I think the legal consequences would by far surpass any of the more practical aspects mentioned here ? Something of value NOT owned (ultimately) by a person of a government ?
What if there is a flaw in the decision software and something goes horribly wrong ? Would not be the first time... The software is written by someone, the decision to use the software was done by someone, etc. Still if really no one else is to blame, then the government is responsible ? So if nobody else then they 'own' it ?
It also means abandoning the risk taking and therefore profit potential / earning power for people.
Challenging.
And once solved why not use the same concept for rental apartments, clothes ?
From what I read almost none of the dynamics mentioned require self-ownership... Except for the self ownership part. Most of the dynamics at play could still be in play with a corporate owner. And after laying out the R&D funds, what incentive would a corporation have to layer in the extra self-ownership code instead of just skimming profits and playing a more centralized coordination role?
Although the article talks down the use of AI at face value, my gut reaction to this is; Right, that would be the vehicle's responsibility.
In regards to risk specifically, I would like to believe risk/route evaluation at the least is considered as part of a profit algo for the vehicle itself. Yes, this is a assuming a more abundance of data points available to the vehicles itself — but I'd like to assume infrastructure to streets and safety are being enhanced along side of the vehicle(s) themselves.
And I think for the sake of the thought exercise you have to consider what happens to related markets — specifically in this scenario, risk. If vehicles are no longer being sold to people, neither is auto insurance (at least in today's form).
Ultimately companies (the shares) are owned by persons and managed by persons. This is very fundamental to our (Western ?) system. Its nearly impossible to start a company or have a bank account without identifying the Ultimate Beneficial Owner (UBO). Only possible in a very few (questionable) places.
Yes we have ways to deal with that: taking their money and/or freedom. Don't forget that revenge is an important element. Taking the 'freedom' of a car (scraping it ?) probably feels less satisfying then putting a man in jail. Especially if, for example, a family member is unnecessary and severely harmed.
'Bad luck' and 'it could happen to anyone' might be sufficient in the past, but nowadays we like to make someone (and not something) responsible. A person to take away his freedom or a big company to get a lot of money. A car has neither.
> Most writers acknowledge that the deed is lost or no longer exists, if in fact it ever did. Such a deed, even if it did exist, would have no legal standing. Under common law, the person receiving the property in question must have the legal capacity to receive it, and the property must be delivered to—and accepted by—the recipient.
I think the BBC _journalist_ forgot completely to mention this tiny issue regarding a legal person. Fun to read, but more scifi than reality...
Potentially significant bits from the above wikipedia page:
The original transfer of the deed would not be valid because
'Such a deed, even if it did exist, would have no legal standing. Under common law, the person receiving the property in question must have the legal capacity to receive it, and the property must be delivered to—and accepted by—the recipient.'
and ' Athens-Clarke County confirms that the tree is in the right-of-way, and is thus “accepted for care” by municipal authorities;'
however,
'Regarding Jackson’s deed, one writer noted at the beginning of the 20th century, “However defective this title may be in law, the public recognized it.”[13] In that spirit, it is the stated position of the Athens-Clarke County unified government that the tree, in spite of the law, does indeed own itself.[14]'
So, while if you wished to do this with some land you owned, and a tree, it probably would not be considered legally valid in the US (at the moment), but there might be a chance you could get it to be effectively the case?
A tree can't give consent, ergo it can't be on the receiving end of a contract.
That the public plays along has more to do with novelty and tradition than with a serious legal precedent. Saying that the tree de facto owns itself despite the lack of legal underpinning is like saying Emperor Norton was de facto the emperor of the United States because people accepted his fake money as legal tender and acknowledged his claims.
Consider if Emperor Norton would have engaged in an armed conflict with the US government, for example. It'd be pretty obvious that he would be treated as an American citizen, not as a foreign head of state. Likewise if the Tree That Owns Itself's supposed legal status were ever at odds with the well-being of humans (to a degree that dwarfs its cultural value), it's pretty obvious who's side the courts would be on.
At best, the tree could be treated as a corporation with the municipal government or the public acting on its behalf.
This is a cool concept for lots of things, maybe even big things like companies and utilities. A for-profit entity that operates just like a normal one, but is programmed to not to simply maximize profits but to maximize social good. It would work well on things that don't require too much human thinking to run (just balancing equations and running equipment), but are normally hindered by self-interest.
The vision I had was that these things wouldn't be run by some strong AI trying to figure out what's best, but just companies and machines that don't need complicated ideas to run. Just application specific "AI" that's good enough to balance the books and keep everything running, save enough for a rainy day, and keeping the consumers interests in mind.
But I guess if left unchecked, as technology improves, things would tend towards Skynet.
Wait, do you seriously think that Sun, Bell, Ericsson (java, c, erlang) set out to maximize social good? Of course everybody knows that the internet is powered by well wishes and puppy dreams, not costly telecommunication hardware.
For what time interval does your algorithm maximize profits/social good/etc? We've seen what maximizing profits for the next earnings report does. Ten years? For our lifetimes? Our children's?
Too short, and you miss some big opportunities. Too long, and the algorithm might hedge against all the unknowns and miss some big opportunities.
How about an ownerless house that keeps its own rent money? An ownerless McDonald's franchise? Ownership is not just about money. Driverless cars don't really change anything about the concept of ownership.
If you think driverless taxis are such a great idea and you want them to be operated in a non-greedy way, start a non-profit, or start a corporation and make it public so that anyone who wants can invest and share in the profits, and pay dividends. If you insist "nobody" owns and operates the fleet of cars, it will probably end up being the government, who already programs the fare boxes in taxis.
Several authors have written about algorithmically generated & operated companies or entities (Charlie Stross, the Ghost in the Shell) either as litigation shields or entities which outlived their owners but continued to operate.
The idea is certainly odd, to essentially create autonomous and roughly speaking self-sufficient/directed artificial creatures, but it's consistent at least.
The thought that makes me saddest i think is just that the core quality we'd imbue humanity's first autonomous creations with would be capitalist wage-slavery (a literal non-hyperbolic one in this case).
Saddling people with debt, then paying them a minimum wage. Effectively they become slaves unable to leave their employer due to dependencies on things such as lunch programs, health insurance, leave benefits.
This system of wage-slavery is enforced today through police, lawyers, bankers and politicians working in accidental concert to ensure that you must go into debt (by renting or buying a house), and that you must work for whatever wages you can attract (because no money means no place to live), or be thrown in gaol for the sin of being poor.
There are towns in the USA re-introducing debtors prisons ("poor houses") where people who have defaulted on debt repayments can live in penury, having their wages garnished by the state before they get hold of the money themselves.
Is that really sad given that these 'creations' would not be in any way self aware? The idea is far-fetched, but the programming required is possible today. This wouldn't be advanced AI.
For some entities it makes sense (the lobsters in Accelerando clearly benefit from it and their emancipation does not do any harm) but for society as a whole it is surely a dangerous idea.
Are you intentionally changing the subject, or do you equate self ownership to "ownerless"? Or is it that you are skipping ahead logically because you don't recognize a non-human entity as being capable of self ownership? If that is the case, that is the more important issue that needs to be discussed.
The difference is that self driving cars are pretty close to coming to fruition. It's a pretty natural step to assume a car that can realize it's broken and hire a mechanic to repair it.
Houses aren't nearly that far along yet.
And ownerless taxis don't need to be non-greedy. The algorithms that make the most money and are able to produce the most offspring (either by buying other cars or paying off a factory to build more) will probably help the economy the most. If the car is effective at making money, 1 car can eventually turn into 1000 with little human management. A non-greedy car might not be able to achieve that.
This is an interesting idea, probably ahead of its time. I can't help but think of this quote: "Because the people who are crazy enough to think they can change the world, are the ones who do."
If AI starts to sort of work, we may get to see some kind of corporation run by computers, owned by other similar corporations. Probably out of some financial haven such as the Seychelles or Cyprus. This is more likely in the financial space than in the self-driving car space.
I could see someone setting this up and retaining some minority ownership so that they get dividends, while the thing runs itself.
Competitive markets always end up with zero economic profit, all the 'profit' is compensation for the efforts and risks in allocating capital.
If there is no need for a human doing the allocation of capital anymore, and the market demand is so liquid that the capital risks are zero, then yeah we'll get autonomous profitless (in economic and accounting sense) cab networks.
This is not a 'new' type of corporation. It's just a corporation with one more task automated.
> StorJ is not able to find new hosting environments on its own, due to a lack of sufficiently powerful AI— but it can purchase the knowledge from humans: When an instance of StorJ is ready to reproduce it can announce a request for proposal: Who will make the best offer for a script that tells it how to load itself onto a new hosting environment and tells it all the things it needs to know how to survive on its own there?
The 'AI' doesn't have to be smart it enough to do everything itself, it just has to be smart enough to be able to hire a human to help it. This step could be broken down into lots of mturk-style tasks, each done by a different human:
1) Find hosting providers who accept Bitcoin
2) Evaluate the service provided by these hosting providers
3) Sign up for an account on this hosting provider
4) Setup an account on this server with this SSH public key
You could even deal with legal issues that way - have a retainer with a law firm, and if any issues crop up send them an email.
The article very explicitly states that this is not about self aware cars. Merely about cars which are programmed to make enough money to pay for their own gas/electricity, repairs and insurance. And probably repay cost of being built to its manufacturer too.
to own itself, first someone would have to give that ownership (and consequent profit) to it. Why would someone do that in the first place? Will humans not care about their own needs anymore?
Would self-owned cars have a need to know the implications of its actions? Legalities?
Will they – like humans – be able to take on small losses in order to survive? Can they be expected to uphold the law like humans are? What if they act as getaway drivers for a bank robbery? Drug mules?
Would their programmers be legally responsible for their illegal actions? Or would they, as the buyers and evaluators of the software, be responsible? Or the programmers of the evaluation software?
The article mentions that the cars could hire programmers to improve them. Why not also let them hire drivers - that way there would be no need to wait for the technology of driverless cars to become feasible.
The biggest problem with driverless cars will not be the safety technology but the people controlling them.
It's not hard to imagine a scenario where most regular cars have been outlawed or restricted in favour of safer driverless ones and bus services are replaced by fleets of self driving cars. It wouldn't be long before restrictions are be placed on the journeys for the sake of profit.
Even if the self driving cars were to stay in an individuals possession I can imagine restrictions being placed by the manufacturers or insurance companies, almost like a drm for cars.
I would suggest that current trust laws, where legal fictions like companies can be trustees, could be a foundation for creating a programmatic trust deed for cars been trustees.
The trustee (car) is given legal title to the trust property (car, software, service contracts), by the settlor (manufacturer company), but is obligated to act for the good of the beneficiaries, humans.
Here in Boston, The Subway system has been completely shutdown by the snow and the executive in charge of the system has been fired. The MBTA performs poorly and consistently goes further and further into debt.
I wonder what a system built of self-interested-interested transport robots would look like, especially in times of disruption and crisis.
What? No, that's silly. Even if the US grants personal rights to corporations (see Hobby Lobby), it's a bit much to imagine any court of law granting rights to an "autonomous" non-entity created by a third party.
Until we develop truly sentient machines, it's preposterous to think that a "thing" could own itself in any meaningful way (even corporations have humans acting on their behalf).
Also, what is the incentive to do this? It's far more likely that autonomous cars will be handled like train and bus systems.
And finally, as a programmer, any notion of (non-sentient) computer programs being granted autonomy is a nightmare to me. Yes, we can make very nice semi-autonomous toys, but the real world is simply far too complex to expect any machine to be able to handle all of it. This is why we have emergency overrides.
Our tolerance for errors made by machines is far lower than our tolerance for errors made by humans, anyway. If a machine autonomously made a mistake that resulted in the loss of human life, it'd be an outrage even if no human could have fared any better under the same circumstances. The liability alone would be a headache.
64 comments
[ 2.8 ms ] story [ 132 ms ] threadWhat if there is a flaw in the decision software and something goes horribly wrong ? Would not be the first time... The software is written by someone, the decision to use the software was done by someone, etc. Still if really no one else is to blame, then the government is responsible ? So if nobody else then they 'own' it ?
It also means abandoning the risk taking and therefore profit potential / earning power for people.
Challenging.
And once solved why not use the same concept for rental apartments, clothes ?
In regards to risk specifically, I would like to believe risk/route evaluation at the least is considered as part of a profit algo for the vehicle itself. Yes, this is a assuming a more abundance of data points available to the vehicles itself — but I'd like to assume infrastructure to streets and safety are being enhanced along side of the vehicle(s) themselves.
This argument applies to people too and we have ways to deal with it other than people being owned.
I think the BBC _journalist_ forgot completely to mention this tiny issue regarding a legal person. Fun to read, but more scifi than reality...
The original transfer of the deed would not be valid because 'Such a deed, even if it did exist, would have no legal standing. Under common law, the person receiving the property in question must have the legal capacity to receive it, and the property must be delivered to—and accepted by—the recipient.'
and ' Athens-Clarke County confirms that the tree is in the right-of-way, and is thus “accepted for care” by municipal authorities;'
however,
'Regarding Jackson’s deed, one writer noted at the beginning of the 20th century, “However defective this title may be in law, the public recognized it.”[13] In that spirit, it is the stated position of the Athens-Clarke County unified government that the tree, in spite of the law, does indeed own itself.[14]'
So, while if you wished to do this with some land you owned, and a tree, it probably would not be considered legally valid in the US (at the moment), but there might be a chance you could get it to be effectively the case?
That the public plays along has more to do with novelty and tradition than with a serious legal precedent. Saying that the tree de facto owns itself despite the lack of legal underpinning is like saying Emperor Norton was de facto the emperor of the United States because people accepted his fake money as legal tender and acknowledged his claims.
Consider if Emperor Norton would have engaged in an armed conflict with the US government, for example. It'd be pretty obvious that he would be treated as an American citizen, not as a foreign head of state. Likewise if the Tree That Owns Itself's supposed legal status were ever at odds with the well-being of humans (to a degree that dwarfs its cultural value), it's pretty obvious who's side the courts would be on.
At best, the tree could be treated as a corporation with the municipal government or the public acting on its behalf.
That is how Skynet happens.
But I guess if left unchecked, as technology improves, things would tend towards Skynet.
Too short, and you miss some big opportunities. Too long, and the algorithm might hedge against all the unknowns and miss some big opportunities.
If you think driverless taxis are such a great idea and you want them to be operated in a non-greedy way, start a non-profit, or start a corporation and make it public so that anyone who wants can invest and share in the profits, and pay dividends. If you insist "nobody" owns and operates the fleet of cars, it will probably end up being the government, who already programs the fare boxes in taxis.
Several authors have written about algorithmically generated & operated companies or entities (Charlie Stross, the Ghost in the Shell) either as litigation shields or entities which outlived their owners but continued to operate.
The idea is certainly odd, to essentially create autonomous and roughly speaking self-sufficient/directed artificial creatures, but it's consistent at least.
The thought that makes me saddest i think is just that the core quality we'd imbue humanity's first autonomous creations with would be capitalist wage-slavery (a literal non-hyperbolic one in this case).
This system of wage-slavery is enforced today through police, lawyers, bankers and politicians working in accidental concert to ensure that you must go into debt (by renting or buying a house), and that you must work for whatever wages you can attract (because no money means no place to live), or be thrown in gaol for the sin of being poor.
There are towns in the USA re-introducing debtors prisons ("poor houses") where people who have defaulted on debt repayments can live in penury, having their wages garnished by the state before they get hold of the money themselves.
You mig have other rationalisations for it.
By what and who's metric? There is no stable, verifiable entity "society as a whole".
Are you intentionally changing the subject, or do you equate self ownership to "ownerless"? Or is it that you are skipping ahead logically because you don't recognize a non-human entity as being capable of self ownership? If that is the case, that is the more important issue that needs to be discussed.
Houses aren't nearly that far along yet.
And ownerless taxis don't need to be non-greedy. The algorithms that make the most money and are able to produce the most offspring (either by buying other cars or paying off a factory to build more) will probably help the economy the most. If the car is effective at making money, 1 car can eventually turn into 1000 with little human management. A non-greedy car might not be able to achieve that.
The ultimate absentee landlord?
I could see someone setting this up and retaining some minority ownership so that they get dividends, while the thing runs itself.
If there is no need for a human doing the allocation of capital anymore, and the market demand is so liquid that the capital risks are zero, then yeah we'll get autonomous profitless (in economic and accounting sense) cab networks.
This is not a 'new' type of corporation. It's just a corporation with one more task automated.
There are tons of complications before you get to the point of worrying about machines that might kill people if something goes wrong. :)
> StorJ is not able to find new hosting environments on its own, due to a lack of sufficiently powerful AI— but it can purchase the knowledge from humans: When an instance of StorJ is ready to reproduce it can announce a request for proposal: Who will make the best offer for a script that tells it how to load itself onto a new hosting environment and tells it all the things it needs to know how to survive on its own there?
The 'AI' doesn't have to be smart it enough to do everything itself, it just has to be smart enough to be able to hire a human to help it. This step could be broken down into lots of mturk-style tasks, each done by a different human:
1) Find hosting providers who accept Bitcoin
2) Evaluate the service provided by these hosting providers
3) Sign up for an account on this hosting provider
4) Setup an account on this server with this SSH public key
You could even deal with legal issues that way - have a retainer with a law firm, and if any issues crop up send them an email.
But they would be programmed to seek self-improvement in order to avoid becoming obsolete."
No, no, no: we are not going to apply evolution, we will simply let the fittest survive!
Will they – like humans – be able to take on small losses in order to survive? Can they be expected to uphold the law like humans are? What if they act as getaway drivers for a bank robbery? Drug mules?
Would their programmers be legally responsible for their illegal actions? Or would they, as the buyers and evaluators of the software, be responsible? Or the programmers of the evaluation software?
In fact most of it is being done currently, except the owner-less bit.
It seems a common mind screw up.
People think when technology X is released all these other advanced technologies will come with it.
Voice recognition is an obvious example.
Just because a computer can understand the words perfectly, it still has no idea what you are saying.
For it to do that it'd have to be close to passing the Turing test.
This fallacy should have a name if it doesn't already.
It's not hard to imagine a scenario where most regular cars have been outlawed or restricted in favour of safer driverless ones and bus services are replaced by fleets of self driving cars. It wouldn't be long before restrictions are be placed on the journeys for the sake of profit.
Even if the self driving cars were to stay in an individuals possession I can imagine restrictions being placed by the manufacturers or insurance companies, almost like a drm for cars.
The trustee (car) is given legal title to the trust property (car, software, service contracts), by the settlor (manufacturer company), but is obligated to act for the good of the beneficiaries, humans.
I wonder what a system built of self-interested-interested transport robots would look like, especially in times of disruption and crisis.
Until we develop truly sentient machines, it's preposterous to think that a "thing" could own itself in any meaningful way (even corporations have humans acting on their behalf).
Also, what is the incentive to do this? It's far more likely that autonomous cars will be handled like train and bus systems.
And finally, as a programmer, any notion of (non-sentient) computer programs being granted autonomy is a nightmare to me. Yes, we can make very nice semi-autonomous toys, but the real world is simply far too complex to expect any machine to be able to handle all of it. This is why we have emergency overrides.
Our tolerance for errors made by machines is far lower than our tolerance for errors made by humans, anyway. If a machine autonomously made a mistake that resulted in the loss of human life, it'd be an outrage even if no human could have fared any better under the same circumstances. The liability alone would be a headache.