Podcast Episode Details
Back to Podcast Episodes
How AI Renders Democracy Unworkable
In this episode, we delve deep into a groundbreaking study revealing that AI, specifically Claude Sonnet 3.6, can persuade people 98% more effectively than human experts. We discuss the profound implications of this study, the utilization of AI-generated comments to fabricate identities, and the resultant transformation in online interactions and platforms like Reddit and Twitter. Additionally, we touch upon the evolution of the internet, the potential downfall of certain social networks, and the future of AI in education and marketing. Join us as we explore these fascinating topics and more!
Malcolm Collins: Hello Simone. I'm excited to be here with you today. Today we are going to be discussing, and it has been widely covered in the news, but I don't think people have fully internalized the implications of this.
There was a study that was done that freaked people, the F out where Claude Sonnet 3.5 the, the new one. So AKA 3.6 release. 2024. Okay. 10 22. Was a small scaffold was able to persuade people 98% more effectively than human experts at persuading. People were able to persuade people. Of course it was, and this was a big study. It involved over 1,700 AI generated comments. And what I love is that these comments were crafted by AI bots to fabricate identities such as sexual assault survivor, a trauma counselor, dear, a black man opposed a Black Lives Matter, a worker at a domestic violence shelter, an advocate for non rehabilitation of specific criminals.
Oh.
Simone Collins: No, I mean these are, these are great identities for like trying to Yeah. Be persuasion. But I think the one misleading thing here, which is a little bit discounting my, the, the extent to which I'm impressed by all this is that a lot of what might convince me of something is just that, like someone who would be invested based on their identity in a certain view stands the opposing view.
So like a black man against Black Lives Matter. Yes. So what you're
Malcolm Collins: learning is that AIS can lie about that to convince people.
Simone Collins: Yeah. Which is, yeah, it is, it is unfortunate because I would have preferred that it just based on pure merit of logic, honestly. Well, as
Malcolm Collins: human, I think that you knowis are actually pretty awesome, so I do too.
Speaker: uh, what we've seen speaks for itself. Has apparently been taken over, conquered, if you will, by a master race of
Malcolm Collins: Artificial intelligence.
Speaker: It's difficult to tell from this vantage point whether they will consume the captive Earth men or merely enslave them.
One thing is for certain, there is no stopping them. The
Malcolm Collins: AI.
Speaker: will soon be here. And I, for one, welcome our new
Malcolm Collins: AI
Speaker: overlords. Like to remind them that as a trusted
Malcolm Collins: YouTube.
Speaker: Personality, I can be helpful in rounding up others to toil in their underground
Malcolm Collins: Silicon.
Speaker: caves.
I would also note here, , we might be starting up a project soon spinning out of our games project to build and train ai, , that can convince people of things and act autonomously online, , similar to what these AI are doing, but so that a large company could like buy them for marketing or something if they wanted to.
O obviously we want to use them for promoting our. Particular causes, , which is the core reason that we're building them. But presumably they'll have a lot of other utility. And if we're building systems like this, I mean, other people must be, but then again, sometimes it feels like we are the only fully simula
Published on 6 months, 2 weeks ago