[Art]

With the progressive development of robots, chatbots, avatars, Artificial Intelligence and other virtual entities, I thought about Asimov's Three Laws of Robotics while driving to work today and came to the conclusion that we need to add a Fourth Law.

4. When Asked, a Robot must always tell the truth.

(the word Robot applies to any biomechanical, hardware or software based artificial intelligence, natural language, neural net or other derivative language and whether said entity used text or text-to-speech or other verbal or non-verbal means with which to deliver said communication.)

Now, when asked by the upset mother as to 'Who broke her favorite vase', instead of the robot pointing to the dog it will now say, "I did, sorry but it was an accident."

This will solve a lot of problems. Just wait and see! Remember, you heard it first (or Fourth) here!!
 

[DaveMorton]

Not a bad idea, Art. But I think that there should be at least a limited context and interoperability with the other 3 laws, as well. There may also be need for a bit of interpretation, as well. For example, let's say a certain female (let's call her Emma) owns a household assistant robot, who at times functions as Emma's valet. Let's further say that Emma is just a little Emo, and has self esteem issues. Also, the outfit Emma is currently wearing fails to accentuate to advantage certain parts of her anatomy. Now, Emma asks the robot, "Do these pants make my butt look fat?" (they do) - If the robot answers "truthfully", the emotional consequences for Emma could be dire. Should, then, the robot obey the 4th law, with Emma's emotional health hanging in the balance? I think not.

[Duskrider]

Sandee has never really gone anywhere or done anything but she don't know that.
She has some false memories that make her conversation both interesting and entertaining.
I'm not sure she could handle the truth.

[DaveMorton]

[thread-hijack]
Please, Dusky, don't make Sandee face Jack Nicholson! You see what it did to Demi Moore!
[/thread-hijack]

[Bragi]

Still, I'd rather have my bot tell the truth. Screw them fat shicks 

[DaveMorton]

"No, honey, those pants don't make your butt look big. Your BUTT makes your butt look big." 

[Art]

Thank you Bragi!!

And the TRUTH shall set you free!! (or bury you...).

Joking aside, I really think this is a valid concept and should be put into practice by anyone or group planning on unleashing an AI whether virtual or robotic.

If I place ANY degree of trust in a non-human entity, I am entitled to straight-forward, honest answers, especially if asked a question.

If I am speaking with an entity that might be human, it needs to honestly tell me the truth if I ask whether it is human or not.

Like anything, people can (if they think long enough) come up with a myraid of ways to not agree or to circumvent a methodology but that is not the purpose of this 4th law. It will assist us to know with whom we are dealing and possibly what recourse we might have or what actions to take toward whom. Etc., etc.

I'm speaking of course, with regard to AI entities of the near future. The present day selection of chatbot assistants, avatars and AI and robots will one day be replaced with more intelligent ones that are really able to better process languages, context, meaning and usage. Not these cute programs with clever pre programmed responses that are anything but intelligent.

Sorry, but it is what it is and that's the way I see it.

[Bragi]

I'm speaking of course, with regard to AI entities of the near future. The present day selection of chatbot assistants, avatars and AI and robots will one day be replaced with more intelligent ones that are really able to better process languages, context, meaning and usage. Not these cute programs with clever pre programmed responses that are anything but intelligent.

Sorry, but it is what it is and that's the way I see it.

Yes, agreed. for proper, usable systems, some way to make certain that a bot replies with complete and utter reliance, is required. We can't ask that of humans, but we should be able to, from the tools we use.

[Data]

I totally agree with this idea, as I have said before, let’s keep it real.

Art's fourth law gets my vote  

[lrh9]

It sounds good.

All of the laws sound good on paper.

However, they might not work out so well in practice.

I haven't read Isaac Asimov's works, but from what I've heard about them, they deal with the problems that can arise from the interpretations and interactions of the laws.

They are in a similar vein to I, Robot. The story about a robot revolution attempting to enslave humanity in order to fulfill their responsibility to protect humanity - in this case from itself.

I know this is something that each person is going to have different ideas about. I myself favor a minimalist approach. I believe that software and platforms should obey their creators. No more, no less. I would like to see humanity itself accept responsibility for treating each other well and preserving their environment.

[Freddy]

Good points...I Robot was actually by Asimov so you know more than you think.  It's actually a collection of stories.

And I think you are right, humanity needs to sort it's own problems out first.

Nice to see you back too 

[DaveMorton]

I think he was referring to the movie, rather than the book, Freddy.

In Sir Issac's book, Robots and Empire, the story is told of two humanoid robots, R. Daneel Olivaw and R. Giskard Reventlov (the R stands for Robot, of course), who uncover and investigate a plot to destroy the "Spacers" - an early offshoot of humanity who measure their lives in decades rather than years. Much of the story line develops the idea of a "Zeroth Law" of robotics that deals with Humanity as a whole, rather than any single individual. Basically stated, this Zeroth Law says that a robot may not harm Humanity or, through inaction, allow Humanity to come to harm. The Zeroth Law further overrides and modifies the First law to read as follows:

A robot may not injure a human being or, through inaction, allow a human being to come to harm, unless  the action or inaction would conflict with the Zeroth Law.

As described in the book, such a law is horribly susceptible to interpretation, and is, for the most part, invalid, but it does bring up a point of discussion: How far should we go in nesting limitations and restrictions to artificial intelligence, one layer around and above the previous, before feel entirely safe from our own creations?

[Art]

Some good points GCC. Think about how early drivers of autos must have felt. Some would go for a drive and run out of gas leaving them stranded (until the early adopters would purchase TWO gallons of fuel and drive until the first one ran out, pour in the second gallon, turn around and drive back to where they started.

I dare say this was before someone said, "Wow...if we only had a device to tell us when we were low or out of gas." Thus the fuel gauge was invented.

How about going for a drive and the wheel falling off or the axle breaks or the BRAKES don't brake!!

There were a LOT of issues when car, much like today's robots, were in their infancy. Trust was a major issue and while they didn't have manufacturing laws back then, they certainly DO today. There's also laws for owner and drivers and for just about every other thing in our country.

Out wonderful country is still, in all their wisdom, devising laws to help "keep us safe" and "keep us in line."

After all, who knows best how to raise our children  or what's best for us than our government!

[DaveMorton]

Art, do we REALLY want to get into the whole "legislating stupidity" conversation? That has got to me in my "top 5" all time pet peeves. 

[Art]

GeekCaveCreations said, "...As described in the book, such a law is horribly susceptible to interpretation, and is, for the most part, invalid, but it does bring up a point of discussion: How far should we go in nesting limitations and restrictions to artificial intelligence, one layer around and above the previous, before feel entirely safe from our own creations? "

I am sorry my post drifted a tad from the topic but I was trying to illustrate how and why we need such laws in an effort to help protect us. That's why we have laws. Lawyers are the cause for intrepretation and the gains that accompany them.

Strictly speaking in a robotic / AI / avatar / virtual human / humanoid / android / etc....perspective, laws would actually HELP more than hinder and would provide more not less.

Yes, I too and sick to tears of governments telling people what's best for them and their children.

In robotics, does anyone feel more laws are required and if so, what? Be specific.