Why Darwinian evolution? Because it’s hard to know if it’s suffering otherwise?
I’m proposing a society in which living things, or sufficiently detailed emulations of them, and especially sapient ones, have preferred moral and legal status. I’m reasonably confident that for something complex and mobile with senses, Darwinian evolution will generally produce mechanisms that act like pain and suffering, for pretty obvious reasons. So I’m proposing a definition of ‘suffering’ rooted in evolutionary theory, and only applicable to living things, or emulations/systems sufficintly closely derived from them. If you emulate such a system, I’m proposing that we worry about its suffering to the extent that it’s a sufficiently detailed emulation still functioning in its naturally-evolved design. For example I’m suggesting that a current-scale LLM doing next-token generation of the pleadings of a torture victim not be counted as suffering for legal/moral purposes: IMO the inner emulation of a human it’s running isn’t (pretty clearly based on parameter count to, say, synapse count) a sufficiently close simulation of a biological organism that we should consider it’s behavior as ‘suffering’: for example, no simulations of pain centers are included. Increase the accuracy of simulation sufficiently, and there comes a point (details TBD by a society where this matters) where that ceases to be true.
So, if someone wants a particular policy enacted, and uses sufficient computational resources to simulate 10^12 separate and distinct sapient kittens-girls who have all been edited so that they will suffer greatly if this policy isn’t enacted, we shouldn’t encourage that sort of moral blackmail or ballot-stuffing. I don’t think they should be able to win the vote or utilitarian decision-making balance just by custom-making a lot of new voters/citizens: it’s a clear instability in anything resembling a democracy or that uses utilitarian ethics. I might even go so far as to suggest that that Darwinian evolution cannot have happened ‘in silico’, or at least that if it did it must be a very accurate simulation of a real physical environment that hasn’t been tweaked to produce some convenient outcome. So even if they expend the computational resources to in-silico evolve 10^12 separate and distinct sapient kitten-girls who will otherwise suffer greatly, that’s still moral blackmail. If you want to stuff the electorate with supporters, I think you should have to do it the old-fashioned way, by physically breeding and raising them — mostly because this is expensive enough to be impractical.
I’m proposing a society in which living things, or sufficiently detailed emulations of them, and especially sapient ones, have preferred moral and legal status. I’m reasonably confident that for something complex and mobile with senses, Darwinian evolution will generally produce mechanisms that act like pain and suffering, for pretty obvious reasons. So I’m proposing a definition of ‘suffering’ rooted in evolutionary theory, and only applicable to living things, or emulations/systems sufficintly closely derived from them. If you emulate such a system, I’m proposing that we worry about its suffering to the extent that it’s a sufficiently detailed emulation still functioning in its naturally-evolved design. For example I’m suggesting that a current-scale LLM doing next-token generation of the pleadings of a torture victim not be counted as suffering for legal/moral purposes: IMO the inner emulation of a human it’s running isn’t (pretty clearly based on parameter count to, say, synapse count) a sufficiently close simulation of a biological organism that we should consider it’s behavior as ‘suffering’: for example, no simulations of pain centers are included. Increase the accuracy of simulation sufficiently, and there comes a point (details TBD by a society where this matters) where that ceases to be true.
So, if someone wants a particular policy enacted, and uses sufficient computational resources to simulate 10^12 separate and distinct sapient kittens-girls who have all been edited so that they will suffer greatly if this policy isn’t enacted, we shouldn’t encourage that sort of moral blackmail or ballot-stuffing. I don’t think they should be able to win the vote or utilitarian decision-making balance just by custom-making a lot of new voters/citizens: it’s a clear instability in anything resembling a democracy or that uses utilitarian ethics. I might even go so far as to suggest that that Darwinian evolution cannot have happened ‘in silico’, or at least that if it did it must be a very accurate simulation of a real physical environment that hasn’t been tweaked to produce some convenient outcome. So even if they expend the computational resources to in-silico evolve 10^12 separate and distinct sapient kitten-girls who will otherwise suffer greatly, that’s still moral blackmail. If you want to stuff the electorate with supporters, I think you should have to do it the old-fashioned way, by physically breeding and raising them — mostly because this is expensive enough to be impractical.