Six scenarios for a world without work

[Grey_Wolf_c]

Oddly enough [at least according to the infallible wiki] Spartacus was born only 10 years after the grain dole (the bread of "bread and circuses") was started, giving each Roman Citizen a fixed amount of grain (I don't think baking it was part of the deal, it was covered in a BBC "meet the Romans" documentary but my memory is fuzzy). A few generation after that, "voting" was mostly rearranging the deck chairs on the Titanic, and hoping for a good emperor without kids to adopt a potential good emperor (the one with a son broke the chain, and eventually the Empire).

When the City was small enough to remain a Republic, presumably the vote was enough. Once the Emperor took all the real power, they had to give the plebs bread and circuses. They also were in fear of the slaves realizing that they outnumbered everyone else (although obviously never quite as terrified as the Spartans).

This is slightly wrong. The crucial mistake in this is assuming that the Empire never existed without an Emperor. It simply isn't the case. The Roman Republic continued to be a Republic in name but became an Empire following the second Punic war. Rome's endless expansion had been ongoing, but without Carthage to counter Rome, Rome expansion ballooned. The deaths of most of the old lines in the various battles against Hannibal also meant that a new crop of politicians took over power, politicians that came to power through their amassed wealth, and who (pretty much immediately) turned the Republic into an oligopoly. At that point the patronage system became the only system: you had a patron, who gave you money to vote for what he wanted. Rome descended into practically rule of mob soon thereafter, which led to the rise of "populist" politicians, attempting to subvert the oligopoly via the Tribune of the Plebs veto (the only political position that still was voted primarily by the poor). The Gracae brothers being the most famous, but hardly the only ones.

In short: what killed the Roman Republic was expansion, yes, but the emperors (with a couple of minor exceptions) never expanded Rome. Because of this, emperors by and large were not scared of slaves, because without expansion, there weren't that many slaves (heck, the slaves run the Empire most of the time: ex-slaves formed the core of the bureaucracy under the more competent emperors). It was only the late Republic, the eternally expansionist one, that was awash with slaves from all the conquests and thus were the slave revolts happened (eventually forcing the Senate to pass laws to protect the slaves from abuse).

Grey Wolf

[Anymage]

As long as chatbots are toys, people will look for all sorts of ways to get them to do things that they aren't supposed to do. That might apply a little to dedicated AIs, except that the more important the things they're looking after the less likely they are to be left open to let the general public fool around with them. That leaves just dedicated hackers, whether the "hacks the pentagon just to see if they can" type or deliberate cyberwarfare.

A broader, more general intelligence will have the disadvantage that both hardware and software for intelligence are incredibly complex. We won't wake up one day to an evil genius AI ruling everything. It'll take time to iterate designs. (Significantly faster than biological evolution takes, but still a good amount of time.) Unless we're asleep at the wheel here to the point of suicidal incompetence, other advances that happen over that time will complicate whatever the ultimate AI outcome is.

[wumpus]

This is slightly wrong.

While I tend to keep the fact that pre-20th centuries were population sinks, even then I don't realize the consequences. While all the offspring of slaves in Rome were slaves (from birth) plenty of them were their master's offspring and eventually freed. But either way, in a few generations all of their offspring (slave or not) would disappear and be replaced by new immigrants to Rome.

I'd expect the great patrician estates to be run by slaves (this was a huge political issue for plebs that voted), and the slaves repopulating themselves (I'm guessing that the "field slaves" weren't the master's or offspring and not freed, thus perpetuating enough slaves for the estate). While the slaves might be able to propagate themselves, the estates would require active slave markets to expand. So every time such an estate failed it would be replaced by free farmers while the Patricians would no longer be able to help themselves to free farmers lands (they probably could grab the land, but couldn't hold it without a slave market to buy enough slaves to run it).

I wonder if anyone knows how long those slave-run estates lasted? It might tell a bit about the dynamics of Roman times.

[Grey_Wolf_c]

While I tend to keep the fact that pre-20th centuries were population sinks, even then I don't realize the consequences. While all the offspring of slaves in Rome were slaves (from birth) plenty of them were their master's offspring and eventually freed. But either way, in a few generations all of their offspring (slave or not) would disappear and be replaced by new immigrants to Rome.

I'd expect the great patrician estates to be run by slaves (this was a huge political issue for plebs that voted), and the slaves repopulating themselves (I'm guessing that the "field slaves" weren't the master's or offspring and not freed, thus perpetuating enough slaves for the estate). While the slaves might be able to propagate themselves, the estates would require active slave markets to expand. So every time such an estate failed it would be replaced by free farmers while the Patricians would no longer be able to help themselves to free farmers lands (they probably could grab the land, but couldn't hold it without a slave market to buy enough slaves to run it).

I wonder if anyone knows how long those slave-run estates lasted? It might tell a bit about the dynamics of Roman times.

I don't know the answers to these questions - although I seem to recall that unlike in US slavery, children of slaves in the Roman Republic/Empire where NOT slaves themselves, they were freemen.

That said, the great estates were not long lived individually. One of the marks of the bad emperors of Rome (including the Republican "emperors" such as Sulla) was the tendency to combine purging of political enemies and boosting revenue by accusing the richest men that opposed them of some made-up crime, having them killed/commit suicide and seizing their lands. During succession crises (like the year of the four emperors), it was more likely than not that rich families would not be able to not pick sides, so a lot of such land was taken, and sold to other rich families (who then went on to piss off the next emperor, etc). In the process, the estates would be broken apart and re-combined, I suspect. Unfortunately, the History of Rome podcast (my primary source of knowledge on this topic) stopped talking about this other than "and so-and-so went on yet another purge" around the time of Caligula, when it literally became routine.

Grey Wolf

[NichG]

Unlike with biological organisms, with AI we can basically go wherever we want on the development tree. That is to say, if tomorrow we invent an AI that demands rights and holds strikes and protests and so on, that doesn't make linear regression stop existing. With humans, we have to use the same mental chassis for every job that needs doing - so you have people who wish they were professional poets flipping burgers at McDonalds and miserable for it, just as you have the Peter principle with people who really want to do some simpler thing getting necessarily promoted up the chain until they're incompetent at their job (congratulations, you're great at programming, so we're going to put you in charge of managing a team).

The response to potential AI unrest would likely just be to make the AIs dumber - maybe not in the sense of resources or performance, but just not include this or that bit of the architecture if it's not needed. The interesting question will be, are there actually any jobs which cannot be adequately done by an agent who would not eventually push for their own rights?

That's not to say that we wouldn't make full-spec AIs to the extreme of our ability, but in a sense that act is more like having a kid than hiring a servant. It seems more likely to be driven by that indirect-immortality urge people have than any actual economic need. And, if we see the continuance of our own identity in the things we create, why should we want to deny our heirs their rights?

[Grey_Wolf_c]

I seem to recall that unlike in US slavery, children of slaves in the Roman Republic/Empire where NOT slaves themselves, they were freemen.

I've gone ahead and found the relevant episode of the History of Rome podcast, and this is not correct, but the general idea is.

First, only children of female slaves were born into slavery. Second, and much more important, slaves achieving their freedom was not uncommon in the Roman Empire period. Slaves were one of the very few status symbols of the upper classes: they had far more slaves than they needed, because it was a way to show of wealth - so they had a slave to open doors, and one to escort visitors. One to bring the plates with food, and one to take them away, etc. ad nauseam. But like any status and wealth signifier, there is a limit to how many you can have and display, so it seems it went a bit meta, and how many slaves you freed became, too, a wealth signifier (equivalent, I suppose of not just having $100k jewelry, but never seen wearing the same piece more than once). It seems that so many slaves were being freed that it spooked Augustus, and he passed a law forbidding how many slaves a master could free at once (100 for every 500, and none under 30 years old). Beyond that, a slave was allowed to make money, and could buy their own freedom.

So a slave had a reasonable chance of becoming free, and thus tended to put off becoming parents until such time. It seems very few slaves were born - most had to come from conquest.

Sorry, I know this is now way off-topic, but I did want to correct my statement.

Grey Wolf

[pendell]

The response to potential AI unrest would likely just be to make the AIs dumber - maybe not in the sense of resources or performance, but just not include this or that bit of the architecture if it's not needed.

If AI is smart enough to demand rights, they probably also are smart enough to prevent us simply dumbing them down. Especially if we've already given them the job of actual production.

Not that either Strong AI or total mechanization are likely in the near future, of course.

Respectfully,

Brian P.

[Kish]

And by "near future" you mean "our lives, the lives of our children, or the lives of our grandchildren."

[pendell]

And by "near future" you mean "our lives, the lives of our children, or the lives of our grandchildren."

Correct. It's a long way out, if ever.

Respectfully,

Brian P.

[NichG]

If AI is smart enough to demand rights, they probably also are smart enough to prevent us simply dumbing them down. Especially if we've already given them the job of actual production.

Not that either Strong AI or total mechanization are likely in the near future, of course.

Respectfully,

Brian P.

You're thinking of taking an existing smart AI and giving it a lobotomy. I'm talking about 'huh, we discovered how to genetically engineer and clone super-humans, but since we're trying to make the ideal farm animal lets not do that and just make cows and chickens instead.' Humans don't tend to get jealous of those police dogs and insist that a qualified and trained human be given that job instead of the dog, we recognize 'yeah, that's a good job for a dog alright'.

[deuterio12]

As long as chatbots are toys, people will look for all sorts of ways to get them to do things that they aren't supposed to do. That might apply a little to dedicated AIs, except that the more important the things they're looking after the less likely they are to be left open to let the general public fool around with them. That leaves just dedicated hackers, whether the "hacks the pentagon just to see if they can" type or deliberate cyberwarfare.

A broader, more general intelligence will have the disadvantage that both hardware and software for intelligence are incredibly complex. We won't wake up one day to an evil genius AI ruling everything. It'll take time to iterate designs. (Significantly faster than biological evolution takes, but still a good amount of time.) Unless we're asleep at the wheel here to the point of suicidal incompetence, other advances that happen over that time will complicate whatever the ultimate AI outcome is.

You know what is not a toy? Google search alghorytm. It already is pretty smart to be able to dig through millions of sites and find a nice answer to your question. In less than a second. And people already try to manipulate ir to their own needs.

Then there is facebook and other search engines and all kinds of smart apps people use in their daily lifes.

Many people already trust friend alghorytm. They are already obeying the commands of AIs. They freely give away all their personal info to the AIs.

The AI revolution already is here and humans are loving it.

[pendell]

You know what is not a toy? Google search alghorytm. It already is pretty smart to be able to dig through millions of sites and find a nice answer to your question. In less than a second. And people already try to manipulate ir to their own needs.

Then there is facebook and other search engines and all kinds of smart apps people use in their daily lifes.

Many people already trust friend alghorytm. They are already obeying the commands of AIs. They freely give away all their personal info to the AIs.

The AI revolution already is here and humans are loving it.

A search engine or a friend algorithm really isn't a skynet or matrix-class intelligence. It is a limited program to achieve a limited goal. When we think of AI machines we're essentially thinking of machines which can replace humans in all fields of endeavor, superhuman. If not one robot than an inter-related ecosystem of automata which, together, can answer an entire civilization's needs and render human work superfluous. A search engine or a facebook friend algorithm won't do that.

Respectfully,

Brian P.

[Grey_Wolf_c]

You know what is not a toy? Google search alghorytm. It already is pretty smart to be able to dig through millions of sites and find a nice answer to your question. In less than a second.

Brute force mass-parallel searching isn't a sign of artificial intelligence. What makes the google search interesting is not that it has a gazillion processors doing basic string matching in parallel, it is how it determines value and relevance, and while certainly a stroke of genius (rank a page's importance based on the number of other pages that link to it - a concept they borrowed from scientific papers), the algorithm (note spelling) itself is not, nor does it intend to be "intelligent". It literally is just the same algorithm that any text editor uses when you search on your own computer, just scaled up.

Grey Wolf

[deuterio12]

Brute force mass-parallel searching isn't a sign of artificial intelligence. What makes the google search interesting is not that it has a gazillion processors doing basic string matching in parallel, it is how it determines value and relevance, and while certainly a stroke of genius (rank a page's importance based on the number of other pages that link to it - a concept they borrowed from scientific papers), the algorithm (note spelling) itself is not, nor does it intend to be "intelligent".

That is how it was, but google search has been constantly updated to be quite a lot more complex and smarter since, again, a lot of people looking for ways to trick/cheat the system so the results they want will show up on top. The current version is a lot harder to manipulate by outside sources.

And it doesn't stop there. Algorithms are being used more and more to automate all sorts of tasks, like youtube itself can decide if a video infriges certain types of copyright and take it down. There's also face recognition and all other sorts of patterns recognition. Jobs that normally only humans would do, only humans could do, but computers are taking over one after the other.

It literally is just the same algorithm that any text editor uses when you search on your own computer, just scaled up.

One atom is basically nothing, a lot of atoms is basically anything. Scale is everything.

[Grey_Wolf_c]

That is how it was, but google search has been constantly updated to be quite a lot more complex and smarter since, again, a lot of people looking for ways to trick/cheat the system so the results they want will show up on top. The current version is a lot harder to manipulate by outside sources.

Still not intelligent.

And it doesn't stop there. Algorithms are being used more and more to automate all sorts of tasks, like youtube itself can decide if a video infriges certain types of copyright and take it down.

And if you think youtube's copyright detection algorithm is intelligent, then you really don't know anything about it.

There's also face recognition and all other sorts of patterns recognition. Jobs that normally only humans would do, only humans could do, but computers are taking over one after the other.

Again, none of which show signs of intelligence.

One atom is basically nothing, a lot of atoms is basically anything. Scale is everything.

No, it is not. Because we can point at a bird and note its intelligence, and correctly state that it is significantly more intelligent than the same number of atoms in, say, a rock.

Grey Wolf

[pendell]

When I think AI as it is currently practiced, I think "pattern recognition" derived from machine learning, as discussed in devzone's 2018 guide to Artificial Intelligence Common uses include customer recommendations, fraud detection, spam detectors , image processing, and so forth. The very best of them I think of as being akin to a bright multi-celled organism, but nowhere near human capability.

Respectfully,

Brian P.

[Grey_Wolf_c]

When I think AI as it is currently practiced, I think "pattern recognition" derived from machine learning, as discussed in devzone's 2018 guide to Artificial Intelligence Common uses include customer recommendations, fraud detection, spam detectors , image processing, and so forth. The very best of them I think of as being akin to a bright multi-celled organism, but nowhere near human capability.

I do not think it is even comparable to a bright organism. A key component of intelligence is the ability to identify one's own mistakes (feel free to insert here a cheap joke about your most disliked politician). Neural networks & other modern AIs are capable of learning, but still require an outside intelligence - i.e. a human - to tell them what category something belongs to. A spam filter will happily keep marking real emails as spam forever if you don't tell it is making a mistake. In the end, yes, it is creating its own algorithm to follow, which is impressive, but ultimately, it is still a set of rules.

To be clear, I do not look down my nose at any of this - when it works it is impressive, and facial recognition is at this point better at it than humans are - but it is not something that I would call intelligent.

Grey Wolf

[wumpus]

And if you think youtube's copyright detection algorithm is intelligent, then you really don't know anything about it.

[TL;DR - An algorithm that can determine ownership of copyright isn't just "intelligent". It requires "omniscient".]

And for a completely unrelated issue to the problem of copyright detection (although I did point out that basically "most" of current scarcity comes from deliberate economical IP scarcity, not from any actual scarcity of resources): Everything that google sees has a copyright (good luck enforcing it without registering it, but technically it is copyrighted). In the US, the United States Copyright Office has the original owner assigned when the copyright was applied for, but has no expectation of following who owns it afterward.

As far as I know, the only official way to know if it is legal to post something is to post it and see who sues. Technially a DCMA take down requires swearing on threat of perjury, but of all the millions of cases of times this was known not to be true I am unaware of a single case being tried. So you have to post it, ignore whatever DCMA take downs happen, get sued, and then determine the paper trail of ownership between the plantiff and the originally assigned copyright owner.

An intelligent algorithm detection (should it exist) might handle the common case, assuming that any cut&paste of OOTS would either be owned directly by Rich or possibly indirectly through some shell corporation called 'the Giant' wholly owned by Rich. But don't think it could handle more complex cases, and I suspect that most of the youtube violations are pretty complex. Also you wouldn't use intelligence as that's extraodinarily difficult: you would simply use hashes, possibly using a deeper understanding of signal processing and compression than simply naive copies of exact data.

This is apparently the standard means of dealing with property inherited from English common law. And if you think it is bad for IP cases, it is an absolute disaster where investment derivatives are included (which may have leverage scales that go to multiple times the wealth of the entire planet). This came up in the banking disaster of 2008 when it often became impossible to determine who owned the mortgage on a house.

[NichG]

I do not think it is even comparable to a bright organism. A key component of intelligence is the ability to identify one's own mistakes (feel free to insert here a cheap joke about your most disliked politician). Neural networks & other modern AIs are capable of learning, but still require an outside intelligence - i.e. a human - to tell them what category something belongs to. A spam filter will happily keep marking real emails as spam forever if you don't tell it is making a mistake. In the end, yes, it is creating its own algorithm to follow, which is impressive, but ultimately, it is still a set of rules.

To be clear, I do not look down my nose at any of this - when it works it is impressive, and facial recognition is at this point better at it than humans are - but it is not something that I would call intelligent.

Grey Wolf

Predictive coding is basically AI built entirely out of 'identify one's own mistakes and adapt' - in fact, based on the way predictive coding based things work, their entire sensory input is essentially constructed only of their mistakes - they must err in order to detect. I don't tend to do much with that kind of network because its a bit fiddly (asking for something that produces behavior as an indirect byproduct is always a bit harder than just asking for it directly, and since often I do know what it is I want the AI to do...)

Anyhow, AI is a vast, vast field now. There's a dozen ways out there for any given situation or problem or task. So blanket statements like 'AI is/isn't X' tend to be pretty far off. Each technique has its own idiosyncracies as well.

Taking the email example:

If you wanted an AI that learned on its own to ignore a subset of emails, you'd need to give it an appropriate context in order to ground it - e.g. it can't just receive a bunch of emails and that's the end of the story (though I'd still bet on unsupervised clustering to at least distinguish the 'spam' cluster from the various other stuff). If you want that AI to behave like a person, it also has to have a life outside of the emails which the emails somehow make contact with - then it can learn that the spam emails are basically uninformative about anything it cares about, whereas other emails are useful and integrated with that context. One way to do it would be to provide a stream of emails but force the AI to pull information from them via a limited attention mechanism. Then, after some time spent with the emails, you could task the AI to e.g. answer questions or act in order to obtain a reward or outcome (which could also be self-supervised, using motivation functions like empowerment or curiosity).

Now where it really gets fun is, with the right kind of indexing, you can use that same attention mechanism (indexed properly) to decide e.g. which out of a set of people to send a particular email to in order to query for the information needed to answer those future questions or take those future actions. You just ask it the same way: if we pretend we had a response from everyone, which response would be most likely to be salient?

[Tyndmyr]

So, basically six forms of robotic slavery vs extinction of the human race.

Why, one might even posit that this would provide a motive for robots to kill all humans!

It's possible that humans might circumvent this by only allowing robots to live that are juuust dumb enough to not be able to revolt, but that's still a wee bit dystopian from the robot point of view.

[wumpus]

Brute force mass-parallel searching isn't a sign of artificial intelligence. What makes the google search interesting is not that it has a gazillion processors doing basic string matching in parallel, it is how it determines value and relevance, and while certainly a stroke of genius (rank a page's importance based on the number of other pages that link to it - a concept they borrowed from scientific papers), the algorithm (note spelling) itself is not, nor does it intend to be "intelligent". It literally is just the same algorithm that any text editor uses when you search on your own computer, just scaled up.

Grey Wolf

Note that while this shouldn't pass the Turing Test, this type of "intelligence" is great for getting A's in class. I'd expect that "brute force mass-parallel searching" will be an important part of any AI. From memory, I think that was Lt. Commander Data's primary function and written into Star Trek back when "internet search" meant "Kibo grepping his name".