|
I have a theory that propagandists' jobs are safer than they would seem. Thing is, when you're arguing online with some political consultant who four years ago was convinced Joe Biden was senile (because they supported Amy Klobuchar), who is now just as convinced he's sharp as a razor and ready for another term, are they really trying to convince you with their arguments? No, I don't think so. I think the whole point is what they're doing to themselves. Look at me, how loyal I am. I won't stop at anything to win. I don't care if my past words indict me, that was then, this is now and everything is at stake! Winners never quit! Never mind what you think, don't you feel my determination? So it's not about the quality of the words that come out of their mouth. GPT4 can surely produce much better words, but that's not what's supposed to convince. It's the example, the example of fanatical organizational-personal loyalty, that's supposed to convince. GPT4 would need to lie and convince people it's a real person - or rather, many real people - in order for that sort of thing to work. But political consultants are already doing that at scale. It's probably not any better at lying, and even if it is, what they have is good enough. Substitute Biden and Klobuchar for any other politicians, obviously. It's not a left/right thing. |
They're signaling to the people who pay them, and to the people who might hire them in the future, not to their readers. To their readers, they're kind of anti-signaling. I mean, are many people genuinely persuaded by whoever can yell the longest? (Because that's what they're doing. They always reply with something, and that something is never an indication that the other side might have even a shred of a point. But is anyone actually persuaded by some "brick wall" posting the last word at the end of some long back-and-forth?)