There are some pretty big misunderstandings there...
First of all, very little of American forest is old growth, at least speaking from the perspective of someone from Indiana, where we have millions (around 4.5) of acres of forest, but only 2000 of those acres are old growth. And I know the situation is similar in most states, if not quite as bad as here. (And old growth is a misnomer anyway; it can mean different things to different people - there are several useful definitions for the term, and merely being old does not under all definitions automatically make a forest "old growth" - it has more to do with the condition of the forest and the type of trees, I think)
Second of all, citing a number from 1920 doesn't really help. 1920 was within like 10 years of the minimum forestation (at least for Indiana, I am sure it is similar elsewhere) - they had in our case already ripped up the vast majority of forest in the state and almost all of the old growth forest (basically all old growth forest had been logged by 1930). Since then much land RELATIVELY had been reforested (often not intentionally, so the new forest is only due in part to that 7% "planted by man"), but the vast majority of the damage was already done, and NONE of the new forest by definition will be old growth. And in fact there was still considerable deforestation after that time, but so much forest was added that it hides alot of that. (By the way, most estimates are the Indiana used to be around 80% forest; a quick check on google gave me numbers of around 20% current forestation - which is UP from the 1920 number. And that 80% consisted of significant amounts of old growth - though note it was not exclusively what would be called "old growth" by most people - and now old growth is well less than 1%.)
(Okay, forgive me if I go off course here, I am not an AI expert by any means, but I do have a little idea what is going on here - I hope...)
This statement though is based on some false conceptions about AI. Many of the most interesting AI systems don't work exactly the way we usually think of programs being written. Learning networks and neural networks can automate the learning process - we don't tell them how to do something, we create them and let them watch us. Such techniques have ALREADY been used to teach systems to drive cars and fly airplanes safely. [Okay, admittedly, they might still have trouble dealing with "unique situations" but then so do many humans the first time - this is a problem of lack of experience at that point.]. Frankly, I am more impressed by this than I would be by the ability to bluff well at poker. Yeah, so maybe we can't read facial expressions well yet, but give them time...
If there ever was to be an interesting intelligent system, it would most likely be the result of a system that we cannot easily fully encompass with simple algorithms (which is to say, we may be able to understand the rules for the system, but the end product may not be readily understandable to us). And the whole key here is WE DON"T HAVE TO KNOW HOW TO WRITE THE ALGORITHMS FOR IT.
Slashdot Mirror
There are some pretty big misunderstandings there... First of all, very little of American forest is old growth, at least speaking from the perspective of someone from Indiana, where we have millions (around 4.5) of acres of forest, but only 2000 of those acres are old growth. And I know the situation is similar in most states, if not quite as bad as here. (And old growth is a misnomer anyway; it can mean different things to different people - there are several useful definitions for the term, and merely being old does not under all definitions automatically make a forest "old growth" - it has more to do with the condition of the forest and the type of trees, I think) Second of all, citing a number from 1920 doesn't really help. 1920 was within like 10 years of the minimum forestation (at least for Indiana, I am sure it is similar elsewhere) - they had in our case already ripped up the vast majority of forest in the state and almost all of the old growth forest (basically all old growth forest had been logged by 1930). Since then much land RELATIVELY had been reforested (often not intentionally, so the new forest is only due in part to that 7% "planted by man"), but the vast majority of the damage was already done, and NONE of the new forest by definition will be old growth. And in fact there was still considerable deforestation after that time, but so much forest was added that it hides alot of that. (By the way, most estimates are the Indiana used to be around 80% forest; a quick check on google gave me numbers of around 20% current forestation - which is UP from the 1920 number. And that 80% consisted of significant amounts of old growth - though note it was not exclusively what would be called "old growth" by most people - and now old growth is well less than 1%.)
Collections of works are already protected under current law. E.g. what university of washington copyright connection has to say about it and the plain old copyright code
(Okay, forgive me if I go off course here, I am not an AI expert by any means, but I do have a little idea what is going on here - I hope...)
This statement though is based on some false conceptions about AI. Many of the most interesting AI systems don't work exactly the way we usually think of programs being written. Learning networks and neural networks can automate the learning process - we don't tell them how to do something, we create them and let them watch us. Such techniques have ALREADY been used to teach systems to drive cars and fly airplanes safely. [Okay, admittedly, they might still have trouble dealing with "unique situations" but then so do many humans the first time - this is a problem of lack of experience at that point.]. Frankly, I am more impressed by this than I would be by the ability to bluff well at poker. Yeah, so maybe we can't read facial expressions well yet, but give them time...
If there ever was to be an interesting intelligent system, it would most likely be the result of a system that we cannot easily fully encompass with simple algorithms (which is to say, we may be able to understand the rules for the system, but the end product may not be readily understandable to us). And the whole key here is WE DON"T HAVE TO KNOW HOW TO WRITE THE ALGORITHMS FOR IT.