Alejandro1 comments on Welcome to Less Wrong! (2012) - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (1430)
This is a very good response, that allows us to make our disagreement more precise. I agree that choosing menstruation, or its hypothetical unisex counterpart, is unreasonable because it is too early. I disagree that birth is too early in the same way. Pretty much everyone in our society agrees that 12-year olds cannot meaningfully consent to sex (especially with adults), whereas many believe 6-month old children to be people -- in fact, many believe fetuses to be people! You might say that they are obviously wrong, but the "obviously" is suspicious when so many disagree with you, at the very least for Aumann reasons.
To put it in another way: What makes you so certain that birth is so far off from what is reasonable as a line for personhood, when you are willing to draw your line at 10 months? That is much closer to birth than 17 is to 12 years old.
Also, I think your analogy needs a bit of amending: the relevant question is, if there was a visible unisex menstruation happening at 17 years old, and an established tradition of taking that as the age of consent, why on earth would a society change the law to make it 16 years and 2 months instead?
While true, I suspect most or all of those people would have a hard time giving a good definition of "person" to an AI in such a way that the definition included babies, adults, and thinking aliens, but not pigs or bonobos. So yes, the claim I am implicitly making with this (or any other) controversial opinion is that I think almost everyone is wrong about this specific topic.
Only if you think development happens linearly. From my knowledge of biology - which is extremely suspect, so do correct me if I'm wrong - the changes between 0 months after birth and 10 months after birth are vastly larger than the changes between 12 years after birth and 17 years after birth.
Of course our selection of analogies is reflective of our positions. From my point of view, the most relevant analogy would be a visible unisex menstruation happening at 30 years old*, and an established tradition of taking that as the age of consent, and I'm arguing that no, it should really be 16.
*(This is based on the assumption that you go through about as many developmental changes between ages 16 and 30 as you do between 0 months and 10 months, which - again from my extremely suspect recollection of biology - is roughly correct.)
(edit:formatting)
One rough effort at such definition would be: "any post-birth member of a species whose adult members are intelligent and conscious", where "birth" can be replaced by an analogous Schelling point in the development in an alien species, or by an arbitrary chosen line at a similar stage of development, if no such Schelling point exists.
You might say that this definition is an arbtrary kludge that does not "carve Nature at the joints". My reply would be that ethics is adapted for humans, and does not need to carve Nature at intrinsic joints but at the places that humans find relevant.
Your point about different rates of development is well taken, however. I am also not an expert in this topic, so we'll have to let it rest for the moment.
Problem: There's no particular reason to expect speciation to be as widespread or as clear-cut as it is in the case of Earth and humans in particular. Certainly not for machine intelligences.
It might so happen that there could be software written for the computer I'm typing this on which could give it intelligence and consciousness. (Unlikely, but not out of the realm of possibility.) Should this machine be considered a person?
The reason I'm being so nit-picky is that what I consider the natural definition (namely, "an agent capable of intelligence and consciousness", or something like that) doesn't have this problem at all. I think it's a problem your definition has only because you were forced to deviate from the natural definition to include something that doesn't really seem like it belongs in that group - namely, newborns.
For computers, hardware and software can be separated in a way that is not possible with humans (with current technology). When the separation is possible, I agree personhood should be attributed to the software rather than the hardware, so your machine should not be considered a person. If in the future it becomes routinely possible to scan, duplicate and emulate human minds, then killing a biological human will probably also be less of a crime than it is now, as long as his/her mind is preserved. (Maybe there would be a taboo instead about deleting minds with no backup, even when they are not "running" on hardware).
It is also possible than in such a future where the concept of a person is commonly associated with a mind pattern, legalizing infanticide before brain development seats in would be acceptable. So perhaps we are not in disagreement after all, since on a different subthread you have said you do not really support legalization of infanticide in our current society.
I still think there is a bit of a meta diagreement: you seem to think that the laws and morality of this hypothetical future society would be better than our current ones, while I see it as a change in what are the appropriate Schelling points for the law to rule, in response to technological changes, without the end point being more "correct" in any absolute sense than our current law.
Well, yes. This seems obvious to me.
I think I must have been unclear - the machine I'm currently typing on should obviously be a person, just because it has the potential to become a person? That seems absurd to me.
Oh, of course. I've taken it that you were asking about a case where such software had indeed been installed on the machine. The potential of personhood on its own seems hardly worth anything to me.
As a data point for your statistics, I think that a 12-year old can meaningfully consent to sex. When it comes to issues of pregnancy and having children, the consequences are greater and I don't think such yound people can consent to this, but fortunately sex and children can be kept separate today with only weak side effects.
I think that a 12-year old from a society with sensible policies would be able to give meaningful consent, but for some reason an enormous amount of work has been put into keeping American 12-year olds dangerously ignorant. That needs to be fixed first.