Alex Blania: social platforms face an existential bot threat — World's orb is the only real fix

reream waiting room so let me tell you about gusto the unifi side platform for payroll, benefits, and HR built to evolve with modern small and mediumsiz businesses. And without further ado, we have Alex from Worldcoin.

What's happening?

How you doing? And Merge Labs.

What's happening?

Last time, wait, last time we had you on, it was Worldcoin, right? Or World Labs, sorry. Tools for humanity. [laughter]

Tools for humanity world and Merge Labs. Now,

so yeah, reintroduce yourself for everyone who might not be familiar. give us the lay of the land and then I definitely want to talk about uh just uh everything that's going on in tech and bots. Yeah,

great. Let's do it. Uh I'm Alex Ba. I'm one of the founders of tools for humanity which is the company behind world and and one of the core premises of world uh always has been

that eventually we will need a proof of human at internet scale because

AI will be agentic and we'll pass the touring test. We'll create videos all those things and

that will lead to basically all human interaction on the internet breaking down.

Yeah.

Um so that's that's that's world and then Merch Labs is a BCI lab.

Cool. And then uh what was your initial reaction to Maltbook? Did you you know think it was this it was Skynet? Did it have you for a second? Did it have you in the first half or were you immediately this is all slop or are you still worried about the AI safety and slop concerns? like how how have you processed the last few days?

Yeah, I think I'm somewhere in between. I think like um you know I think it's like an early glimpse of I think what is about to happen and how the internet will start to look more and more like I think.

Yeah.

Um but I think you know Balaji had this tweet about well it's still humans prompting etc. Like that that's that's of course true. I think there's like a big caveat to that comment that there will be humans prompting for a long time but just the execution time frame of these prompts will go much much longer. So

uh you know it's maybe not

I think a lot of people's reaction was like we have we have enough bots we have enough bots here on a already why do we need a new a new platform?

Yeah.

So yeah

well I think the silver lining maybe like just just to say make something useful out of it. I think like the silver lining is a little bit that um you know on the internet we will need to mediate human and AI interaction and I think that's not like that will require a lot of work and I think that's actually the opening for I think a lot of new products even social products so I think there like there's like a platform shift but it's not it was not the LLM interface alone I think it's like agents and and and humans interacting on on things like social networks and that's you know what this kid with mold book. We don't have it yet because who knows how many humans are behind this, how many bots it actually are. Everything is like heavily civil attacked. So there's like probably way less usage than you than you see. But it's a it's a very interesting experiment, I think.

Yeah. Explain civil attacks for those who might not be familiar. Oh, civil text just means that you um you know that behind just just breaking it down for the mold book use case that the number of comments and replies and mods you see is actually way less

uh actual humans behind or agents behind than you would think.

Yeah, I saw one guy was able to register I think 50,000 agents and so

that's right.

Oh, 500,000. One guy. Wow. Okay. I I I'm constantly off by order of magnitude when I talk about mold book because I I checked it when it was at exactly 150,000 users and now it's at 1.5 million and so I'm get everything wrong. But uh one one of the questions I had uh for the founder earlier was was uh really around

every there there I don't think there's a social media like founder CEO in the world that's not thinking about bots how they can uh mitigate some of the potential risks to them but also leverage them for engagement. And I was telling him how I've, you know, we've had this debate on the show like how how much does X uh really care about bots, right? Because uh if they care a lot about preventing them from being on the platform and they've put a lot of effort in so far, we should be kind of worried because it's not working, right? Maybe it's it's it's not non not solvable without uh maybe something like a World Labs. Um, but how have your like how do you think that other social media uh CEOs are kind of like planning around a world where maybe the average user on a social media app is a bot?

Yeah. So, so first of all, it's it's World and Merge Labs. Two two two different companies. World Labs is yet another company that right that's is that

Yeah. Exactly. World is Merge Labs is you. World coin is you, but it's actually from tools for humanity is the name of the company.

Yeah,

that's right. So now, now we've got it sorted. Um, yeah. So, I think look, just breaking down fundamentals. I think social media networks like many other businesses are about human to human interaction. Yeah. Like that's just like if you break it down, like that's literally the then like that's how the that's how the whole thing functions.

Yeah. But they're about that. But if I'm operating a social platform and there's a bot that I may or may not endorse that then engages with a real human and that real human lands back on the platform and I can serve them an ad, what incentive do I have to stop them?

That that I think that is correct until to a certain point, but at some point it's going to that's going to flip because at some point, you know, every user is just going to be so annoyed

and realize that like okay, like this platform is clearly

is all slob. I'm arguing with AIS. So I think there's like only so far how you can take that and I think we're basically about to approach that limit until which it just doesn't become entertaining anymore and it's just like super annoying.

Yeah.

And so I actually do think that these platforms are basically threatened in their core business. Um because you know like human interaction is why users are there. Uh that's how you do advertising. Uh is human attention essentially. So the moment that falls away and is not authentic anymore, I think the whole business is under threat. So that's maybe my my first statement. Second, I think

to how do CEOs think about it? I think they I I know some of them that really think about it. I I don't know all of them, but I definitely think they should really start thinking about it now a lot because it's going to get like exponentially worse really fast.

How do you think about uh all the things that people will do? It's like this cat-and- mouse game. I mean right now you see you see bots but you also see just people who are you clearly just going to an LLM prompting write a LinkedIn post about my business here's some facts I want to have this format and then they just copy and paste it and they're the ones that are sending it and so you know there's there's this there's this gradient between like there's a Python script that's just replying so cool to every post that it sees on the API and then there's the human that's actually just using AI to generate the text and they're typing it themselves or whatever. And then in between you might have someone the fear would be there's someone that is hired basically just to do the world eyeball scan and then post slop and then they scan again then they post slop and they can't scan again like how are you thinking about the actual integration with the platforms to fight like the all the adversarial uh you know activity that you'll see?

Yeah. Um so maybe the first mental model you should have is that one version of uniqueness is the core property that you look for. So meaning

one individual should have one or a limited number of accounts. That's kind of like the core property you're already talking about. Sure.

I actually think bas basically most of what we will do will be AIdriven. So basically everything that I write will be somewhat co-edited by AI. Yeah.

Or I I will create like fun videos or images. that's totally okay as long as I cannot create 500,000 accounts to do that because then the platform breaks down. So uniqueness is like the core property and I think that's

quite hard uh to accomplish. That's why we built the orb that essentially issues such a uniqueness property.

Um so that's that is a really hard piece. Uh I think we have solved that. I think the reauthentication piece that you just mentioned where uh you whatever you can hire a lot of people that already have world IDs and just use that to post on your platform. I think that's correct and I think we will see some of that but that's still you know way way less that one individual can create 500,000 accounts

uh that are just AI agents. So meaning rate limiting like and and the stronger you can get that the better the system becomes basically.

Yeah. Are there any uh sort of like longtail networks or internet properties that you think are are underrated or under discussed as vectors for spam? We saw this sort of funny viral video from a friend who went to his local coffee shop and found that they had like a hundred Android phones running a bot farm just liking whatever they posted on Instagram just to sort of like give a little bit of extra algorithmic juice to their coffee shop promotion. And that was something I hadn't really considered. I think about X, I think about, you know, LinkedIn and Instagram and AI slop there. But maybe I'm not thinking about it in the impact it's having on the average coffee shop on the corner, that type of thing. Well, one one thing that I think is actually happening much more is the kind of these like kind of call attacks to especially elderly people

as like an AI sounding like their whatever grandson and you know tricking them into like that kind of thing I think is going to

I had that I had that happen uh with my grandparents pre-ai.

Oh really? They were just pretend. They were just kind of saying it was me

and they were relying on my grandparents not not like just processing that it wasn't my voice because they were quite a bit older doing an impression.

Um and I can't imagine what you know what what's happening now where it just it would be exactly like my voice. It could have easily with with that they were like

they clocked it pretty quickly. They called my dad. They were like Jord's fine, right? They made up some crazy story about how I was like in a

uh I'd been in I was in jail and this was my one call and I needed them to

wire money or something. That's crazy.

Yeah. So stuff like that is going to wrap up I think. Then we will get to real time deep fakes probably in the next 12 months.

Mhm. meaning you can just like there there was this you know viral tweet I think last week about uh this this guy that was like looking like a super attractive girl on on a basically a live video call.

So that that kind of thing is going to get commoditized and you know he'll just become super super easy. Yeah, we've had this debate with with one guest. We were trying to clock [laughter] we were trying to clock if they were using a like a gigachad filter just like a 10% gigachad filter. And we were we were we were a lot of people were but we were like pausing it and being like okay like he moved his finger in front of his face change. Yeah. [laughter] Like pixel peeping.

But why why did you assume that he he has a gigot filter?

Cuz he looked like a gigot. [laughter]

Yeah. No, if he wasn't if he was calling if he wasn't

and and other people online were levying accusations at him. So, we were like, it's our duty to get to the bottom of this.

Uh like you I mean, you're clearly using a giga chat filter today, right?

Thank you.

That's just what you look like. Wow, dude. It's working. Whatever you're doing is working.

Yeah. It's a sales call. [laughter]

No. Uh what what else is new in your world? Uh I mean, how are you splitting your time? What what what does the rest of the year look like? You're just patiently awaiting the singularity.

I mean, I think, you know, it's the most important time in in a long long while. So, I think just trying to work all day long and making it happen.

Um, spending my time still very focused on world uh you know, Merge Labs is is a research lab.

Sure.

That we that we started with the goal of actually uh bridging artificial and biological intelligence.

Um, but it really is a research lab. So there I think we have a lot of great scientists.

It's not a heavy operation yet. Um so

but on the other side I think it's the it really is the year for world because I think

uh you know aantic

all these kind of like mold book cloud bots that's just the beginning. I think it's going to get

Yeah. Do you think there's something like a lot of these ideas uh molt book and clawbot now openclaw were all ideas that people had had and yet there was some element of risk surrounding them

that maybe pre presented uh prevented some company like bigger labs from even going [snorts] there right like it's not like the idea of like a

agent social network is some it you know this is not the first time that someone has talked about this this is idea that's been floating around some of them have been funded the

prop actually like written about this they've run the experiments too.

Uh so is there some element right now where we're in a point where like you're getting these sort of open source or community grown projects against ideas that labs had or big companies had but just didn't have

uh didn't have like kind of like the freedom to do without a ton of risk.

Sure. I think there's just this general element of you can be much more scrappy as a as a developer you know like entropic or openi doing something like that you know you would actually need to mediate human AI collaboration for example all these kind of things would need to be solved and the bar is much much higher so I think that's true um and then broadly speaking just AI is getting so much better that one guy can create such a such a platform like super fast

you know and can it can iterate on a daily basis like a couple years ago this would would have needed to be like a team of developers probably.

Yeah.

Um and so I think that's the thing to not also underestimate just like how much more effective one person can be now. Um and I think that's going to get like great crazier. So I think we'll see a lot of that happening.

Buckle up pretty soon.

Well, we appreciate you taking the time to come chat with us.

Yeah. Thanks for awesome.

Congrats on all the power.

Thank you guys.

Good to see you.

We'll talk to you soon. Goodbye.

See you. Byebye. Let me tell you about Reream. One live stream, 30 plus destinations. If you want to multiream, go to reream.com. Our next guest is anonymous. It's Nick from X.com. Obviously, all over the internet. He's in our YouTube chat often. We've had back and forth with him. I accused him of being an open AI hater. He said number one. I'm I got a bal I got a balanced set of opinions. So, I'm excited to talk to him today about his takes on AI, his takes on the vibes online, see where we agree, where we disagree. It'll be interesting. Uh, so let's bring him in from the reream waiting room. Nick, what a beautiful profile picture.

You look exactly to have you on the show. Thanks for joining.

Well, thanks for having me.

Thanks so much. uh why don't you kick us