4 ms·

This still implicitly asserts that the murderers are to blame for something. Well, this will quickly turn into the nuances of definitions of words. Blame can h

This still implicitly asserts that the murderers are to blame for something. Well, this will quickly turn into the nuances of definitions of words. Blame can have several subtilely different meanings, so call it blame if you want. I will avoid that problem by saying that we go from asserting that a person did something wrong and could and should have done otherwise to asserting that a person did something wrong. "Protecting others" carries an implicit assertion that failing to abide by this principle means you are broken [...] Yes, if we have the principle that one person should not harm another and you do not follow that principle, then in a deterministic universe we would say that you are not behaving as expected, that you are broken if you want. Well, there is probably some nuance required here, are you really broken and will keep harming others or was this more like an exceptional situation making you misbehave once but unlikely to do so again? "[...] rather than that the principle itself is mistaken, as would be true if you really believed blame didn't exist. Take the example of some manufacturing robots, if one of them breaks and keeps crashing into others, then you will remove and replace the broken one, you are not going to blame it or assign any responsibility to it, because one of the wires connecting a sensor got lose. I do not see why enforcing a policy, categorizing behavior into acceptable and unacceptable behavior, requires blame in any form. We just categorize, we do not even have to look why something ended up in one category or the other. A related but different question is what our policy is, where the line between acceptable and unacceptable behavior lies. And then you still have to account for whether the transgressor was coerced rather than was acting from their beliefs and desires (aka made a free choice). That falls under the point mentioned above, more nuance, whether the behavior was situation specific or is likely inherent. But that is again independent of free will - whether I am determined to do bad things or freely decide to do bad things, one should try to prevent me from doing bad things. And it matters even less in case of coercion, or maybe it is even more complicated in the case of free will. There is not really anything to be said about deterministic coercion to do a bad thing, but if you have free will, maybe you should have resisted more? This will not always apply, but at least sometimes you will have a choice to resist the coercion and probably experience some negative consequences in return, or to go along a cause negative consequences for others.

naasking 3y ago
> a person did something wrong and could and should have done otherwise to asserting that a person did something wrong. These statements seem equivalent. To say someone did something wrong is to say they should have done something else. > then you will remove and replace the broken one, you are not going to blame it or assign any responsibility to it Yes, you are literally saying this robot is to blame for the problem. The only difference with intelligent beings is that they can understand what they did wrong if they are given feedback, and thus change their future behaviour, and that's why holding them responsible requires a different kind of action than holding a dumb robot responsible. When AI advances to the point that the robot too can understand this, then it would make sense to also hold it responsible in a similar way to humans. Note that responsibility and blame don't necessarily entail punishment. This is a common mistake. In many prior discussions, I've found people jumping to hard determinism rather that Compatibilism simply because they don't like punishment, but nothing in Compatibilism necessarily requires punishment.
- danbruc 3y ago
  To say someone did something wrong is to say they should have done something else. They important word is could not should. Yes, you are literally saying this robot is to blame for the problem. I am saying the robot caused the issue - for which you can use the word blame if you want to - not that the robot is responsible or could have done otherwise - for which I would reserve the word blame most of the time - which he could not as the issue was caused by a lose cable. [...] than holding a dumb robot responsible. I don't think you can hold a robot responsible without stretching the definition of holding responsible quite far.
  - naasking 3y ago
    > They important word is could not should. I disagree emphatically with the inclusion of "could" in the sense that you are using it. When a mother scolds her child for misbehaving, she's saying the child should have acted otherwise, not that the child could have chosen differently if you rewound time and presented the child with the same choice again. When laypeople agree that someone "could" have done otherwise, they're not agreeing with nondeterminism, they mean that any mentally competent person with common knowledge would have chosen differently given the same circumstances, ie. that there was no mental, health-related or coercive element that was forcing them to choose wrongly. And this should be obvious if you think about it: we have centuries of legal precedent based on exactly this reasoning, and most people find it perfectly reasonable. That's why free will in law will never be overturned, despite what some hard determinists think. Whether people are deterministic is simply irrelevant to such judgments. > not that the robot is responsible or could have done otherwise Frankfurt debunked the principle of alternate possibilities' relevance to responsibility more than 40 years ago. I don't know why people still think it's relevant. > I don't think you can hold a robot responsible without stretching the definition of holding responsible quite far. This sort of exchange is not at all uncommon: Frank: what was responsible for the server downtime? Alice: an update messed up the routing tables on our routers. The intended meaning is very clear: use "who" for blameworthy agents, and "what" for blameworthy non-agents. You can of course say this sort of informal dialogue is imprecise and not in line with formal definitions, but nobody would find it a stretch or confusing.

naasking 3y ago

> a person did something wrong and could and should have done otherwise to asserting that a person did something wrong. These statements seem equivalent. To say someone did something wrong is to say they should have done something else. > then you will remove and replace the broken one, you are not going to blame it or assign any responsibility to it Yes, you are literally saying this robot is to blame for the problem. The only difference with intelligent beings is that they can understand what they did wrong if they are given feedback, and thus change their future behaviour, and that's why holding them responsible requires a different kind of action than holding a dumb robot responsible. When AI advances to the point that the robot too can understand this, then it would make sense to also hold it responsible in a similar way to humans. Note that responsibility and blame don't necessarily entail punishment. This is a common mistake. In many prior discussions, I've found people jumping to hard determinism rather that Compatibilism simply because they don't like punishment, but nothing in Compatibilism necessarily requires punishment.

danbruc 3y ago
To say someone did something wrong is to say they should have done something else. They important word is could not should. Yes, you are literally saying this robot is to blame for the problem. I am saying the robot caused the issue - for which you can use the word blame if you want to - not that the robot is responsible or could have done otherwise - for which I would reserve the word blame most of the time - which he could not as the issue was caused by a lose cable. [...] than holding a dumb robot responsible. I don't think you can hold a robot responsible without stretching the definition of holding responsible quite far.
- naasking 3y ago
  > They important word is could not should. I disagree emphatically with the inclusion of "could" in the sense that you are using it. When a mother scolds her child for misbehaving, she's saying the child should have acted otherwise, not that the child could have chosen differently if you rewound time and presented the child with the same choice again. When laypeople agree that someone "could" have done otherwise, they're not agreeing with nondeterminism, they mean that any mentally competent person with common knowledge would have chosen differently given the same circumstances, ie. that there was no mental, health-related or coercive element that was forcing them to choose wrongly. And this should be obvious if you think about it: we have centuries of legal precedent based on exactly this reasoning, and most people find it perfectly reasonable. That's why free will in law will never be overturned, despite what some hard determinists think. Whether people are deterministic is simply irrelevant to such judgments. > not that the robot is responsible or could have done otherwise Frankfurt debunked the principle of alternate possibilities' relevance to responsibility more than 40 years ago. I don't know why people still think it's relevant. > I don't think you can hold a robot responsible without stretching the definition of holding responsible quite far. This sort of exchange is not at all uncommon: Frank: what was responsible for the server downtime? Alice: an update messed up the routing tables on our routers. The intended meaning is very clear: use "who" for blameworthy agents, and "what" for blameworthy non-agents. You can of course say this sort of informal dialogue is imprecise and not in line with formal definitions, but nobody would find it a stretch or confusing.