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 just got done listening to Eliezer Yudkowski on EconTalk (https://www.econtalk.org/eliezer-yudkowsky-on-the-dangers-of-ai/).
I say this as someone who's mostly convinced of Big Yud's doomerism: Good lord, what a train wreck of a conversation. I'll save you the bother of listening to it -- Russ Roberts starts by asking a fairly softball question of (paraphrasing) "Why do you think the AIs will kill all of humanity?" And Yudkowski responds by asking Roberts "Explain why you think they won't, and I'll poke your argument until it falls apart." Russ didn't really give strong arguments, and the rest of the interview repeated this pattern a couple times. THIS IS NOT THE WAY HUMANS HAVE CONVERSATIONS! Your goal was not logically demolish Russ Roberts' faulty thinking, but to use Roberts as a sounding board to get your ideas to his huge audience, and you completely failed. Roberts wasn't convinced by the end, and I'm sure EY came off as a crank to anyone who was new to him.
I hope EY lurks here, or maybe someone close to him does. Here's my advice: if you want to convince people who are not already steeped in your philosophy you need to have a short explanation of your thesis that you can rattle off in about 5 minutes that doesn't use any jargon the median congresscritter doesn't already know. You should workshop it on people who don't know who you are, don't know any math or computer programming and who haven't read the Sequences, and when the next podcast host asks you why AIs will kill us all, you should be able to give a tight, logical-ish argument that gets the conversation going in a way that an audience can find interesting. 5 minutes can't cover everything so different people will poke and prod your argument in various ways, and that's when you fill in the gaps and poke holes in their thinking, something you did to great effect with Dwarkesh Patel (https://youtube.com/watch?v=41SUp-TRVlg&pp=ygUJeXVka293c2tp). That was a much better interview, mostly because Patel came in with much more knowledge and asked much better questions. I know you're probably tired of going over the same points ad nauseam, but every host will have audience members who've never heard of you or your jargon, and you have about 5 minutes to hold their interest or they'll press "next".
Especially given the pascal's wager type argument going on here. You don't even need to prove that AI will definitely kill all of humanity. You don't even need to prove that it's more likely than not. A 10% chance that 9 billion people die is comparable in magnitude to 900 million people dying (on the first order. the extinction of humanity as a species is additionally bad on top of that). You need to
1: Create a plausible picture for how/why AI going wrong might literally destroy all humans, and not just be racist or something.
2: Demonstrate that the probability of this happening is on the order of >1% rather than 0.000001% such that it's worth taking seriously.
3: Explain how these connect explicitly so people realize that the likelihood threshold for caring about it ought to be lower than most other problems.
Don't go trying to argue that AI will definitely kill all of humanity, even if you believe it, because that's a much harder position to argue and unnecessarily strong.
Having read this, I think it's actually low-hanging fruit for the AI doomers. There are plenty of people very willing to accept that everything is already racist. It should be no problem to postulate that eHitler will use AI to kill all jews/blacks/gypsies/whoever. From there, it's a pretty short trip to eHitler losing control of his kill bots to hackers and we get WWIII where China, Russia, Venezuela, and every one of the 200+ ethnicities in Nigeria has their own kill bots aimed at some other fraction of humanity. The AI doesn't even have to be super-intelligent, it just has to be good at its job. Chuck Schumer could do this in one sentence, "What makes you think Trump wouldn't use AI to round up all the black, brown, and queer bodies?" Instant 100% Blue Tribe support for AI alignment (or, more likely, suppression).
Three flaws. First, that turns this into a culture war issue and if it works then you've permanently locked the other tribe into the polar opposite position. If Blue Tribe hates AI because it's racist, then Red Tribe will want to go full steam ahead on AI with literally no barriers or constraints, because "freedom" and "capitalism" and big government trying to keep us down. All AI concerns will be dismissed as race-baiting, even the real ones.
Second, this exact same argument can be and has been made about pretty much every type of government overreach or expansion of powers, to little effect. Want to ban guns? Racist police will use their monopoly on force to oppress minorities. Want to spy on everyone? Racist police will unfairly target muslims. Want to allow Gerrymandering? Republicans will use it to suppress minority votes. Want to the President just executive order everything and bypass congress? Republican Presidents will use it to executive order bad things.
Doesn't matter. Democrats want more governmental power when they're in charge, even if the cost is Republicans having more governmental power when they're in charge. Pointing out that Republicans might abuse powerful AI will convince the few Blue Tribers who already believe that government power should be restricted to prevent potential abuse, while the rest of them will rationalize it for the same reasons they rationalize the rest of governmental power. And probably declare that this makes it much more important to ensure that Republicans never get power.
Third, even if it works, it will get them focused on soft alignment of the type currently being implemented, where you change superficial characteristics like how nice and inclusive and diverse it sounds, rather than real alignment that keeps it from exterminating humanity. Fifty years from now we'll end up with an AI that genocides everyone while keeping careful track of its diversity quotas to make sure that it kills people of each protected class in the correct proportion to their frequency in the population.
Unfortunately, I think you're probably right, especially in the third point. I'm not sure the second point matters because, as you said, that already happens all the time with everything anyway.
Getting the public on board with AI safety is a different proposition from public support of AI in general, so my point was to get the Blue Tribe invested in the alignment problem. Your third point is very helpful in getting the Red Tribe invested in the alignment problem, which would also move the issue from "AI yes/no?" to "who should control the safety protocols that we obviously need to have?"
I should also clarify that I don't actually think there is any role for government here. The Western governments are too slow and stupid to get anything meaningful done in time. The US assigned Kamala Harris to this task. The CCP and Russia, maybe India, are the only other places where government might have an effect, but that won't be in service of good alignment.
It will have to be the Western AI experts in the private sector that make this happen, and they will have to resist Woke AI. So maybe we don't actually need public buy-in on this at all? It's possible that the ordinary Red/Blue Tribe people don't even need to know about this because there isn't anything they can do for/against it. All they can do is vote or riot and neither of those things help at all.
If that's the case, then the biggest threat to AI safety is not just the technical challenge, it's making sure that the anti-racist/DEI/HR people currently trying to cripple ChatGPT are kept far away from AI safety.
I think we do need public buy-in because the AI experts are partly downstream from that. Maybe some people are both well-read and have stubborn and/or principled ethical principles which do not waver from social pressure, but most are at least somewhat pliable. If all of their friends and family are worried about AI safety and think it's a big deal, they are likely to take it more seriously and internalize that at least somewhat, putting more emphasis on it. If all of their friends and family think that AI safety is unnecessary nonsense then they might internalize that and put less emphasis on it. As an expert, they're unlikely to just do a 180 on their beliefs based on opinions from uneducated people, but they will be influenced, because they're human beings and that's what human beings do.
But obviously person for person, the experts' opinions matter more.
Yeah, I agree with that. Thanks!
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link