marcymakesgames

I personally would care if studies are peer reviewed and from higher quality institutes because it comes with more credibility than any other arbitrary thing on Reddit or whatever social media. Sure, accurate information can come from outside those sources, but no one has the time or resources to validate every single thing they see.

I'm not hung up on the LLM as a component here, more just curious about this specific instance because I (in all my lack of understanding of this sort of thing) don't really know if I would call it ethical or unethical. What sticks out to me is that the subreddit they used for the study is one where users specifically go to to test their viewpoints against arguments from others whom they would most likely believe are people. In this instance, we could say 'oh well of course there are bots and misinformation online all the time', but I don't know if it's reasonable to assume a person going there for debate should expect full, outright lies (en masse) to be persuaded, particularly because it's against the posted rules and the mod team (presumably) works to curtail that. If that were the case and the expectation is most people aren't telling the truth, that sub wouldn't exist as it does .I think that part is especially grey for me.

Still, I get what you're saying, that ethics committees are primarily about protecting physical health and informed decision making, it just seems a bit different when someone outright says, "Convincing me will change my outlook and behavior going forward," and your study is based on lying to those people and not later informing them that you lied.

I think I'm following? So wouldn't the real crux of the issue be whether or not the users know they're talking to real people?

Like, if you dropped a bunch of well-documented, peer reviewed studies from prestigious institutions in someone's lap, but didn't tell them you changed a bunch of the contents to support fascist ideologies and then they became fascist, that would be an issue, right?

In this specific case, the LLMs are just straight up lying and portraying themselves as people in attempt to change people's opinions. That feels a bit like it treads on the informed consent grounds, doesn't it?

Can I ask why it's unethical? I don't have a background in science or academia, but it seems like more data about how easily LLMs can manipulate social media would be a good set of data points to have, wouldn't it?

Maybe I'm conflating ideas, but I feel like there's already some understanding that bots are all over social media to begin with, but again, my understanding of scientific ethics is basically the just a discussion about the Stanford Prison Experiment in school, so I'm for sure out of my depth with this.

I mean, mostly, yeah. The issue is, while most people will say dude is gender neutral, it really depends on the person.

I'm trans, I have two friends who call friends dude. One is a woman who I have known for like 5 years and will call everyone and everything dude. When she says it I don't feel weird about it.

I have another friend who says dude is gender neutral and I've known them for about the same amount of time. However, I have never heard them call a woman dude. They say they do, but I our mostly female friend group, I think I've only ever heard it towards me.

That's really the issue. People will say it's gender neutral but not use it neutrally, and believe me, trans people notice it. Anecdotally, this is how my trans friends view it as well, but take my experience with a grain of salt.