Programming in Ethics

...responsibility...?

Responsibility for determining culpability and liability is that of the legal profession, and usually happens some time after the event. Responsibility at the actual scene of an accident falls to the emergency services, which are there to provide an emergency response!

And is really there any such thing as an 'accident'?

Ho, hum.

Hi PWSlack,

Just suppose the "Johnny Cab" that Doug Quaid is riding in has a collision with "Benny Cab" who is driving around without a passenger. Quaid is not injured, of course, he's Swartzie. Johnny and Benny belong to different cab companies. Both cabs are totaled. Which cab is responsible for the accident? Let me rephrase; which cab caused the collision?

Their navigation software is in affect whether or not they have a passenger, so let's don't automatically jump on Benny (I have no idea why he's driving around without a fare. Maybe he's cruising for girl cabs or something).

Should their software include ethical modules that will cause them to accept responsibility for the collision, or should each point its headlights at the other and demand retribution, i.e., "blame" software instead of "ethics" software.

I guess when you come down to it, the cab owners will fight it out in court, and if thats true, then what's the use of "robotic ethics"? Seems more like an oxymoron.

-John

<...when you come down to it, the cab owners will fight it out in court...>

Q.E.D.

The legal profession makes its income out of other people's misery!

"Intelligent"? "smart"? These are words used by the "news media" and marketeers to sensationalise and hype technology. A complex machine is is still a machine thus the designers, owners and operators have responsibilities under existing law. There may no longer be a driving seat inside a hi-tech vehicle but someone, somewhere powered up that vehicle and must take responsibility for it until it is safely powered down.

I could not agree more completely! The operator of the machine has responsibility for it while they are running it. But there is also a responsibility the engineers that designed and programmed the machine to make it safe. Modern aircraft controls are an excellent example, in flight operation can be left completely in automated control but the pilot is ultimately responsible for its safe operation .

Mike

Of course you're correct. The machines should be made as safe as possible. However, the question is should the machines have a built in set of ethics, i.e., "ethical behavior"? Should the machines be programmed with a sense of right and wrong and thus not do anything that is "wrong" as per their programmed definition of wrong?

I know this could really get squirrelly but lets approach it not from the standpoint of engineer/operator responsibility but rather how the machine should perform once it is doing an autonomous task.

-John

Setting aside the fact that a machine cannot be endowed with "ethics", even if it were possible, who decides the ethos? The designer must still accept full responsibility for the "ethical simulation rules" ie what is "right" and "wrong". It seems as though some would seek to avoid responsibility for any carnage caused by their technology let loose upon the world.

As I said Guest, the whole thing could get squirrelly.

Lets just assume, for a moment, that an autonomous android could be built. I'm talking about one that could learn, and be influenced by its environment. Would it know the meaning of punishment for wrong deeds? Would it care? Could a sense of felt altruism be initially incorporated into it?

We seem to be headed toward Asimov's Three Laws of Robotics,

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 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.

There is a long, long way to go before any program or programming method could make any of these decisions. Maybe in another 100 years?

Mike

Is #3 the "Prime Directive" then?

You said "There is a long, long way to go before any program or programming method could make any of these decisions. Maybe in another 100 years?"

I tend to agree with you on that MrRoboto, however at the rate technology is advancing don't be too sure though.

The one thing I am sure of is that no robot will ever possess true consciousness. Emulation, perhaps, but that's all.

-John

A machine cannot "know". Any operating logic, even that intended to endow the machine with an illusion of autonomy or survivability is generated by the designer and will contain flaws, features etc that say far more about the designer than the machine. As for "altruism", my interpretation may well be different from yours , so who is right? Who will judge? Worst case scenario might be that the machine armed with "altruism" decides that it is in our best interests that we are all put out of our misery. I wonder why, in a world where many people seemingly lack empathy, humility . . and altruism, some are so keen to endow machines with allegedly human traits. We should fix ourselves first.

Hi FlyBlue,

You said "some are so keen to endow machines with allegedly human traits. We should fix ourselves first."

I don't disagree with you on that but technology, and that includes robotics, will march on. Man will always attempt to imprint himself on his surroundings. It's his way of achieving a legacy. In short, man's main purpose in life is to give birth to himself.

Some will try to do it via robotics.

-John

Hi Johnjohn,

I accept your idea that "Man will always attempt to imprint himself on his surroundings. It's his way of achieving a legacy", if only to to say to some far off future that "we were here!". However, I feel that often for many people, it is the fear of death - the fact that all our bones turn to dust one day (a great leveller) - that creates a possibly unhealthy obsession with developing robotics in the hope that technology will enable them to live forever. Some might say that is their choice - fine by me - provided the development of such technology is not to the detriment of those of us who are comfortable with our low tech fate. It is not too hard to imagine a future hi-tech vs. low-tech polarisation of global society.

Maybe some of us will become a direct ancestor to one of our own children here.

Not likely but who knows...

In UK law, the driver of a road vehicle is legally responsible for its safe operation. If the driving task is delegated to a software-based system, does that mean that the programmer is now the driver and the person sitting in the seat closest to where the steering wheel used to be is now simply a passenger, in law? Discuss!

London Underground's 'Victoria Line' and the 'Docklands Light Railway' are automated transport systems, among a number of other fine examples worldwide. Fortunately, the automation needed to keep a limited number of similar vehicles from colliding with each other along a linear transport corridor are relatively simple today, given that random obstructions like pedestrians, stray cows and wobbling cyclists are largely absent. On rail systems the legislation is different; any person wilfully causing a threat to a train can face prosecution, even if their car is written off as a result!

Now consider developing that automation onto the road network, with an enormous variety of vehicle shapes and sizes, some automated and some not. Think about the potential obstructions to the safe passage of a potentially fully-automated road vehicle and it becomes obvious what an enormous task the system programmer would be set with, quite apart from the sensor developer and the maintenance tecnician. And who would want to tackle that task, faced with the possibility of being dragged into a court after every collision?

The skills acquired in becoming a safe, licensed road vehicle driver are not easily learned and often undervalued, if accident statistics are anything to go by.

^ There's a bug in the program. Or it might be a sensor malfunction. Sorry....

Interesting.

I just heard on the news the other day that there's a proposal to largely replace airport traffic controllers with an automated (computerized) system. According to the report, planes are presently kept 3 to 5 miles apart. With the automated system, planes could be separated by no more than about 1 mile. Supposedly this is intended to decrease airline delays, etc.

At airplane speeds, 1 mile does not give a pilot much time to react to a software glitch. I think I would rather stick with humans in this case, thank you very much.