This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
I think that likening the rationalists treatment of AI to the anti-finasteride crowd is a bit unfair to the former.
Now, AI has been a theme with rationalists from the very beginning. It would not be totally unfair to say that our prophet wrote the sequences (e.g. A Human's Guide to Words) as an instrumental goal to be able to discuss AI without getting bogged down in pointless definitional arguments. That was almost two decades ago, in the depth of the AI winter.
Scott Alexander wrote about GPT when it was still GPT-2, it was the first time I heard about it. It is fair to say that AI is the favorite topic on LessWrong, with Zvi minutely tracking the progress with the same dedication previously allocated to COVID. Generally, the rationalists are bullish on capabilities and bearish on alignment. But I feel that Eliezer's "dying with dignity strategy" haha-only-serious April's fool is overconfident in a way which is not typical of LW. In practical terms, it does not matter much if you think that p(ASI) is 0.15 and p(doom) is 0.1 or if you think they are 0.95 and 0.9 respectively.
We do not have a comprehensive theory of intelligence. We have noticed the skulls of the once who have predicted that AI would never beat a chess master, succeed at go, write a readable text, create a painting which most people can not distinguish from a human work of art and so on. This does not mean that AI will reach every relevant goalpost, reverse stupidity is not intelligence, after all.
We are in the situation where we observe a rocket launch without the benefits of any knowledge of rocketry or physics. Some people claimed the rocket would never reach an altitude of more than twice its own length, and they were very much proven wrong. Others are claiming that it would never reach 1km, and they were likewise proven wrong. From this, we can not conclude that it will obviously accelerate until it reaches Andromeda, nor can we conclude that it will not reach Andromeda.
Wrt the AI 2027, the vibes I remember getting from browsing through it is that it mostly Simulacrum level two, and came across as the least honest things which Scott ever co-authored. The whole national security angle is very much not what keeps LW up at night -- if China builds aligned ASI, they have a whole light cone to settle. What will happen to the US will just be a minor footnote in history. But the authors recognized that their target audience -- policymakers in DC -- will likely be alienated by their real arguments about x-risk. By contrast, national security is a topic which has been on the mind of the DC crowd for a century, so natsec was recruited as an argument-as-a-soldier.
Well, I am rather sure that there is a great rebuttal to these arguments somewhere on Less Wrong, so I remain unconvinced.
Well, the version of Pascal's wager offered by the AI safety people is that (1) the current AI boom might lead to AGI which is much smarter than humans are, and (2) that aligning such AI systems will be hard. You assign a probability to both of these, then multiply this by the QALY cost of killing all humans (or an even higher cost if you care about humanity's far future, which many do), and you get a number how seriously you should take AI x-risk.
Of course, you can simply pick your AGI probability to be 1e-50, but then I might claim that you are overconfident, and ask what other past correct predictions you have made which might make me rely on your predictions instead of everyone else's.
If you pick 1% for both numbers, then a one-in-10k chance to wipe out humanity still seems like a big fucking deal.
In Scott's last 24 non-OT posts on ACX, I have counted four AI stories (2x geoguesser, which is more "AI as a curiosity" and 2x AI 2027, which is more doom and gloom). While I am sure that some of the ACX grants go to AI safety, he is also funding plenty of other projects, which would be totally irresponsible if his p(doom) was 0.9. If this is him showing religious fervor, it is not very convincing.
I will concede that AI alignment is his pet thing much more than it is Scott's, and as of late he has been very bullish on p(doom). Still, I have found him to be a smart, engaging writer. Most of the ideas from the sequences could also be picked up elsewhere, but he did do a great job of communicating all these ideas and putting them in one place. For some light reading of his, see if you like HPMoR.
By and large, the ratsphere does not share his high confidence on p(doom), I think, because they were trained by their prophet to update based on the strength of arguments, not to blindly follow their prophet.
More options
Context Copy link
More options
Context Copy link