|
|
|
|
|
| |
Posted by Walter Bushell on September 9, 2008, 5:24 pm
Please log in for more thread options
>
> I'd want a robot that was capable of self-replication. With near-term
> technology it'd be rather large, but you're allowing for quite a lot of
> technological advancement so in this case it wouldn't need to be. Maybe
> even something as flexible as Stargate's Replicators.
>
> With such a system it would be straightforward to industrialize the
> whole solar system. As much habitat as we want, as much energy as we
> want, whatever megastructures suit our fancy. And, presumably, they
> could vacuum and clean the dishes as well. Just make sure to program
> them better than the typical SF robot-run-amuck.
Ah, there's the rub. Being without such a limitation would be a survival
advantage and once started could take over the entire system.
|
|
Posted by Bryan Derksen on September 9, 2008, 7:22 pm
Please log in for more thread options
Walter Bushell wrote:
>> With such a system it would be straightforward to industrialize the
>> whole solar system. As much habitat as we want, as much energy as we
>> want, whatever megastructures suit our fancy. And, presumably, they
>> could vacuum and clean the dishes as well. Just make sure to program
>> them better than the typical SF robot-run-amuck.
>
> Ah, there's the rub. Being without such a limitation would be a survival
> advantage and once started could take over the entire system.
Which is why you program it well. Perhaps we can't make it
mathematically impossible for such a system to run amuck, but you can
make it arbitrarily unlikely. Add as many bits of error-correction to
the genome as you like, and if one family of robots does run amuck have
the billions of other loyal robots that replicated correctly help to
destroy it. That'll turn such mutations into a serious survival
_dis_advantage.
We've managed to keep many other species of self-replicating organisms
domesticated without any of them rising up in revolt yet, I don't see
any reason why robots would be different in principle.
|
|
Posted by Walter Bushell on September 10, 2008, 9:46 pm
Please log in for more thread options
> Walter Bushell wrote:
> >> With such a system it would be straightforward to industrialize the
> >> whole solar system. As much habitat as we want, as much energy as we
> >> want, whatever megastructures suit our fancy. And, presumably, they
> >> could vacuum and clean the dishes as well. Just make sure to program
> >> them better than the typical SF robot-run-amuck.
> >
> > Ah, there's the rub. Being without such a limitation would be a survival
> > advantage and once started could take over the entire system.
>
> Which is why you program it well. Perhaps we can't make it
> mathematically impossible for such a system to run amuck, but you can
> make it arbitrarily unlikely. Add as many bits of error-correction to
> the genome as you like, and if one family of robots does run amuck have
> the billions of other loyal robots that replicated correctly help to
> destroy it. That'll turn such mutations into a serious survival
> _dis_advantage.
>
> We've managed to keep many other species of self-replicating organisms
> domesticated without any of them rising up in revolt yet, I don't see
> any reason why robots would be different in principle.
Mostly by close watching. If a Scottish Collie is found to have killed a
sheep, all his or her get (all descendants) are killed. This has been
going on for many years.
|
|
Posted by Bryan Derksen on September 11, 2008, 12:29 am
Please log in for more thread options
Walter Bushell wrote:
>> We've managed to keep many other species of self-replicating organisms
>> domesticated without any of them rising up in revolt yet, I don't see
>> any reason why robots would be different in principle.
>
> Mostly by close watching. If a Scottish Collie is found to have killed a
> sheep, all his or her get (all descendants) are killed. This has been
> going on for many years.
Yup. And some supervision is probably going to be needed for responsible
husbandry of self-replicating machines, too. We'll have some major
advantages though. The machines will be designed by us initially, not
domesticated from existing species, so we won't have to worry about
dealing with undesirable 'wild' traits. The machines themselves can help
with their own supervision - for example the scenario I mentioned
earlier where loyal machines could be ordered to help destroy any
disloyal ones that crop up. Putting in lots of redundant kill switches
would help too. And so on.
The Jurassic Park scenario where life forms are seemingly directed by
some invisible hand to instantly take advantage of any chance at
'revolt', no matter how unlikely, is just fantasy. I see no fundamental
reason why we couldn't get along perfectly well with self-replicating
robot servants.
|
|
Posted by Mike H on September 11, 2008, 2:34 am
Please log in for more thread options
Bryan Derksen wrote:
> Walter Bushell wrote:
>>> With such a system it would be straightforward to industrialize the
>>> whole solar system. As much habitat as we want, as much energy as we
>>> want, whatever megastructures suit our fancy. And, presumably, they
>>> could vacuum and clean the dishes as well. Just make sure to program
>>> them better than the typical SF robot-run-amuck.
>> Ah, there's the rub. Being without such a limitation would be a survival
>> advantage and once started could take over the entire system.
>
> Which is why you program it well. Perhaps we can't make it
> mathematically impossible for such a system to run amuck, but you can
> make it arbitrarily unlikely. Add as many bits of error-correction to
> the genome as you like, and if one family of robots does run amuck have
> the billions of other loyal robots that replicated correctly help to
> destroy it. That'll turn such mutations into a serious survival
> _dis_advantage.
>
> We've managed to keep many other species of self-replicating organisms
> domesticated without any of them rising up in revolt yet, I don't see
> any reason why robots would be different in principle.
Then again, they don't have to revolt to take over.
Cats have been silently controlling us for thousands of years.
Robots will be taught to do the same.
|
| Similar Threads | Posted | | We can make you talk. | November 18, 2005, 10:42 am |
| Re: !!!!! Make Over $200 per day in Autopilot !!!!! 2^oLh=0uR+ | December 29, 2007, 3:41 pm |
| Re: !!!!! Make Over $200 per day in Autopilot !!!!! ]O*/h?3f)Z | January 11, 2008, 4:29 pm |
| easy way to make money | March 21, 2006, 9:44 am |
| Have shop. Bored. Looking for something to make. | July 5, 2006, 6:02 pm |
| Re: Make Money in Two Weeks - Act Now!!! | January 11, 2008, 4:29 pm |
| Re: !!!!! Make Over $200 per day in Autopilot !!!!! or;&f_I1We | January 15, 2008, 3:28 pm |
| All make-ups will be elderly broken satisfactions. | December 2, 2007, 4:25 am |
| how to make a rpbot for painting an object | August 21, 2008, 12:06 pm |
| how to make a stepper motor step gradually? | June 2, 2006, 6:17 am |
|
|
|
XML site map
> I'd want a robot that was capable of self-replication. With near-term
> technology it'd be rather large, but you're allowing for quite a lot of
> technological advancement so in this case it wouldn't need to be. Maybe
> even something as flexible as Stargate's Replicators.
>
> With such a system it would be straightforward to industrialize the
> whole solar system. As much habitat as we want, as much energy as we
> want, whatever megastructures suit our fancy. And, presumably, they
> could vacuum and clean the dishes as well. Just make sure to program
> them better than the typical SF robot-run-amuck.