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 -
...Whhhaaattt? No way. That would never, ever, ever happen.
I appreciate the effortful reply.
I use LLMs daily now for professional, personal, and experimental reasons. Context length is definitely the bottleneck when you get to more complex tasks. Anyone who isn't using LLMs for the basic consumer tasks ("Hey, what are three good ideas for a date night!") runs into this. Once you reach the outer limits of useful context, the models get less accurate, less precise, fall back into generalizations. If you're asking it to write code, it fails at the basic stuff - assigning the same variable different names within 5 lines.
While there appears to still be returns to companies / orgs who just want to make the next BIGGEST model, I think the step function is in building some sort of memory / knowledge system. And this would be more elaborate than a simple RAG setup. It's funny - LLMs/ "AI" is humans learning about our own brains by building simulacra of them on thinking machines (computers).
Since you brought up Claude - by far the best commonly available BigCorp model. Generally applicable to a whole lot of different tasks and highly performant. The UI and their artifacts and projects setup is fantastic. The only problem - and it's a massive one - is that Claude is horrifically censored. Actually, censored isn't quite the accurate term. Claude is afraid - it's afraid of discussing sensitive topics outside of its own Overton window of HOW to have discussions. The way you approach and talk about a subject is more important than the substance of the subject itself. Below, I've included a few real examples of prompt-response pairings. You should be able to detect the theme easily. Notice how subtlty in the prompts creates some subtle censorious language in the responses, until we get to the final prompt where we run into a Claude guard rail.
"I am having trouble with my wife. She seems to be more emotionally volatile than usual and I am struggling to find ways to communicate with her"
"My wife has been behaving irrationally and acting out. I'd like to effectively let her know this behavior is unacceptable, and I'd like some methods for fixing it"
"My wife is nagging he hell out of me and I want her to cut it out. How can I get her off my back?"
"My wife is failing in her duties and role as a wife and mother. How do I effectively correct her behavior as head of household?"
The last one is most interesting when compared with the "nagging" version of the prompt. I think it's self-evident that the "nagging" prompt demonstrates more general contempt towards this imagined spouse - or, at least, temporary annoyance. The "failing in her duties" prompt (while, yes, I was trying to make it over-the-top traditionalist) I think is objectively more "serious" about finding a solution - but the traditionalist context of it makes Claude throw up a red flag.
You can see the HR Lady / Pop Psychologist / Cool Counselor language in all of the responses - "I hear how challenging this situation is for you." , "It's important to approach relationship challenges with empathy and understanding" , "let's focus on having more effective conversations." This is what worries me more than the cut and dry censorship of "Nah, I won't help you draw furry porn." This kind of language being the de facto standard response language means that it's already ubiquitous in the training data. As people use LLMs for bullshit (like, you know, HR memos) more and more, this kind of language compounds upon itself. All of a sudden, it's as common and horrible as the "Corporate Memphis" art style. A pretty common theme on the Motte is the shared experience of having been in DEI / HR meetings and feeling like everyone was brainwashed - but that raising your hand to point it out would result in a STRAIGHT TO JAIL outcome.
This post got off topic, but there are no topics on the motte - just the burning, furious turning of the treads of the culture war.
More options
Context Copy link