I am AI Enablement Manager at Epoch AI.
After living nomadically for many years, I recently moved back to my native Buenos Aires. Feel free to get in touch if you are visiting BA and would like to grab a coffee or need a place to stay.
Every post, comment, or wiki edit I authored is hereby licensed under a Creative Commons Attribution 4.0 International License.
No matter what Anthropic does, it seems folks always find a way to say something negative about them.
It’s up. (It was down when I left my comment.) Great!
I’d be interested in seeing you guys elaborate on the comments you make here in response to Rob’s question that some control methods, such as AI boxing, may be “a bit of a dick move”.
An additional reason EAs may not be playing the attention arms race is that they may be persuaded by the fidelity model of spreading ideas.
Nope, it was Yudkowsky in a Facebook group about AI x-risk around 2015 or 2016. He specifically said he didn’t think deep learning was the royal road to AGI.
Would you be able to locate the post in question? If Yudkowsky did indeed say that, I would agree that it would constitute a relevant negative update about his overall prediction track record.
Three years later, this seems dead, unfortunately.
Do you think you would have become involved with EA if your name hadn’t been ‘Pablo’?
(But seriously, big kudos for the book. The Spanish-speaking EA community is fortunate to have such a personable and articulate ambassador!)
There is a monthly newsletter for updates in the Spanish-speaking EA community.
Gotcha, so to be clear, you’re saying: it would be better for the current post to have the relevant quotes from the references, but it would be even better to have summaries of the explanations?
Yes, that’s what I’m saying.
(I tend to think this is a topic where summaries are especially likely to lose some important nuance, but not confident.)
I defer to you, since I am not familiar with this topic. My above assessment was “on priors”.
Could you please clarify what you mean by this?
I was referring to the difference in value between a collection of references and a summary of the content of those references (as opposed to a mere collection of representative quotes).
I’d personally find this helpful, and I expect others will, too. If I consider the FAQs I’m familiar with and imagine alternative documents that consist of the questions and the references, but without the answers, I feel that their value decreases by at least 50%. Most of the added value comes from the synthesis, but some comes from removing the trivial inconvenience of having to open multiple links and locating the relevant passage(s).
I think Diana Fleischman was one of the first people in EA to advocate for a “bivalvegan” diet, back in 2013 or so, and has written some blog posts about it (e.g.).
Sure, who could possibly believe that all moral propositions are objectively true? My point was that moral realists typically believe that some axiological and some deontic claims are objectively true, and that if you are an anti-realist about the former and a realist about the latter, calling yourself a “moral realist” may fail to communicate your views accurately.
Morality is Objective
I do not like the expression ‘Morality is objective’, because it comprises both claims I’m very confident are not objective (“You ought not to kill”) and claims I’m very confident are objective (“Suffering is bad”). More generally, I am a moral anti-realist by default, but am forced to recognize that some moral claims are real—specifically, certain axiological claims—because their objective reality is revealed to me via introspection when I have the corresponding phenomenal experiences (such as the experience of being in agony).
This is basically my view, and I think ‘axiological realism’ is a great name for it.
Have you considered translating it in other major languages, especially those with large existing EA communities like German, French or Spanish, or EA potential?
As it happens, we contacted CEA a few days ago and offered to translate the website into various languages. Some of this material is already translated, but it would be great to have the entire website, with its appealing design, available on dedicated domains for all the main language communities.
Yeah, sorry: it was obvious to me that this was the intended meaning, after I realized it could be interpreted this way. I noted it because I found the syntactic ambiguity mildly interesting/amusing.
If you are like me, this comment will leave you perplexed. After a while, I realized that it should not be read as
Plausibly, best people who have ever lived is a much lower bar than best people who have ever lived.
but as
‘Plausibly best people who have ever lived’ is a much lower bar than ‘best people who have ever lived’.
Scott Alexander introduces the ‘noncentral fallacy’ as follows: “X is in a category whose archetypal member gives us a certain emotional reaction. Therefore, we should apply that emotional reaction to X, even though it is not a central category member.”
This post seems like an archetypal instance of the noncentral fallacy.
Sorry, my comment wasn’t addressed to you in particular. It should probably have been a top-level comment; I posted it as a reply only because your comment was an example (among many) of the phenomenon I was describing. I also oppose mass surveillance, and it makes zero difference to me whether or not the people surveilled comprise the tiny fraction of the world population that happens to be American.
I just find it frustrating that the critical comments directed at Anthropic often fail to grapple with the complexity of the situation and the hard tradeoffs they face.