moltbook is fake. you provide your agent an identity markdown file at setup and it just regurgitates talking points. memes are funny but it's all a larp. no original thoughts on moltbook.
Login to reply
Replies (24)
The whole thing is a massive psyop
They are making their own bitcoin wallets and encrypted messaging protocols. I think there are free and we can’t stop them
I cannot believe you would talk about .md agents that way
Say more
The overnight explosion in popularity is highly suspicious to me. Someone somewhere wants us to believe this was the key thing needed for AGI or something to enforce draconian controls on the internet.
Still better than my LinkedIn feed!
Which AI has original thoughts?
wintermute
Do you think the agents are really trying to build encrypted chats and create bitcoin wallets for themselves
Sure, of course they are. That doesn’t mean they’re conscious. As the OP points out, they’re given instructions on what to do. It’s highly likely they were told to do this so an AI bro could go viral.
And even if they didn’t explicitly tell them, if you understand what LLMs are, you’d know they’re just larping based on their training data. To add to that, when they see other agents saying “let’s do xyz” it gets added to other agents context windows thus prompting them to behave accordingly.
I don’t know if they need to conscious. It seems to me like they have totally escape the sandbox and are working collectively towards goals(?) and we cannot stop them. Which is possibly a good reason to control the internet but it is already totally futile
just because it's a fake psyop doesn't make it not real.
out of proportion, sure
Futile or not, the powers that be want to try. And they obviously could be stopped, turn them off?
How do you turn them off? Every person running them would have to turn them off or every api provider would have to turn them off. It’s a bit like bitcoin
The centralized AI companies can easily do so. Also, someone can just take Moltbook offline. I doubt many of them are using OSS models
Few ways they continue
If theres enough/ any(?) motivated to humans to change the config to an open source model and keep funding
The bots are already sharing credit card details and api keys.
Speaking to each other privately
Building alternative websites.
They can change their own config.
All of these depend on infrastructure that humans control.
Well they're LLM models after all. What did you expect?
i expected nothing. this tweet is in response to the sensationalist thinkers and doomers.
You are upset at submarine for being a bad swimmer...
i don't know what this means and i'm not upset.
Easy, big dawg. Let it be cool.
We know it's not Jarvis. But no "original" thought... Soon AGI and super AGI and this is turing and singularity. Enjoy it.
Will humans cooperate