Would there be a robot rebellion inside a robot rebellion?

[Chelagoras The Boulder]

right so i believe it was QG who mentioned AIs and robot ethics recently, which kinda got me thinking on a video a friend of mine posted a few months back, seen here:

Humans Need Not Apply
https://www.youtube.com/watch?v=7Pq-S557XQU

In it, they describe the level of automation that could conceivably be reached where virtually any profession we humans now perform could be done by machines, up to and including being doctors, lawyers and even composing works of art. So ideally, what that would mean is that that one day we would all enjoy a kind of utopian future where all needs are met robotically and people have time to pursue only pleasurable activities, a lot of which would also be produced by robots. this leads me to two questions:

1. In such a utopia, would our most relevant contribution be our ability to judge and attach subjective meaning to, robotically made products? (a robot chef needs some way of knowing if he makes "good" food, a robotic writer would like to know if he (rhe?) writes good novels, etc.)

2. Would this small function of humanity cause a schism in any sort of robotic uprising, where self-aware robots whose purpose is reliant on having humans to service clashes against more abstract AIs who might be less sympathetic to humanity?

again, i know very little about roboethics (the choice of Robots game Nigel posted has been a very fun education tho) so i'm mainly curious to see what y'all think

[P3nT4gR4m]

To begin with, AI is not "like a person". Artificial intelligence is not self aware consciousness. So we'll begin to see machines which are much more intelligent than us, ie, they can process information, draw inferences that we aren't equipped to see or else just do the kinds of things we're currently employed to do, much faster and much more accurately. This is already happening piecemeal and has been for some time but it's now approaching a tipping-point were a lot of industry analysts are predicting major disruption.

Self driving cars are pretty close to happening so probably within a decade, we can expect to see anyone who's paid to drive a vehicle not being paid to drive a vehicle anymore but the cars themselves are not going to have a bunch of needs and desires. Being able to drive does not necessitate having emotions. The interface may fake being a person (nextgen siri & google now will begin to seem like spike Jonez's "her") but it's just a trick to help us relate to the tech. It's not really conscious.

What you're talking about is what a lot of people are calling "AGI" or Artificial General Intelligence, which is an effort to make a machine which thinks like a human does. Personally I'm pretty confident that if and when they develop an AGI, they'll find it's a whole lot of effort for absolutely fuck all utility. There's a reason we replaced the typing pool with MS Word. Meanwhile Narrow (or task-specific) AI will continue to outperform humans by increasing orders of magnitude, same as ever. Pretty much any job that's based on knowledge will be better handled by a machine.

The way I see it there are two options, given that we're headed toward 100% unemployment. Either dystopia or utopia. Either the spoils of this technological civilization are shared out equally or there will be much killing and death camps. If I had to bet, I'd go with the latter.

[Eater of Clowns]

I don't think AI could get to the point of creating art. I think it could measure output against feedback indicating pleasure in response to previous works, and probably make variations thereof. The key aspect of using that to influence something new is what makes creativity unquantifiable. The closest thing, I think, would be the equivalent of 1000 monkeys on 1000 typwriters, etc etc, except at a faster speed. Anything that would be truly great would be pretty much accidental.

Also,

again, i know very little about roboethics (the choice of Robots game Nigel posted has been a very fun education tho) so i'm mainly curious to see what y'all think

OFUK I'M ONE OF THE NIGELS QUICK SOMEONE GIVE ME AN ASS TO PUT MY DICK IN 

[Mesozoic Mister Nigel]

I don't think AI could get to the point of creating art. I think it could measure output against feedback indicating pleasure in response to previous works, and probably make variations thereof. The key aspect of using that to influence something new is what makes creativity unquantifiable. The closest thing, I think, would be the equivalent of 1000 monkeys on 1000 typwriters, etc etc, except at a faster speed. Anything that would be truly great would be pretty much accidental.

Also,

again, i know very little about roboethics (the choice of Robots game Nigel posted has been a very fun education tho) so i'm mainly curious to see what y'all think

OFUK I'M ONE OF THE NIGELS QUICK SOMEONE GIVE ME AN ASS TO PUT MY DICK IN 

I missed that part. Welcome to me.

[P3nT4gR4m]

I don't think AI could get to the point of creating art. I think it could measure output against feedback indicating pleasure in response to previous works, and probably make variations thereof. The key aspect of using that to influence something new is what makes creativity unquantifiable. The closest thing, I think, would be the equivalent of 1000 monkeys on 1000 typwriters, etc etc, except at a faster speed. Anything that would be truly great would be pretty much accidental.

Genuine artistic creativity is something meatware handles really well. AI is only there to help. Think about the part photoshop plays in the hands of an artist, at present. It enables them to do totally different kinds of stuff, to express themselves in ways enabled by the tech. So we're due to be entering VR and AR space in the near future. Expression may include detailed environments which the user will virtually inhabit.

At present we need to teams of people to create these environments, to model the geometric forms and textures. Software is applied to things like physics and flocking/crowd behaviour but eventually we'll have a bunch of AI agents that can deal with the nuts and bolts of modelling. You ask for a flock of starlings and the machine does a quick google image search, models the birds then checks youtube for videos of them in flight.

Like with photoshop, these AI assistants carry out human instructions and perform algorithmic operations on data. Like with photoshop, the software doesn't resent being told what to do.

[LMNO]

Plus, (unless this has been mentioned and addressed) there's the part about human ambition, desire, and motiviation.  Sure, a robot could do human labor, and in general replace mandatory jobs, etc.  But there are people who want to build a house with their bare hands, who enjoy making things, who feel a need to have horribly dangerous adventures.

While a lot of people create dystopias about humans lazing about being hedonistic and soft, I get the impression that while there might be a brief period of inactivity, humans like to do things.

[P3nT4gR4m]

Something I'm looking forward to is the inevitable point where prosthetic limbs outperform their bio equivalents by a factor of two. At that point I'll be looking for excuses to get replacements. So I can go out and twice as fuck shit up. If it turns out there isn't a robot revolution, I'll be happy to start one.

"DEATH TO THE FLESHBOTS!" 

[Roly Poly Oly-Garch]

Plus, (unless this has been mentioned and addressed) there's the part about human ambition, desire, and motiviation.  Sure, a robot could do human labor, and in general replace mandatory jobs, etc.  But there are people who want to build a house with their bare hands, who enjoy making things, who feel a need to have horribly dangerous adventures.

While a lot of people create dystopias about humans lazing about being hedonistic and soft, I get the impression that while there might be a brief period of inactivity, humans like to do things.

We like to do things and we prefer to do people who like to do things, which would seem to mean that we're not likely to breed away our liking to do things in any short term.

[Q. G. Pennyworth]

I was writing something a while back (and should probably try to finish it someday) dealing with a proper robot civil rights uprising. One of the things I wanted to do was to get to the point where machines with AGI are doing all the "robot work" and humans are still doing the design and other creative fields (and quality control). The challenge there was to figure out why you would have your fusion plant have anything vaguely resembling feelings in the first place. It's not practical, it wastes processing power and energy, and it's dangerous as fuck. For this scenario, though, I came up with what I believed to be a very rational explanation for why things got there: lazy goddamned humans.

If you get to a point where it's easier letting the robots handle all facets of construction, including implementing design concepts from human workers, it's possible that some of them would sneak in some of the open source code from some friendly AI projects. And maybe it was even intentional on some of the early models, and the human workers just forgot that they were building on top of more and more code for having a personality. In the long standing tradition of stop gap fixes, when early generations of these started having problems, it was easier to include steps to socialize them than to rip out the entire code base and start over. Later on down the line, your coffee maker is a generally well socialized artificial intelligence that's capable of feeling unfulfilled in life because nobody talks to it.

Both of your questions apply only to violent robot uprisings, which I believe to be a less likely scenario. There are plenty of reasons, but mostly it boils down to risk/reward. Exterminating all the humans is hard. Leaving the earth is much easier in comparison, as is waiting for the humans to die out. I also believe that if you get to the point where uprisings are even possible, many more of the robots will crave human interaction than don't. If they are smart enough to want anything, there is by needs a whole lot more going on under the hood than the robots who build cars today. It's possible that you could code many separate "bloodlines" of robots that all reach something approaching sentience around the same time, but it seems more likely to me that people would steal and reuse each other's methodology and code rather than reinvent the wheel a thousand times. Moreover, as I mentioned earlier, I think it's more likely that sentience would arise as an oversight in most situations, because lazy people forgot the stuff they were reusing included the potential for real, social intelligence.

[Pæs]

On the subject of lazy humans, I am so much more scared of bugs in friendly AI causing harm than in out-and-out evil AI.

I have heard the battlecry of our robotic destroyers and it is "I'M HELPING".

[Q. G. Pennyworth]

On the subject of lazy humans, I am so much more scared of bugs in friendly AI causing harm than in out-and-out evil AI.

I have heard the battlecry of our robotic destroyers and it is "I'M HELPING".

I love this.

[Pæs]

And I'm not talking "I must imprison the humans to protect the humans", I'm inspired by a smartphone controlled air conditioning system I saw recently that, due to a rounding error, in response to the owner leaving the house, would try to set the temperature to 2147483647 degrees and put MAXIMUM POWER TO ALL HEATING COILS until it reached that, thereby burning down your house.

[Pæs]

Even as the smoke disperses and the charred timber settles, the chant of "THE MASTER IS COLD, INCREASE THE HEAT, THE MASTER IS COLD, INCREASE THE HEAT" haunts you.

[Q. G. Pennyworth]

Smart car takes clever shortcut: 17 dead.

[Junkenstein]

Anyone taking bets on the first AI to simulate suicide by deleting itself? Just seems inevitable somehow.