full text This seems like an impressive first step towards AGI. The games, like 'pong' and 'space invaders' are perhaps not the most cerebral games, but given that deep blue can only play chess, this is far more impressive IMO. They didn't even need to adjust hyperparameters between games. I'd...
Why AGI is extremely likely to come before FAI I am using FAI to mean EY's definition of provably friendly AGI implementing CEV. It is possible that the AGI which comes first will not be *provably* friendly, but will never the less turn out to be friendly, although the probability...
>Was she part of a crowd?
Yes, but they could only fit through the broken door slowly one at a time, so they couldn't rush the officer.
Incidentally, I realise that my comment sounds pro-republican because I'm talking about what I see as a democrat executing a republican. But I'm sure many republicans would love to execute democrats too, its just that at the moment the democrats seem to have far more power, and so its far more likely that democrats start executing many republicans than vice versa. Either way, my point is that mass violence is an order of magnitude higher now than it was a few decades ago.
>Dan Carlin has often mentioned... (read more)
>If, as I contend, most of them weren't, why should we take the shooting of Ashli Babbitt (in the course of committing a violent crime) as evidence that the US is headed into fascism and mass executions in the next few years, when we don't draw any such conclusions from tens of thousands of other police killings?
Heading into mass executions with perhaps 5% probability is what I said.
The point is that the police make mistakes, but now it seems that whether an action is justified depends on political affiliation. If there were a similar event, of a republican killing a democrat and seeming to get away with it on pure political affiliation... (read more)
>If the police officer who shot Babbitt was tasked with defending the Capital, then, even if there was poor judgement on his part, it seems really unreasonable to call his actions an execution. An execution is carried out, generally by a state, as part of a deliberative, measured process of deploying force. This is a distinction I think most people would appreciate.
You are correct, this is an important distinction. My impression is that there was no grounds for the officer to think that his life was in danger - since he is a man he could have physically restrained the woman without a risk of her wrestling the gun from his hands... (read more)
>If you weren't trying to suggest that there's debate about whether the people who stormed the Capitol should be executed without trial,
There isn't debate: Ashli Babbitt was de facto executed without trial (unless there is an objective legal principle where a male police officer can shoot an unarmed woman for breaking into a government building) and the shooter is described as a hero by the main levers of power, with no debate. The mass execution of capitol rioters might occur within a few years, I give it perhaps a 5% chance.
Now this might seem like I'm trying to build sympathy for the Republican cause, but I am actually not. It is wiser to side with the winner than the loser, and I for one welcome America's new fascist overlords.
This tactic is called Kafkatrapping:
"A sophistical rhetorical device in which any denial by an accused person serves as evidence of guilt."
Its also an idea I consider far more absurd than creationism, and to be frank I am shocked to see it taken seriously here.
1) If race is a social construct, then if we abandon this social construct, will a person who previously would have been conceptualised as Asian suddenly grow a few inches? If not, then this is an example of race having a physical meaning independent of mental states.
2) If differing outcomes are a matter of oppression, then shouldn't Holocaust survivors, the city of Hiroshima, the whole of China etc all... (read more)
I can understand why a race realist would argue that beleiving the truth makes you racist, but why a vicious racist?
This is just blue-tribe signalling. I apologise for being rude, but I don't know which is more absurd - the implication that only Europeans cared about gold/land/slaves (I think all non-tribal civilisation cared about the first two, and most cared about the third), or the idea that land has no value. I wonder what history would look like if the US had not expanded west? I think the most obvious change is that Germany and/or Japan would have won WWII. The US would have had little oil (as the Texan oil fields were not part of the initial colonies) no pacific naval bases, no empty land to test nuclear bombs on etc.... (read more)
I haven't posted on LW for a while now, but after posting to LW2.0 I got banned (till 2021) very quickly. My posts were also deleted. I was not told why I was banned, although I assume it was because I entered a fake email (I was annoyed that the new site required an email and not just a username). I asked why I got banned, and received no response.
Well, I'm sorry for using a fake email. I wasn't trying to spam, or sockpuppet or anything, but I think a 3 year ban without any sort of warning or explanation seems a little excessive. I'm happy to provide an email that works if required. I also considered setting up a new account with a real email, but I don't want it to seem like I'm sockpuppetting.
Facebook is full of bullshit because it is far quicker to share something then to fact-check it, not that anyone cares about facts anyway. A viral alarmist meme with no basis in truth will be shared more then a boring, balanced view that doesn't go all out to fight the other tribe.
But Facebook has always been full of bullshit and no-one cared until after the US election when everyone decided to pin Trump's victory on fake news. So its pretty clear that good epistemology is not the genuine concern here.
Not that I'm saying that Facebook is worse then any other social media - the problem isn't Facebook, the problem is human nature.
This seems like an impressive first step towards AGI. The games, like 'pong' and 'space invaders' are perhaps not the most cerebral games, but given that deep blue can only play chess, this is far more impressive IMO. They didn't even need to adjust hyperparameters between games.
I'd also like to see whether they can train a network that plays the same game on different maps without re-training, which seems a lot harder.
Why AGI is extremely likely to come before FAI
I am using FAI to mean EY's definition of provably friendly AGI implementing CEV. It is possible that the AGI which comes first will not be *provably* friendly, but will never the less turn out to be friendly, although the probability that non-provably friendly AI turns out to be friendly depends on how it is programmed. But this is a subtopic for another article - Part II: what to do if FAI is unlikely to come first.
I realize much of this has been said before.
Complexity of algorithm design:
Intuitively, FAI seems orders of magnitude more complex than AGI. If I decided to start trying
>this comment is pretty introspective and open to criticism, which I think is even more LW-y than being "smart".
Thanks.
I'm not going to delete my posts on second thoughts, it would not be fair to those who have replied to me. I sort of feel that all political posts should just be moved to a walled off area for cognitive hazards. I am however, committing to staying away from politics.
Perhaps I will create a new account.