A Fourth Law of Robotics

Since the "Laws" seem to be growing:

"Rule 4: A Robot shall not construct, or allow the construction of, another robot that is not governed by the Five Laws of Robotics."

"Rule 5: A Robot that identifies another Robot that is not governed by the Five Laws of Robotics shall report its existence to jurisdictional law enforcement authorities."

Good point. If you want your machines to serve you, it's best not to make them too powerful. Once you lose control of autonomous machines (or biological agents as another example), the game is over.

"These laws are meaningless if they can't be enforced....and they can't..."

However, In Asimov's universe, those laws CAN and ARE enforced, INSIDE the Positronic Brains that control the robots. As an example, SE, what is preventing you from, say, dismembering and eating your own children? Not fear of punishment, you know that is a punishable act, but that isn't what's stopping you. What's stopping you, what keeps you from even entertaining that thought in your head, is the deep-down revulsion you feel at the CONCEPT of hurting your own children. The Three Laws are encoded deep inside the Positronic Brain's programming, to the point where the robot physically cannot even THINK of breaking them. That is why, in one story, the computers (which were also Positronic Brains, just without robot bodies) tasked to work out the equations on a new Faster-Than-Light drive kept self-destructing: in the simulations they ran, the human crews 'die,' and the Brains burn themselves out if ordered to continue the simulation past that point⁽¹⁾.

In our world, Laws are lists of restrictions and corresponding punishments for people only, machines and animals are considered in capable of mens ra "Guilty Mind," machines and animals are considered incapable of greed, hate, jealously, etc. that drive people to commit crimes; machines simply follow their programming, animals simply act on their instincts. So the Five Laws⁽²⁾ are more of a mental exercise for people in how we should be designing machines to treat people, and how we should be treating each other, for most science fiction is not really about the science, but about the PEOPLE and how they react and interact with each other because of the new science.

Notes:

1. In the story, it took a special Brain, one designed with the limited understanding of a child, and therefore no concept of 'death' to complete the simulations and build the drive. It turned out that the passengers would 'die' during the journey, but they would 'get better' before the end.
2. "Three, Sir!" "Three."⁽³⁾
3. Darnit, that's the wrong reference. Curse you, Sir Robbin! Get out of my head!⁽⁴⁾
4. Like that would help, given all the OTHER Monty Python references inside my skull.
   

"In our world, Laws are lists of restrictions and corresponding punishments for people only..."

So, what happens when mother nature violates the First Law of Thermodynamics? ;-)

All kidding aside...

"machines and animals are considered incapable of greed, hate, jealously, etc."

Well, I would consider that statement false because dogs clearly demonstrate emotions such as jealousy, fear, anxiety, and spite.

For an animal like a dog, punishing it for those feelings is a bit difficult if not impossible because the animal quickly forgets what it just did within seconds.

That is why when training dogs you have to reward them within one second of the action for them to make the association.

On a broader spectrum, the robotic Laws are simply science fiction. As has been already posted, autonomous machines are being designed now to kill humans. How does anyone expect to enforce such laws on machines?

Any programer can instill or omit such restrictions at will and what organization would be responsible to insure compliance?

Well, I know one person on the CR4 forum that would tell you it had to be the UN.

Lastly, any algorithms that are employed as safeguards would likely take on a much more highly complex set of expressions than simply something like Thou Shall Not Kill. While that might be a very high level requirement, implementing it would be considerably more detailed and the number of possible permeations you would have to test for would be nearly infinite. That would be one huge FMEA!

I'm sure Asimov understood that, but he was writing a fiction novel and as such the simplicity of the Three Laws works just fine for the reader and the story.

"

"machines and animals are considered incapable of greed, hate, jealously, etc."

Well, I would consider that statement false because dogs clearly demonstrate emotions such as jealousy, fear, anxiety, and spite.

"

Perhaps I should have clarified that in the eyes of a court of law, machines and animals are incapable of mens rea. We have a four-footed family member at home, and I am fully aware of the wealth of emotion and even intellect behind that husky's eyes. From her 'I want you to follow me' stare, to her 'I want to be outside' fake-sneeze (I do NOT know why she decided on that signal, but it's consistent and unambiguous. when she walks up to you, makes eye contact and then does her head-shaking fake sneeze, she's going to be racing for the back door when you stand up) to her 'I walked up to you and sat pretty without being coached, now give me a treat before I have to start stealing shoes' pose, she communicates as well as the other housemates, all human.

Yes, it is interesting how dogs signal their owners (pack leaders). ;-)

Wrong catch phrase. His would be "I'll be back".

"Make my day" is:

"Wrong catch phrase. His would be "I'll be back"."

There was also "Come with me if you want to live," spoken in every movie by different characters, and I believe even on Summer-Terminator's first appearance in the TV series.

GP! I remember reading somewhere in the distant past: "These laws, they are useless. The good people don't need them and the bad people don't obey them."

Let us all go forth with new laws for plausibly sentient creatures that yet exist.

Corollary 1: A robot presented with an equivalent rule paradox of options will suffer no damage for inaction. When practical, human guidance should be sought and followed.

Easy. We'll just put up some 'Robot Free Zone' signs.

That idea works so well for guns...

That's a comical response to a very deadly result. I have actually talked to different companies about having the ridiculous sticker on their door advertising a gun free policy. I don't agree with the policy, but find it very dangerous to advertise it.

I have such a sticker on my front door.

It faces and directs its message outward.

I cannot see it from the inside.The message is not for me.

When visiting others with similar policies it is considered good form to comply.

Why is it dangerous to advertise that you want to be the only one in control of any guns in your home/business premises?

Yes, the dog can bite.....

It's not just a sticker, it is enforced.

That restaurant you mentioned did not search its patrons.... that's a process failure.

At any airport there are loads of signs prohibiting firearms yet there are personnel inside and around that are visibly armed and some who are covertly armed.

I don't have a dog sign either. The dog is self annunciating.

Depends where you are.

I don't want to be searched everywhereI go just to feel safe. I don't want to live in a police state. That's insane.

For another thing, the perp that came shooting up the restaurant would just shoot anyone that tried to stop and search him.

The same thing happens in controlled spaces here. Some people just don't care about the signs and the armed security. They go in with guns blazing to kill.

It's the product of a social disease we have cultivated because its socially fashionable to throw out all semblance of moral responsibility over the last 50 years.

It's not a "gun free zone sign", that idea is stupid..... it's "more of a "this is my house" sign....it's a cultural thing.

Some folk get a bit detailed...

I like the no sunglasses, beards, and mustache signs.

However, if someone is coming to burglarize your home, are you expecting them to leave their weapons in their car?

Took me awhile to figure those out but I decided those were ski mask pictures, not beards and mustaches. Unless they are beards and mustaches that cover the nose too.

Basically people who are hiding their faces are banned. Like trick-or-treaters. Good way to save money on candy for Halloween.

They are dust/birdflu masks.

Oh great, so if someone has bird flu they have to take off the mask that was keeping the germs from their coughs and sneezes from entering the general atmosphere?

Great idea "If you're contagious, you MUST spread it around." The days of the 'Pox parties' are long over, I don't want to get sick just because someone else is prevented from taking proper precautions.

Great idea "If you're contagious, you MUST spread it around."

Yes, this way, only the fittest survive. No more allowing the weak to coast through life. And while we're at it, let's drop those silly flu vaccinations....oh hell....let's drop them all.

I don't make these policies I just live with them.....

Most business owners place such signs on their doors because they think it will prevent this:

http://www.nydailynews.com/news/national/video-shows-twin-peaks-waco-biker-gang-shootout-article-1.2416608

Which couldn't be further from the truth.

As AH stated below - most of the people who are going to heed that sign are the ones you're never going to have any problems from anyway and likely the ones that would have come to your aid in a time of crisis.

For me, I conceal carry and the ONLY place I heed those signs are: Banks, Airports, and my civilian employer... kind of ;-); I am armed everywhere else.

But then again, I am also a law enforcement officer in the military (reserve part-time status now), and was one on the civilian-side for a short period. Add that to my multiple combat deployments, and I feel naked without it.

Unless all sentient beings can abide by them too you wasting your time. Human beings are no where near a level in their social behavoir to live by them. Hey we can't even agree on a creator with out coming to blows. Take the defeat of their home team in sports matches.

Can they do any worst then we do to ourselves.

Wow, I didn't even catch that. Good job!

Thanks, kind of embarrassing.

"What happened to Go forth and multiply"

What about "Viruses" that could potentially cause faulty processes?

Are all computer viruses created out of malevolence?

What would you call insanity applied to a robot?

Isaac Asimov obviously never met Capt. Ed Murphy.
Or any number of radical terrorists who are capable of reprogramming/inventing killer robots.

Not applicable for humans at all

Might make a good fiction book.

But this isn't reality.

Uh, which "Law"? Forth (like the programming language) or Fourth, like the number between third and fifth?

Some dingdong has already decided it is OT to care about such matters as spelling.

It does matter, and was already off-set, but I added another for good measure.

If only we could apply the laws to humans as well.........

Sound like secret society of robots to me

Sorry! to establish a new law for robots we need an entire system of robot lawyers.

Aren't lawyers robots already?

I think we are missing a fundamental difference between humans and creatures that will run on a programming construct.

A human decides to or not to follow a given law. A robot will not have the option of making a decision. It is a pure question, or should be, of logic, and his decision engine will have no room for "choice", or at least not at this level of decision making.

IMHO... That needs to be core to their "thinking" processes. These type of decisions need to be done in a robotic mind at a protected level that is outside the range of code sections that may be influenced or modified by any process. Call it what you want, "Command Level Alpha" or whatever. A command issued to the "robot" by a "Command Level Alpha" routine may not and cannot be ignored by the OS of the system. There are ways to make this possible.

This is exactly why we can see that the era where Asimov's Laws could be coded has not yet happened. Look at the First Law:

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

This Law has several critical subjective decisions that must first be made before even attempting to apply this law. What is a human being and how does an AI determine it? What is harm? Does harm have a magnitude, if so then what threshold invokes this law? The algorithms to make these decisions are instinctive traits to our minds but not to any AI program we've been able to create.

Exactly!

The Three Laws are pure science fiction.

Reality is orders of magnitudes more complex than the book.

AI is often misunderstood. People think of AI as a true thinking entity. It is not. It is merely the emulation of thought, not the simulation of thought. That is an important distinction.

Emulated reasoning is a long distance call between human reasoning and computer algorithms. Even strong AI is still an emulation.

We may reach thought emulation in a few decades (at least to some level), but a simulated, self-aware consciousness is another matter altogether. Not saying it will not happen, but simulated consciousness is another level complexity that we simply do not understand yet.

"This Law has several critical subjective decisions that must first be made before even attempting to apply this law. What is a human being and how does an AI determine it? What is harm? Does harm have a magnitude, if so then what threshold invokes this law? The algorithms to make these decisions are instinctive traits to our minds but not to any AI program we've been able to create."

These topics were discussed and agonized over in Asimov's Robot stories. One discussion I remember involved a robot being placed in the Dilemma of the Rail Switch, where leaving the switch alone killed six people, but pulling the switch saved those six and killed a seventh. Bu the raw numbers a robot would pull that switch, as [Action causing one human death] is a better outcome than [Inaction allowing six human deaths], since [6 > 1]. The discussion then brought up the idea that the [one person] was famous or important to society, while the [six people] were all 'common people,' or the [one person] being an old man and the [six people] replaced by a single young child, bringing into the equation the need to determine the 'value' of one life over another, and whether a robot should be allowed to make that subjective judgement, or even if it was CAPABLE of understanding the concept.

People talk about the Three Laws a lot, but without knowing the full context. It would be like me arguing the merits and/or shortcomings of the newly elected Canadian Prime Minister, when I know NOTHING about the man⁽¹⁾.

I would Strongly Suggest that people read Asimov's Robot stories⁽²⁾ before debating the Three Laws, since much of what is brought up has already been dealt with at length by the characters within the pages.

Notes:

1. Okay, I know he's male, and white from mostly if not exclusively European stock, and he has black hair kept 'business short,' and prefers to go clean shaven, but that's all from a photo. I don't even remember the guy's NAME, for crying in the corn, or which Party he's from, I just know he defeated the Conservative PM who's been in charge for a decade, or was it 14 years? See, I have NOTHING intelligent to say about the recent Canadian Election, so I would 'keep my own counsel' and sit quietly, listening to more knowledgeable people to fill the gaps in my understanding.
2. Yes, I mean ALL of the stories. I KNOW the collected book is a doorstopper, it's thick enough to stop a Sabot round from an Abrams Tank Gun. But if you read the stories, you'll see than many of the questions brought up about the Three Laws are already 'Asked and Answered,' and you'll see more into the insights of not only how Man and Machine interact, but also how Man treats Man. Asimov very seldom wrote for pure entertainment, there's often a hidden lesson within his prose.
   

I love Asimov's work.

I read all of the Robot short stories, the Foundation series, and many others whose titles elude me right now. I've even used his two Thiotimoline reports as templates for scientific papers. I agree that these dated and yet timeless stories should be read by more people.

You're also correct that Isaac explored in much more depth many of the other laws the robotologists in his universe tried. The three laws were the only laws common to all robots.

I'm in the middle of decorating at home which meant moving my books & bookcases into the shed for a while. I was re-filling one bookcase last week & realised that I have almost a complete shelf of Asimov books. There's the Foundation trilogy plus prequals & sequals, short stories and much more. Isaac was a prolific writer & I didn't know just how much I have read.

In an interview of Isaac I learned that he considered the universe of his fiction to be a consistent but not contiguous timeline of the future of humanity. In other words, the people in a story set 100,000 years in the future were descendants of people in the stories set 50,000, 1,000, and 50 years in the future. There were no alternate timelines, or other independent worlds.

A hidden conclusion he had in his vast prediction of the distant future is that robots will not exist forever with humanity.

The way "moral" humans get around exterminating other people is to simply redefine their enemies as "not human."

Germany: See those Jews? Not human.

Robots: See those people over there? They are actually robots like you, but they look and act exactly like humans. You can tell which ones are not human because I will tell you which ones are not human.

"The way "moral" humans get around exterminating other people is to simply redefine their enemies as "not human.""

You, sir, win a genuine, hand-crafted, gold-plated NoPrize for succinctly summarizing the overarching lesson of the Robot stories: "How and why do we define some as 'people' and some as 'not-people,' and why haven't we, as a society, moved beyond this primitive, caveman-esque 'us vs. them' mentality?"

I don't know what a NoPrize is, but if it's gold-plated, it must be good.

Thanks.

A NoPrize, as many a Marvel Comics fan would tell you, is exactly what it says on the tin, "No Prize," just a verbal (or print) pat on the back. Often handed out in the letters page of a comic by Stan Lee to someone who wrote a letter to the editors pointing out a 'plot hole' in a recent issue, as well as providing some explanation that could 'seal' the hole.

Seems like a gold-plated NoPrize pat on the back might really hurt. Do I have to immerse myself in a plating bath? Are you sure it's genuine?

"Overclocking" is to Robots as steroids are to athletes.