people… and why shouldn’t they be? By automating such messy tasks like bombing villages and gathering intel, human resources can be put to better use, like developing even smarter bombs and smarter robots to deliver them.  If they have proven accurate enough to provide reliable intel to humans in past battlefields, then future battlefields can see these robots given the authority to execute the targets they cue up with fire at will instructions.

The rules of robotics are going out the window and the idea that we could harness them in such a paradoxical triage of constraints is ridiculous to begin with. The very nature of the robot is as a tool, and Asimov even addressed this with the three laws that could be applied to all tools in the way that they apply to robots.. i.e .. Tools should be safe, use of them should be efficient unless it is not safe…ect.  But the main difference is that these are tools that can be programmed to do what their owners demand of them.  This means that robots will do anything or break any law that they are programmed to.

Even a very intelligent robot can be fooled into breaking their own laws by their own will. For example, if a robot was given the choice to save a human life at the expense of its own, would it do so?  Rule one says inaction is a violation so he must take action to save the human. Law Three says he must not destroy himself unless not doing so conflicts with laws one or two. So clearly he would give his ‘life’ for a human, if following the laws. But what if instead of his own life, he had to choose between two lives…. one of a child and one of an escaped criminal. Does (or should?) the robots have the aptitude to make a moral decision based on some subset of underlying values? Or does its brain just fry – unable to process,  uttering over and over again… cannot compute… cannot compute…

We can see here in that example that it will have to make a choice of some sort and either choice will break one of the laws. Action to save one is inaction to save the other.  We can also take from that that a robot would destroy itself if instructed to do so by a human, since he is bound by rule two.  Since his instruction given to him by a human to save both humans, perhaps, in order to avoid making a decision, it would self destruct. A more likely scenario is that it approach it like a task. Do one task , then the other, save one human .. go back and save the other human… but by the time he got back to the second human, the condition no longer existed as there was no longer a living human to save. The first human saved would live and  there would not be time for the second life to be saved, and as far as the droid is concerned , it has fulfilled its obligation to ‘save the humans’.

Chances are that robots will never truly follow the three laws, but may come within close proximity of it. Engineers making robots for uses other than carpet bombing will have to put a lot of thought into the decision tree robots will use to make choices in the absence of a human command.  Until robots actually develop an actual conscious, they will be relegated to executing the will of man… which generally does not even come close to the laws of robotics, thus exposing the fallacies of the Three laws.

In a report recently aired on CNBC with Maria Bartiromo, the issue of  a workforce  in the future was the topic. Their discussion was basically asking  “how are people are going to find employment in the future when so many jobs are becoming automated?”

There was  a lot of interesting concepts, but they concluded essentially that those with education in mathematics and engineering are the most likely to succeed in such a future.  This got me to thinking about what jobs exist right now that can be automated or replaced by machines.  My last article discussed the possibility of   replacing police with artifical intelligence, facial recognitions, cameras, UAV s and modifying social behaviors with ultra invasive tracking networks.  Sounds  a little paranoid come to think of it,  but now I am asking the question  – how many other jobs can replaced? As it turns out , after some research – there are quite a few . I will start with the most visible jobs and build up from there.

Cashiers

So far, many stores have already started experimenting with automated cashiers. You’ve  seen it at Walmart  and Home Depot,  now picture them everywhere you shop.  Bank tellers have long been replaced by ATMs for years and now many newer start-up banks do not even bother to have physical locations!  With the popularity of do-it-yourself- check out systems , you could end up seeing all major food chains offering the same service – as a matter of fact, it is already quite popular in China other countries with dense populations.  Although it would be pretty easy to have customers  punch in their order on  a touch screen -  more than likely , you will build an order on your  iPhone  and it will already have your favorite meal cued up – then just beam it to the automated cashier  while a cook in the back complete the orders.

Eventually we could see all stores using automated tellers and cashiers. It will save the company  money and reduce employee mistakes/theft,  and lower insurance  and wage taxes.  As a result, our already declining level of customer service will reduce even further as we are forced to serve ourselves  – think it won’t happen? when was the last time a service attendant came out and pumped your gas for you?

Taxi / Limo / Bus drivers

The taxi of the future already on display

At some point, vehicles will be self driven using built in GPS and collision detection,  so there’s no reason why the majority of cars won’t be driving themselves and you around someday.  So why would you pay a driver to steer your taxi around town when it can just as easily do it on it’s own.  Safety they will tell you -   After all ,  your car will be able to communicate and  have a dialog with  on-board systems from other vehicles and will be able to monitor speed and braking with precision – a slight upgrade from today’s typical communication protocols witch include honking, flips-offs, and the word “joikof”.

Pilots

UAVs continue getting larger and more sophisticated

OK , lets apply the same concept to airplanes. After all , they can already fly by themselves  good enough to go to another country , snap photos, do flybys, oh … drop bombs on stuff. Lets face it , these things know their way around the sky and with all of the recent press about airline pilots and drinking incidents, its just a matter of time before people starting putting two and two together to realize that they might be better off in a precision piloted aircraft.

Nurses

As it is , many of the devices hospitals are using are already self monitoring and automated.  Still however, we need somebody to attach that stuff to us. It’s nice  getting served our food and medicine while incapacitated, but many hospitals have realized that many of these tasks can be accomplished by a roving robotic nurse. Using automated bots for simple things like delivering medication to a patients room or serving  food can free up nurses to tend to more immediate emergencies – like resetting the robotic surgeon in the other room.

Surgeons

Remote surgery is already common

There have already been several examples of  doctors operating remotely using robotic arms to perform surgeries considered to require extreme fine precision. As object recognition evolves into real world applications, it will become possible for robots with cameras to recognize and map out human organs and blood vessels, allowing them to perform complex procedures using real time data. It’s likely that we can see some of these remote surgeries performed entirely by machines with assistants on standby.  The only problem is, they wont make for very interesting soap operas.

Teachers

Can Smarter Classroom Lead to Smarter Students?