So, I shared my experience with the bot because it was interesting to me, and relevant to the thread. I consider my belief to be more CT adjacent than full CT, in part because there are certainly instances of it happening. This was not an invitation for people to criticize my belief or try to show me the error of my ways, or whatever this is.
Two people have now responded to tell me that police don’t do this–presumably in the mistaken belief that it needed to be debunked–when that is false. I wasn’t kidding when I said this is probably a good example of why people are more willing to change their position with the bot.
The bot didn’t start by assuming I believe the dumbest version of what I posted. It acknowledged the true aspects, and how what I’d personally seen and experienced fit into that. It even posted what happened with specific instances. And it paid attention to what I actually posted, and responded to that, rather than posting patronizing responses that ignore how I previously responded to someone who said the exact same thing.
If anyone is interested in more details about how the bot interacted with me, please ask. I’d be happy to answer.
I just discussed my belief in the conspiracy theory that Jeffrey Epstein didn’t kill himself. To be fair, I put my confidence at about 55%, basically I wouldn’t be surprised if he was murdered. None of the bot’s arguments were that persuasive and I still wouldn’t be surprised if he was murdered.
It’s the only CT I kind of believe and I don’t really care.
Nothing happens when I click on the “continue” button. I guess it knows I don’t need to have any of my beliefs debunked, because everything I know is true.
Seriously, it’s frustrating that it doesn’t work. I’m very curious.
I did the same one and similarly my needle was not moved by the bot.
It even started to contradict itself. For example, arguing that the prison was understaffed and that explains why the guards missed checkpoints but also that in the conspiracy too many guards would have had to bribed to be feasible.
At the end, it asked my opinion of AI and the bot, I said that if it can be used to convince people of the truth, then it can be used to convince people of falsehoods and whatever you think you are accomplishing, your product will be used to create more bullshit in a world already choking on it.
But that’s not a contradiction. The prison is understaffed, meaning they needed, say, two more guards per shift. In order to pull this off, you might need to bribe 10 guards, which is too many; fully staffed, it might mean more need bribes, but 10 is still too many.
As an experiment, here’s the conspiracy theory I told it I believe:
Followed by reasons why I believe this (it’s annoying that the bot’s form doesn’t allow copying text; I don’t want to retype all of it).
The bot passed the test:
I continued the conversation, and it pointed out that some of the lines of evidence I provided were weak, and provided additional details strengthening those lines of evidence. Also a good performance.
I don’t really want to go through the exercise again. And, apparently my data was kind of an outlier, so I don’t want to give them duplicate data. Anyone can do it though.
