This is probably the most ethical you’ll ever see it. There are definitely organizations committing far worse experiments.
Over the years I’ve noticed replies that are far too on the nose. Probing just the right pressure points as if they dropped exactly the right breadcrumbs for me to respond to. I’ve learned to disengage at that point. It’s either they scrolled through my profile. Or as we now know it’s a literal psy-op bot. Already in the first case it’s not worth engaging with someone more invested than I am myself.
Over the years I’ve noticed replies that are far too on the nose. Probing just the right pressure points as if they dropped exactly the right breadcrumbs for me to respond to. I’ve learned to disengage at that point. It’s either they scrolled through my profile. Or as we now know it’s a literal psy-op bot. Already in the first case it’s not worth engaging with someone more invested than I am myself.
You put it better than I could. I’ve noticed this too.
I used to just disengage. Now when I find myself talking to someone like this I use my own local LLM to generate replies just to waste their time. I’m doing this by prompting the LLM to take a chastising tone, point out their fallacies and to lecture them on good faith participation in online conversations.
It is horrifying to see how many bots you catch like this. It is certainly bots, or else there are suddenly a lot more people that will go 10-20 multi-paragraph replies deep into a conversation despite talking to something that is obviously (to a trained human) just generated comments.
It’s easy to point to reasons why this study was unethical, but the ugly truth is that bad actors all over the world are performing trials exactly like this all the time - do we really want the only people who know how this kind of manipulation works to be state psyop agencies, SEO bros, and astroturfing agencies working for oil/arms/religion lobbyists?
Seems like it’s much better long term to have all these tricks out in the open so we know what we’re dealing with, because they’re happening whether it gets published or not.
This is probably the most ethical you’ll ever see it. There are definitely organizations committing far worse experiments.
Over the years I’ve noticed replies that are far too on the nose. Probing just the right pressure points as if they dropped exactly the right breadcrumbs for me to respond to. I’ve learned to disengage at that point. It’s either they scrolled through my profile. Or as we now know it’s a literal psy-op bot. Already in the first case it’s not worth engaging with someone more invested than I am myself.
You put it better than I could. I’ve noticed this too.
I used to just disengage. Now when I find myself talking to someone like this I use my own local LLM to generate replies just to waste their time. I’m doing this by prompting the LLM to take a chastising tone, point out their fallacies and to lecture them on good faith participation in online conversations.
It is horrifying to see how many bots you catch like this. It is certainly bots, or else there are suddenly a lot more people that will go 10-20 multi-paragraph replies deep into a conversation despite talking to something that is obviously (to a trained human) just generated comments.
Yeah I was thinking exactly this.
It’s easy to point to reasons why this study was unethical, but the ugly truth is that bad actors all over the world are performing trials exactly like this all the time - do we really want the only people who know how this kind of manipulation works to be state psyop agencies, SEO bros, and astroturfing agencies working for oil/arms/religion lobbyists?
Seems like it’s much better long term to have all these tricks out in the open so we know what we’re dealing with, because they’re happening whether it gets published or not.
I marketing speak this is called A/B testing.
But you aren’t allowed to mention Luigi
You’re banned for inciting violence.
Free Luigi
Eat the rich
The police are a terrorist organization
Trump and Epstein bff