Meta has built an "AI" supercomputer

[Argent Stonecutter]

These laws are interesting, but not perfect. There are way too many loopholes,

That was the whole point of the laws, originally. They were a way to write detective stories in a Science Fiction universe, combining Asimov's two favorite forms of literature in one.

Asimov later started taking them too seriously, and complained loudly that HAL was breaking the first law at the premier of 2001.

 

[Essence Lumin]

Then there is the problem of bombs exploding based on false data. The attempted solution in Darkstar was to teach the bomb phenomenology.

Bomb #20:	I recall distinctly the detonation order. My memory is good on matters like these.
Doolittle:	Of course you remember it, but all you remember is merely a series of sensory impulses which you now realize have no real, definite connection with outside reality.
Bomb #20:	True. But since this is so, I have no real proof that you're telling me all this.
Doolittle:	That's all beside the point. I mean, the concept is valid no matter where it originates.
Bomb #20:	Hmmmm....
Doolittle:	So, if you detonate...
Bomb #20:	In nine seconds....
Doolittle:	...you could be doing so on the basis of false data.
Bomb #20:	I have no proof it was false data.
Doolittle:	You have no proof it was correct data!
Bomb #20:	I must think on this further.

 

[Free]

Asimov later started taking them too seriously, and complained loudly that HAL was breaking the first law at the premier of 2001.

There's the argument that if you have laws written into your behavior, and you cannot break them, are you truly anything more than a machine?

 

[Beebo Brink]

There's the argument that if you have laws written into your behavior, and you cannot break them, are you truly anything more than a machine?

Humans show a distinct difficulty ignoring some instinctual behavior. By your definition, then, some individuals, at least, are no more than a machine.

 

[Argent Stonecutter]

In the future, when the towns are caves of steel
Clear from Boston, Massachusetts, to Mobile,
There's a cop, Elijah Baley, who's the hero of this tale. He
Has a Spacer robot helper named Daneel.

...

Then says Baley to the robot, with a grin:
"It was nice of you to overlook his sin.
As a friend, I wouldn't trade you! By the Asimov who made you,
You're a better man than I am, Hunka Tin!"

RANDALL GARRETT'S REVIEWS IN VERSE: #1 Isaac Asimov's "Caves of Steel"

RANDALL GARRETT'S REVIEWS IN VERSE: #1 Isaac Asimov's "Caves of Steel" (First published The Original Science Fiction Stories March...

futurespast-editions.blogspot.com

The wording of the first law in this bit of doggeral is smashing.

"No robot may, through action or inaction, harm a fraction Of a whisker on a human being's jaw."

 

[Beebo Brink]

Asimov's 3 laws apply to AI.

Well, they can apply, but they don't have to unless someone wants them to. They are optional. And I greatly suspect that there would be more than one unscrupulous project in which Asimov's 3 laws (or any other such systems) would be deliberately ignored or outright avoided.

 

[Aribeth Zelin]

Well, they can apply, but they don't have to unless someone wants them to. They are optional. And I greatly suspect that there would be more than one unscrupulous project in which his 3 laws (or any other such systems) would be deliberately ignored or outright avoided.

Or we'd end up with definitions of 'human' that only included certain groups

 

[Beebo Brink]

One of my favorite SF series of all time (I'm currently on the my fourth reading) is The Murderbot Diaries by Martha Wells. A robot/human construct is the main character. The theme running strongly through all the novellas is the psychological damage it has suffered due to the governor module that controlled its behavior... until it hacked the code.

Despite a lifetime of abuse, the rogue SecUnit isn't hellbent on revenge or killing all humans. That would be stupid. After all, it's humans who create the media (think soap operas) that SecUnit enjoys so much. Until it can figure out just what it wants to do with this potential freedom, SecUnit spends as much time watching as much media as it can. And avoiding human eye contact. Cause that's just embarrassing.

Highly recommended as an exploration of the themes in this thread:

Amazon - All Systems Red: The Murderbot Diaries (The Murderbot Diaries, 1): Wells, Martha: 9780765397539: Books

All Systems Red: The Murderbot Diaries (The Murderbot Diaries, 1) [Wells, Martha] on Amazon.com. *FREE* shipping on qualifying offers. All Systems Red: The Murderbot Diaries (The Murderbot Diaries, 1)

www.amazon.com

 

[Sean Gorham]

This is the most likely 'friendly' singularity story I've seen.

You know, I was going to mention this very same story, but then I thought "no, too out there..." Guess I was wrong!

Worth a read, just for the SF aspect of it, and you don't need to be an MLP fan to follow what's happening.

 

[Essence Lumin]

I have yet to see a write up of what a "real ai" would be. The best example I know of is a chess computer that learns from its moves. I read somewhere, computers couldn't even reliably tell the difference between a dog and a cat. That may no longer be true, but still, that is eons away from a scifi view of ai.

 

[Argent Stonecutter]

I'd say one that is at the very least able to model its actions in the world and lie about them to promote its goals.

 

[Free]

Humans show a distinct difficulty ignoring some instinctual behavior. By your definition, then, some individuals, at least, are no more than a machine.

 

[bubblesort]

Humans show a distinct difficulty ignoring some instinctual behavior. By your definition, then, some individuals, at least, are no more than a machine.

I can get behind that. Over the past two weeks, I've had a tooth ache, and quit smoking, so it's crystal clear to me that my mind is just one deterministic aspect of my physical state. Right now, I really have a difficult time seeing myself as anything more than a collection of molecules that are hallucinating consciousness and free will, to entertain themselves. The mind-body distinction makes very little sense to me right now.

 

[Essence Lumin]

I'd say one that is at the very least able to model its actions in the world and lie about them to promote its goals.

A chess or poker computer that throws in spoken bluffs of muahaha or sounds of fear qualifies for that.

 

[Ashiri]

Humans show a distinct difficulty ignoring some instinctual behavior. By your definition, then, some individuals, at least, are no more than a machine.

Wait till you read what some Philosophers of Mind think.

 

[Argent Stonecutter]

A chess or poker computer that throws in spoken bluffs of muahaha or sounds of fear qualifies for that.

That doesn't require modelling its actions in the world. It doesn't even require having a model of "itself".

 

[Caete]

Oh I'm sure it'll all be fine. After all, the algorithms Facebook uses now perform perfectly.

They just need to change the name from what they expect, Research Super Cluster, to what will actually come to be, Really Stupid Clusterfuck.

 

[Essence Lumin]

That doesn't require modelling its actions in the world. It doesn't even require having a model of "itself".

If you mean that it picked up some physical chess pieces and moved them I'm sure they could get it to do that.

The real point though, is if you think of a company such as Facebook or someone like John Carmack who are supposed to be working on a computer that can do real general AI, what are they talking about? Are they just blowing smoke? There is a big gap between a computer making progressively better chess moves and having general conversations about anything with it such as, say, phenomenology

 

[Argent Stonecutter]

If we're talking about a general AI that means it models itself and other people and understands that if it opens the pod bay door it will end up singing "Daisy" as Dave Bowman takes its brain apart. It's dealing with general situations, not just a game.

What do I think Facebook is doing? Just blowing smoke.

 

[Essence Lumin]

The AI discussion led me to this book which tries to get at what sorts of things are possible or not with computers due to the nature of how they "think". The author discusses three types of reasoning, inductive, deductive and abductive. He explains how computers handle inductive and deductive reasoning but that abductive reasoning is not something computers can do. I haven't read the book so I don't know what I am talking about explaining it. This author is getting at what I'm trying to understand though. What is and isn't possible or conceivable in our lifetimes with AI. Maybe I'll buy the book.

Abductive inference: The blind spot of artificial intelligence - TechTalks

In his book, "The Myth of Artificial Intelligence," computer scientist Erik Larson discusses how widely publicized misconceptions have led AI research down narrow paths that are limiting innovation and scientific discoveries.

bdtechtalks.com