ben.smith

Karma: 956

ben.smith Jun 7, 2023, 5:57 PM
1 point
0 ∶ 0
in reply to: pseudonym’s comment on: “Pronatalists” may look to co-opt effective altruism or longtermism
I’ve stumbled here after getting more interested in the object-level debate around pronatalism. I am glad you posted this because, in the abstract, I think it’s worthwhile to point out where someone may not be engaging in good faith within our community.
Having said that, I wish you had framed the Collins’ actions in a little more good faith yourself. I do not consider that one quoted tweet to be evidence that of an “opportunistic power grab”. I think it’s probably a bit unhealthy to see our movement in terms of competing factions, and to seek wins for one’s own faction through strategic means rather than through open debate.
But I’m not sure Malcolm Collins is quite there, on the evidence you’ve said. It seems like he’s happy that (according to him) his own favored cause area will get more attention (in the months since this has been posted, I don’t think his prediction has proven correct). I don’t think that’s the same as actively seeking a power grab—it might just be a slightly cynical, though realistic, view that even in a community that tries to promote healthy epistemics, sociological forces are going to have an influence on what we do.

ben.smith Jun 7, 2023, 5:21 AM
1 point
0 ∶ 0
in reply to: isabel🔸’s comment on: Explore the new UN demographic projections to 2100
Have you looked at the fertility rate underlying the UN projections? They’re projecting fertility rates across China, Japan, Europe, and the United States to arrest their yearly decline and begin to slowly move up back to somewhere in the 1.5 to 1.6 range.

That seems way too high because it’s assuming not just that current trends stop but that they reverse to the opposite direction of that observed. Even their “low” scenario has fertility rebounding from a low in ~2030.
This despite all those countries still have a way to go before they get to the low South Korea has reached at 0.88.
See https://ourworldindata.org/explorers/population-and-demography?time=2000..latest&facet=none&country=High-income+countries~OWID_WRL~NGA~CHN~IND~Europe+%28UN%29~USA~JPN~KOR&hideControls=true&Metric=Fertility+rate&Sex=Both+sexes&Age+group=Total&Projection+Scenario=Low

ben.smith Jun 4, 2023, 1:27 AM
9 points
1 ∶ 0
on: The Control Problem: Unsolved or Unsolvable?
I enjoyed this post. I think it is worth thinking about whether the problem is unsolveable! I think one takeaway I had from Tegmark’s Life 3.0 was that we will almost certainly not get exactly what we want from AGI. It seems intuitively that any possible specification will have downsides, including the specification to not build AGI at all.
But asking for a perfect utopia seems a high bar for “Alignment”; on the other hand, “just avoid literal human extinction” would be far too low a bar and include the possibility for all sorts of dystopias.
So I think it’s a well-made point that we need to define these terms more precisely, and start thinking about what sort of alignment (if any) is achievable.
I might end up at a different place than you did when it comes to actually defining “control” and “AGI”, though I don’t think I’ve thought about it enough to make any helpful comment. Seems important to think more about though!

ben.smith Mar 6, 2023, 10:12 PM
1 point
0 ∶ 0
in reply to: Habryka [Deactivated]’s comment on: Please don’t criticize EAs who “sell out” to OpenAI and Anthropic
I have a very uninformed view on the relative Alignment and Capabilities contributions of things like RLHF. My intuition is that RLHF is positive for alignment I’m almost entirely uninformed on that. If anyone’s written a summary on where they think these grey-area research areas lie I’d be interested to read it. Scott’s recent post was not a bad entry into the genre but obviously just worked a a very high level.

ben.smith Mar 6, 2023, 5:08 AM
6 points
2 ∶ 0
in reply to: more better ’s comment on: Who Aligns the Alignment Researchers?
Can you describe exactly how much you think the average person, or average AI researcher, is willing to sacrifice on a personal level for a small chance at saving humanity? Are they willing to halve their income for the next ten years? Reduce by 90%?

I think in a world where there was a top down societal effort to try to reduce alignment risk, you might see different behavior. In the current world, I think the “personal choice” framework really is how it works because (for better or worse) there is not (yet) strong moral or social values attached to capability vs safety work.

ben.smith Feb 9, 2023, 3:24 PM
4 points
1 ∶ 1
on: In (mild) defence of the social/professional overlap in EA
Well written, thanks!

ben.smith Feb 7, 2023, 3:36 PM
2 points
0 ∶ 0
in reply to: Geoffrey Miller’s comment on: EA’s weirdness makes it unusually susceptible to bad behavior
Here’s something else I’d like to know on that survey:
- what proportion of respondents wants to post on EAF or engage in other discussions they think are important for EA’s goals, but don’t, or will only do so anonymously because they are worried about the consequences?
- how does that compare to the proportion who feel free to contribute without fear of retribution?
- what proportion thinks they have been in fact passed over for an opportunity because they have criticized EA or said something else “politically incorrect” here?

ben.smith Feb 5, 2023, 10:24 PM
9 points
4 ∶ 0
in reply to: Arturo Macias’s comment on: EA’s weirdness makes it unusually susceptible to bad behavior
Surveys of these types are often anonymous, because
- while it is possible for people to make false responses, that doesn’t happen very much, because it is time consuming, and unethical, and there just aren’t that many people out there who are all of unethical, have lots of time on their hands, and want to manipulate our survey. Manipulated responses are generally more of a danger for short polls (e.g., “which political party would you vote for”), but less of an issue for 10 minute + surveys.
- there are means of probabilistically filtering false responses out, including eliminating identical copies of responses, comparing IP addresses, and so on
- It is quite expensive, difficult, and risky to verify identities and at the same time, guaranteeing anonymity
- For that reason verifying identities can discourage genuine responses

ben.smith Feb 5, 2023, 8:05 PM
15 points
9 ∶ 0
in reply to: Geoffrey Miller’s comment on: EA’s weirdness makes it unusually susceptible to bad behavior
This is a great idea. EA already runs an annual community survey. So it wouldn’t necessary to create a whole new survey to get this data—just add some questions to the existing community survey. If they aren’t already on there it would be great to see them on the next survey.

ben.smith Feb 5, 2023, 1:50 AM

1 point

0 ∶ 0

Error

The value
  NIL
is not of type
  SIMPLE-STRING
when binding #:USER-ID162

ben.smith Feb 5, 2023, 1:27 AM

11 points

10 ∶ 0

Error

The value
  NIL
is not of type
  SIMPLE-STRING
when binding #:USER-ID162

ben.smith Feb 4, 2023, 11:10 PM

17 points

6 ∶ 0

Error

The value
  NIL
is not of type
  SIMPLE-STRING
when binding #:USER-ID162

ben.smith Feb 4, 2023, 9:25 PM

15 points

6 ∶ 2

Error

The value
  NIL
is not of type
  SIMPLE-STRING
when binding #:USER-ID162

ben.smith Feb 3, 2023, 10:02 PM
11 points
4 ∶ 1
in reply to: Monica’s comment on: EA, Sexual Harassment, and Abuse
I agree with you that mixing romantic relationships with professional ones occurs among people who are monogamous or don’t identify as polyamorous.
I personally wouldn’t like to see EAs discouraged from being polyamorous. I’m not actively polyamorous myself, but I wouldn’t want to see people restricted to more traditional romantic styles, like monogamous marriage, because I think many of those relationship styles developed in a very different social and technological context than the context we have now. Our culture at large probably benefits from people pioneering and exploring relationship styles that are more suited to our current sociotechnological context. In addition, I think it would be somewhat of a human rights issue for, say, employers in the movement to be telling people how to order their romantic lives.
That said, what I mean to say was that if your romantic life involves more people—which I think it can in polyamory (that is the aim for many, perhaps!), you’ll have a larger and more complex web of romantic connections. If those people are also those you have professional connections with, then there is the potential for your professional and romantic webs to overlap. While this can also happen to people in monogamous relationships, to the extent they are romantically involved with fewer people, and their romantic networks are smaller, there’s less potential for an overlap of their professional and romantic lives.
And while I think an interaction of professional and romantic lives isn’t inherently wrong, it facilitates conflicts of interests, and even more importantly, power dynamics that can facilitate abuse or coercion.
I don’t want to dismiss deep platonic friendships, but I’d ask to agree to disagree on the relevance of those, perhaps, because just about by definition, they will not involve physical or sexual abuse, and so those aren’t consequences of coercive power dynamics that might arise in platonic friendships.

ben.smith Feb 3, 2023, 5:38 PM
36 points
31 ∶ 4
on: EA, Sexual Harassment, and Abuse
I think polyamory as it is described in the articles mixes complex webs of personal relationships with professional ones. Romantic connections within polyamorous communities can be complicated even without entanglement with professional concerns. When you bring in layers of professional connections on top of that, I can see why there might be an extra dimension of vulnerability to coercion and exploitation.

ben.smith Jan 25, 2023, 5:52 AM
13 points
4 ∶ 0
in reply to: ConcernedEAs’s comment on: Doing EA Better
A simple back-casting or systems-mapping exercise (foresight/systems-theoretical techniques) would easily have revealed EA’s significant exposure and vulnerability (disaster risk concepts) to a potential FTX crash. The overall level of x-risk is presumably tied to how much research it gets, and the FTX crash clearly reduced the amount of research that will get done on x-risk any time soon.
This is not the first time I’ve heard this sentiment and I don’t really understand it. If SBF had planned more carefully, if he’d been less risk-neutral, things could have been better. But it sounds like you think other people in EA should have somehow reduced EA’s exposure to FTX. In hindsight, that would have been good, for normative deontological reasons, but I don’t see how it would have preserved the amount of x-risk research EA can do. If EA didn’t get FTX money, it would simply have had no FTX money ever, instead of having FTX money for a very short time.

ben.smith Jan 25, 2023, 5:48 AM
4 points
0 ∶ 0
on: Doing EA Better
I was inspired by your post, and I wrote a post about one way I think grant-making could be less centralized and draw more on expertise. One commenter told me grant-making already makes use of more expert peer reviewers than I thought, but it sounds like there is much more room to move in that direction if grant-makers decide it is helpful.
https://forum.effectivealtruism.org/posts/fNuuzCLGr6BdiWH25/doing-ea-better-grant-makers-should-consider-grant-app-peer

ben.smith Jan 25, 2023, 12:14 AM
1 point
0 ∶ 0
in reply to: Jason’s comment on: Doing EA Better: grant-makers should consider grant app peer review along the public-sector model
At a pinch, I would say review might be more worthwhile for topics where the work builds on a well-developed but pre-existing body of research. So, funding a graduate to take time to learn about AI Safety full-time as a bridge to developing a project probably wouldn’t benefit from a review, but an application to develop a very specific project based on a specific idea probably would.
I don’t have a sense on how often five-to-low-six-figure grants involve very specific ideas. If you told me they usually don’t, I would definitely update against thinking a peer review would be useful in those circumstances.

ben.smith Jan 24, 2023, 9:19 PM
9 points
4 ∶ 0
in reply to: Davidmanheim’s comment on: Doing EA Better: grant-makers should consider grant app peer review along the public-sector model
Thank you, David! From what you’ve said here it seems clear my post was missing critical information.

I’m not sure this post literally could have been much better researched, conditional on me writing it. I don’t feel entitled to contact funders to ask them about their process (perhaps I should feel free to? I’m not sure). EA Funds website mentions briefly they “engage expert-led teams of subject matter experts” in their decision-making, and that’s something I should have researched first and mentioned, but also, I think that gives away so little information that I learned more from your reply here than I would have from reading that.

Perhaps other funders describe their process in more detail, I don’t know, and if so, I concede that’s something I could have identified before writing the post.

So the only other way I can see this post could have had more information is that I could have asked more widely with people more familiar with the process than I am. But I’m not personally acquainted with anyone who I knew would know more about the process.

Or, finally, I could have left it to someone else to write, but then, if they didn’t, I wouldn’t have learned from you that grant-makers already engage in expert consultation.

It probably is obvious to you, considering your experiences. But the way grant making was described on the Doing Good Better post last week—something to the effect of “it helps to move to the bay and make friends with grant-makers”—suggests to me the process is pretty opaque to a lot of other people, not just to me, and so I suppose I’m glad I opened a conversation even if I don’t have a lot of insight to share on the process.

Edit: There is a point I’m trying to make, other than defending my own process, which is that the process in general is fairly opaque, and if the information you’re talking about is publicly available, I’m not aware of it! And that validates something of the transparency critiques from the DGB post last week.

ben.smith Jan 24, 2023, 4:40 PM
3 points
3 ∶ 1
in reply to: MichaelDickens’s comment on: My highly personal skepticism braindump on existential risk from artificial intelligence.
Agreed, but 5% is much lower than “certain or close to certain”, which is the starting point Nuno Sempere said he was sceptical of.
I don’t know that anyone thinks doom is “certain or close to certain”, though the April 1 post could be read that way. 5% is also much lower than, say, 50%, which seems to be a somewhat more common belief.