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 have noticed. A lot of hobby spaces have actually banned AI art, and if it's not banned, it's treated with extreme disdain. Any boardgame publisher caught using AI art, for example, gets a social media pile-on. It's a big deal in RPG and self-publishing (where authors and publishers operating on a shoestring obviously find it very tempting to cut costs with AI art). A few traditional publishers have caught flak for using AI art on their covers.
As a bit of an AI enthusiast (I even bought a chunkier GPU for Stable Diffusion), I obviously do not buy the "unethical" argument, but this has become kind of like "Actually, I think racial IQ differences might be real" - not something I can talk about openly with a lot of my circle.
There are a lot of anti-AI arguments, and the ethical/copyright issues are ambiguous, but the bottom line is that artists are, rightfully, afraid of being replaced by a machine. When they complain about how AIs were trained "unethically", ask them if the algorithms improve so much that an AI can be trained entirely on open source or public domain artwork (there have been some efforts to create so-called "ethical" AI models) and produce equivalent results, if they'd be okay with that? They will usually hem and haw and hope that doesn't happen.
I do feel a little bad for artists. I mean, if you had a decent side hustle charging $50 to draw D&D characters, or a more lucrative side hustle drawing furry porn, AI is going to replace you. High end artists will still have jobs, and AI can't really do competent composition or graphic design or a series of pictures with a consistent theme (yet), but the DeviantArt and ArtStation kids are getting hungry and desperate.
It ultimately boils down to money, and they are trying to make it a moral crusade to preserve their livelihood. It is only the threat of being dragged on social media that's preventing more publishers and companies from using AI art, and as AI art gets better and less easily detectable, and more widely accepted, that will change.
I will say that a lot of AI art is just lazy. Like, if you just give a prompt, run 50 iterations, take the best one, and slap it on your cover, it's still probably not going to look very good and it will look like obvious AI generation. Even for my hobby art I do some photoshopping and have learned enough composition to blend elements together - it might still be detectable, but it doesn't just scream "AI." (Then again, I'm not generating anime waifus or furry porn, which is like 90% of AI art as far as I can tell.)
This is coming to other industries as well. Audiobooks, for example, are now pretty lucrative for most authors, and AI voices are becoming nearly as good as human voice actors - and human voice actors are expensive. For self-published authors, it's a no-brainer economically, so narrators and readers are doing their best to make it morally unacceptable to use AI. If the disapprobation fails to kill sales, that entire niche is going to be dead.
It will be some time before AI can replace a lot of other industries, but low-end software development, customer service, and other industries are already being affected. This is what the artists are fighting - not subjective esoteric notions of whether AI art has "soul" or qualifies as "good art."
I had J.C. Denton read me the Unabomber manifesto. 10/10, would critique industrial civilization again.
lmao, thanks for this.
More options
Context Copy link
I'm laughing just imagining what that would sound like.
More options
Context Copy link
More options
Context Copy link
Are you referring to Gemini and its reported 1m context window? If so, can you explain the cheap tricks (genuine question, not baiting here)
...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
More options
Context Copy link
I agree that human actors still do a better job, but even commercial TTS is way better than the robotic Kindle voice of a few years ago, and lots of people were happy to use that. They've got commercial software now for choosing voices and adding tone, emphasis, etc. wtih markup.
It's not quite human-level yet, but voice actors are right to be scared.
More options
Context Copy link
I haven't investigated the AI art options too closely yet, but it seems to me that it would be really useful if I could sketch out a composition and have the AI make it look good. Does this capability not exist yet?
I think you're referring to a style transfer, which could produce remarkably impressive results even five years ago.
More options
Context Copy link
Fundamentally, this has existed for about 2 years, though the software to make it easy to do is more recent. I haven't used Photoshop, but I believe it essentially does that with Firefly, and for free tools, the Krita (freeware) extension to use Stable Diffusion does this pretty well. However, actually getting a "good looking" picture out of it is still something that's not likely to be a one-step process, but rather requiring iterations and intentional inpainting.
What you're talking about is a version of what's referred to as IMG2IMG, which is exactly what it sounds like, and, in fact, it's actually the same thing as TXT2IMG, just, instead of starting with random noise in the case of the latter, you're starting with an image that you sketched. Early on, keeping the structure of the original image was a major struggle, but something like 1.5 years ago, a tech referred to as "ControlNet" was developed, which allowed the image generation to be guided by further constraints beyond just the text prompt and settings. Many different versions of ControlNet exist, including edge detection, line-art, depth map, normal map, and human pose, among others. In each, those particular details from the original image can be used to constrain the generation so that objects you might draw in the foreground don't blend in to the background, or so that the person you drew in a certain pose comes out as a human in the exact same pose. It's possible to run multiple of these at the same time.
Again, in practice, these aren't going to be one-step solutions, with various issues and weaknesses that need manual work or further iterations to make look actually like a good work of art. But in terms of turning, say, a crude mess of blobs into something that looks somewhat realistically or professionally rendered while following the same composition, it's quite doable.
More options
Context Copy link
Yes, it's called in-painting, and Stable Diffusion even has a sketch mode that will turn stick figures into paintings.
More options
Context Copy link
More options
Context Copy link
Me too, but I don't really know what is to be done about it.
I agree with your assessment, it probably is all about money and fear of replacement at the end of the day. But it really is coming for all of us. What are all of us going to do? I merely hope that either we have time to learn new trades that won't get replaced again within our lifetimes, or I hope we will enter a post scarcity society where AI has made everything so cheap that money is meaningless.
I'm a pretty firm believer that the only thing that keeps life so good for so many people is that technological progress keeps making things cheaper and better. And really, I don't even know if I can say that I think that efforts to stop technology through the use of social stigma are bad. Instead I just feel like they're simply doomed to failure. People are going to follow the money. Maybe nuclear power is the one big exception I know of.
The number of people displaced in this manner is probably so trivial anyway. Illustrators tend to serve specific needs for clients, not something easily reproduced by ai. AI art is just the latest iteration of the stock photo.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link