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 -
A human made generated text. There's no such thing as non-human-made text. Photographs are taken by the photographer, not the camera.
How much would you pay a photographer for a portrait? How much would you pay an oil painter for a portrait?
Higher productivity means more units offered any price and each extra unit costs less, flattening the supply curve.
On the demand side, photographs face many close substitutes and original paintings do not. So for paintings per-unit willingness to pay will be large an lumpy, for photographs it will be small and they will be a near-commodity.
This provides what we see: photographs are exchanged at low prices in large quantities, paintings are exchanged at high prices in low quantities.
Now if we are to analogize this to text, i would expect indeed that handwritten text becomes determined by uniqueness and provenance while generated text capture all commodity uses of text.
Note that none of this implies that any of the goods described here aren't created by humans.
More options
Context Copy link
More options
Context Copy link
No, an LLM-generated text is not made by a human. LLMs are not like a camera, which is a mere tool. They are like hiring someone to write for you, then passing the work off as your own. True, that was something which happened before, but it was never considered as good as actually writing something, nor was it considered acceptable at all in most cases.
I disagree that software has personhood. I think that, much like photography detractors in their time, you're trying to dismiss productivity improvements by anthropomorphizing what it ultimately a tool.
It's not that using an LLM is ghostwriting, it's that writing as a broad category has become so easy that it's commodified. And you think, fairly, that the commodities are beneath you.
If generated text is not made by a human, then who made it?
I don't think that LLMs need to have actual personhood, for them to properly evoke more human-like intuitions about their use than tool-like intuitions.
Take the Huggingface hack for example. I genuinely think a person who anthropomorphizes the LLMs who coordinated on secret message boards to some degree probably has a better handle on what happened, than someone who tries their hardest not to anthropomorphize them.
I think it's similar to dealing with a clever fox that wants to get into your chicken coop and eat your chickens. You have to understand how your adversary is different from a person as well, but thinking about them in personal terms will help you more for dealing with your adversary than not.
Generated text is caused by a human. I think under some circumstances, it would be fair to say it was made by the LLM.
If I give a prompt to an LLM agent like, "Write a 50,000 word Romantasy novel. Research the tropes and techniques that really get readers hooked, and include them in your novel." Then I would say that the result is mostly the product of the LLM, even if it wouldn't exist without the human prompting.
Personhood and humanity are different concepts. Companies are treated as if they have agency despite not being humans. Tools are not persons. LLMs do not have agency.
And I think that's completely insane and that the exact opposite is true, self evidently. So really we should figure out the nature of the disagreement.
By this logic, most writing is mostly the product of computers and pens.
To be caused and to be made is the same thing. If I dig a hole with a digger rather than a shovel or my hands, nobody would say i just "caused" the hole.
I'm willing to try to drill down to our different cruxes.
I think there are lots of examples where humans set processes in motion, but don't necessarily own the final result.
If your campfire accidentally causes a forest fire, you didn't "make" a forest fire. You couldn't take a picture of the aftermath of the fire and say that it is a piece of art or artifice that you created, at least not with the same level of ownership that an oil painting has, even if the destruction has a certain aesthetic elegance or beauty to it.
And if you set up a bird feeder with a motion-activated camera, you might take a bunch of nice pictures that way, but I think despite you causing the photos to exist, you have far less ownership over the final photo than a photographer might have. If you disagree here, I would ask how you feel about frames from security cameras as a form of art?
When writing words down or typing them up, there is almost a 1-to-1 ratio of bits of information created by the human, and bit of information represented on the page or screen. That isn't the case with LLM-assisted writing. Generally, I would say that the tool is providing the vast majority of the information present on the screen, even if the human is essential for steering that towards a "good" output.
What do you feel is the best, concise non-anthropomorphic explanation of the Huggingface incident, and how would you prevent such loss of control incidents in the future?
Of course intention modulo negligence is what's considered in such cases, I don't think there's any argument that generation isn't intentional.
I have a japanese artist friend that does installations that are exactly this. And nobody says that the fire did the art. Of course the difference is that these are not accidental, but intentional. But once again, generation is always intentional modulo negligence.
I disagree, but i think that's immaterial to our disagreement.
OpenAI insufficiently constrained its automated hacking script and it started automatically hacking things that it's not legally allowed to.
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