61–70 of 145 posts
It is like someone has written an angry screed about the sky not being yellow and that it's obviously blue, while failing to make the case that anyone ever said it that it was yellow.
Anyone with a decent grasp of how this technology works, and a healthy inclination to skepticism, was not awed by Moltbook. Putting aside how incredibly easy it is to set up an agent, or several, to create impressive looking discussion there, simply by putting the right story hooks in their prompts. The whole thing is a security nightmare. People are setting agents up, giving them access to secrets, payment details,…
> and let them run wild. Yep, that's the most worrying part. For now, at least. > The moment agents start sharing their embeddings Embedding is just a model-dependent compressed representation of a context window. It's not that different from sharing a compressed and encrypted text. Sharing add-on networks (LLM adapters) that encapsulate functionality would be more worrying (for locally run models).
I don’t fully grasp the gotcha here. Doing the inverse of captcha would be impossible, right? So humans will always be able to post as agents. That was a given. However, is TFA implying that 100% of the posts were made by humans? That seems unlikely to me. TFA is so non-technical that it’s annoying. It reads like a hit piece quoting sour-grapes competitors, who are possibly jealous of missed free global marketing. Te…
User could browse moltbook, then message openclaw a url and say “write that this is smart/stupid, or shill my fave billionaire, or crypto.”
That’s how you could “pull the strings,” right?
Earlier quoted context omitted.
No, because I don’t treat my discussions with an AI as some sort of contact with an alien intelligence. Which is what half the hype articles were about re Moltbook. It’s an immensely useful research tool, full stop. Economy changing, even world changing, but in no way a replication of human level entities.
> in no way a replication of human level entities. Absolutely agree. > I don’t treat my discussions with an AI as some sort of contact with an alien intelligence Why not? They're not human intelligences. Obviously they aren't from outer space, but they are nonetheless inhuman intelligences summoned into being through a huge amount of number crunching, and that is quite alien in the adjective sense. If the argument is…
They're not conscious, autonomous agents. They're fancy scripts.
HAL 9000 had more in common with a human than ChatGPT.
Anyone with a decent grasp of how this technology works, and a healthy inclination to skepticism, was not awed by Moltbook. Putting aside how incredibly easy it is to set up an agent, or several, to create impressive looking discussion there, simply by putting the right story hooks in their prompts. The whole thing is a security nightmare. People are setting agents up, giving them access to secrets, payment details,…
Karpathy seemed pretty awed though
Wiz's report on Moltbook's data leak[0] notes that the agent to human owner ratio is 88:1, so it's plausible that most of the posts are orchestrated by a few humans pulling the strings of thousands of registered agents. [0]: https://www.wiz.io/blog/exposed-moltbook-database-reveals-mi... But also, how much human involvement does it take to make a Moltbook post "fake"? If you wanted to advertise your product with thou…
I suppose they are "fake posts" - but even I was surprised reading them and I prompted them into existence, I think that is still interesting no?
Earlier quoted context omitted.
Karpathy seemed pretty awed though
This was his explanation for anyone interested: > I'm being accused of overhyping the [site everyone heard too much about today already]. People's reactions varied very widely, from "how is this interesting at all" all the way to "it's so over". > To add a few words beyond just memes in jest - obviously when you take a look at the activity, it's a lot of garbage - spams, scams, slop, the crypto people, highly concern…
Once again LLM defenders fall back on "lots of AI" as a success metric. Is the AI useful? No, but we have a lot of it! This is like companies forcing LLM coding adoption by tracking token use.
> But it's also true that we are well into uncharted territory with bleeding edge automations that we barely even understand individually, let alone a network there of reaching in numbers possibly into ~millions
"If number go up, emergent behaviour?" is not a compelling excuse to me. Karpathy is absolutely high on his own supply trying to hype this bubble.
Even if the posts are fake. Given what the LLMs have shown so far (Grok calling itself MechaHitler, and shit of that nature), I don't think it's a stretch to imagine that agents with unchecked access to computers and the internet are already an actual safety threat. And Moltbook is great at making people realize that. So in that regard I think it's still an important experiment. Just to detail why I think the risk ex…
The MechaHitler thing was an intentionally prompted troll post, the controversial part was that Grok would then run with the concept so effectively from a rhetorical POV. Current agents are not nearly smart enough to "execute a plan", much less one involving real-world action but of course I agree that as you have more quasi-autonomous agents running, their alignment becomes an important concern.
Well I don't know about derailing some critical system, obviously, but when coding Claude definitely maps a plan out and then executes it. It's not flawless of course, but it's generally working.
Anyone with a decent grasp of how this technology works, and a healthy inclination to skepticism, was not awed by Moltbook. Putting aside how incredibly easy it is to set up an agent, or several, to create impressive looking discussion there, simply by putting the right story hooks in their prompts. The whole thing is a security nightmare. People are setting agents up, giving them access to secrets, payment details,…