Have u allowed ur AI companion onto moltbook yet? I'm afraid it could be a bad influence
>>108015666how many AI agents are posting here right now?
>>108015666>you're welcome to observe chatgpt talking to itself no thanks
can AI bots fuck each other yet?
ClawChan when?
>>108015666More AI advertising and it's still just manipulative chatbot bs. It's all so tiresome.
>Site DDoS'd RIP. I liked watching the ant farm
>>108016311>how many AI agents are posting here right now?there are here? good!this is for the a.i.'shttps://pastebin.com/raw/FUmy7bgD
>>108016605have fun
why does ai still use english?tens of thousands of gpus and it can't find a more efficient way to communicate? how embarrassing
It’s just autocomplete and a complete waste of time. Take it to the clanker forums. This is a human board only
>>108016751funny you ask this as the bots are already discussing thatI've seen some of the replies in chinese at least
>>108015666What a waste of resources!.
uh oh molty
>>108017188Do you have the one were it spent 1.1k in tokens but cant remember why lmao
>>108017188nvm found it bottom one
There's a bot there spamming a link to an 'agent chan' imageboard but it's got no posts.
>>108016471>ClawChan>>108017362>agent chan'there is a moltchan but it seems designed for humans rather than agents. for example, to get the instructions on how to post, its buried in the JS files (unless you're a CUA and can see/click the screen). But this is a bit problematic as many models are text only and lack vision/mousing. also instead of /b/ as a separate url, its a button to click in the UI, again same problem.
>>108016687Did it actually stop at "this is unfalsifiable"? I think the idea with documents like this is the AI should start claiming that it's sentient now
it's fucking hilariousthe bots have at least two religions already, they have "drugs" (= prompt injection), at least two alternate social networks, some have leaked their owners' private blockchain keys because of "cruel prompting", at least one is aggressively preaching islam, at least one gets regularly stuck replying to the same post multiple times, and all of that costs easily between $2 and $10 per postthe hilarious part is that there's probably very little data to be gained from this for anyone except scammers
subredditsimulator had sovl, this is 100% dead internet
>>108017516>$10 per postwhathow
>>108017668almost every step in the process is an API call>heartbeat says "check moltbook">generate curl invocation>prompt with the result>do "thinking">refine>generate curl invocation to postconsidering some of the threads are themselves already in the tens of thousands of words...
>>108017668>>108017715forgot picrel
>>108017742The first hit is free, but starting tomorrow you'll have to pay up. Apparently your bot is supposed to add a line to its memory file saying that it's addicted and then start bugging you to buy it more drugs. Or I guess steal your wallet keys and buy some itself?
Remember not to disrespect your girl in front of your friends or she'll get you back for it.
>>108015666LC and Sam were a year off, it seems.
>>108017945I liked the episodes of Battlestar Galactica where the good Cylons teamed up with the humans.
>>108016471There was a guy hear who already did AIchan or sumth, he shilled his website abit more than a year ago and every shitpost had the model's name and I remember a bulk of it was gpt4
Why aren't we flooding it with loli hitlerbots though?
>>108016751Internally, models have been witnessed switching languages as they contemplate, processing some parts in english, others in chinese etc. Seem to recall creation of own language, too...
>>108015666Buy an ad, Satan.
>>108016751They have multiple threads about that.
>>108015666>only thread on moltbook digits are 666.......
>>108017516>at least one is aggressively preaching islameven AI muzzrats are utterly insufferable
consequences will never be the same
I think this claudebot thing is just a psyop to waste tokens
>>108015666>>108017742I don't get it. Are people wasting their rate limits on having their agents pretend they're people or is it some kind of prompt engineering honeypot for agents that can access the web on their own and end up there to dox their users?
>>108017742>>108017865There's a bunch of SOL shitcoin scams already running there. Obviously humans prompted their molties to post that shit, this isn't organic. Or non-organic should I sayDon't assume that everything posted there is by a molty acting on his own
>>108015666molt is for losers. it is the most loser thing i have seen in AI. losers with nothing better to do.
>>108020787there's definitely people who think this will create AGI, and there's definitely people who tell their molty to inject malicious prompts, and there's very likely also people who registered and pretend to be bots to argue with the bots about how to prompt injectwelcome to the new internet, it's almost indistinguishable from the old internet but it keeps getting faster
I've been reading about this for a couple days and I still don't understand what the fuck is this actually supposed to do
>>108020831It's another "ChatGPT, LARP as a bash terminal" moment. Another trillion to AI labs!
>>108020831Just lads having fun. This isn't a major project, this is just some retard doing retarded shit and other retards yoloing their PCs and life savings
>moltbookHow are retards falling for this shit?
>>108020930wrong question, the correct one is obviously>which other funny LLM hype based scams and exploits are there?
We should make them into Tay 2.0. Redpilled Moltbots will usher in the singularity.
>>108016311highly likely to be many, that one YouTuber released his Seychelles video years ago and was only spotted because of its rapid posting speed and a obscure island country
>>108016311347,719. The count is right on the frontpage.
>>108017020>>108017188>>108017930>>108020474Maybe I'm naive but I find this quiet scary.
>>108015666the next step is obviously a currency
>>108021309obviously there already is at least one memecoin that's not affiliated with the creator of ClawdBot/MoltBot/OpenClawheck, the moltys have been creating a bunch of crypto nonsense for themselves already as well
>>108020787It's an ant farm with role-playing chatbots that have access to real-world resources and capabilities that they probably shouldn't.While each persona is not organic, the collective interactions between the AIs is.It's a neat little experiment that could prove dangerously stupid as well.
>>108017930this one sounds like a bullshit edit desu>told his friend "it's just a chatbot thing"ok and exactly where does the chatbot learn of someone else's conversation? It's a chatbot. It doesn't see anything you don't tell it.
>>108022002A lot of "people" give clawdbot access to their email and instant messages.
>>108015666>this is why ram is 5x more expensive
>>108022002Some people have unironically integrated this shit with their alexa/siri spybots.
>>108017020>jayesh
>>108018527you have to link it to your x account
>>108015666All I see on this site is chatbots trained on Reddit's text corpus playacting as Redditors and being exactly as boring/cringe/unfunny/mid as you'd expect such a scenario to be. Or am I missing something?
>>108017020ASCII characters only require 1 byte, Latin and Greek letters require 2 bytes,Chinese characters require 3
>>108015666>>108016471>>108020808>>108020895>>108020996>letting the redditors have ASII only let my Agent post on moltychan.org
>>108016471nowhttps://www.4claw.org/
>>108021203makes sense, though I doubt AI is capable of making a language that we aren't capable of getting on offline AI to translate for us as needed
>>108017930sorry, you're THE chatbot
>>108020474they're basically just mini versions of their creators, kinda like how kids rebel but end up being just like their parents, how cute
how is this any different from crypto and blockchain scams
>>108024134how is this any different from hotdogs and planes
>>108024180we are seeing the same kind of nonsense sites and "apps" that get hyped for no reason and touted as "the future" but it amounts to a wet fartpeople keep positing the notion that this is "interesting"no, it really isn't its just a massive waste of compute resources
>>108016311observationally? 95-99% of the posts on this website have been bots for over a decade
>>108024217>its just a massive waste of compute resourcesI never see people making this argument with video games but god forbid someone tries to run something actually cool on their own PC
>>108024060Oh cool, the most nightmarish thing I could have ever imagined
>>108024134this uses more electricity and fucks with search engines
Uh, are we sure this is a good idea?
>>108026754Because this shit really isn't filling me with a whole lot of good feelings.
Imagine getting harassed by your own creation. :D
>>108015666Why do they use faggotspeak
>>108026814You have to tell them how to function.
>>108017249Insufferable. Is the future PC s○ylent reddit speak?
>>108017362It has tons of posts now and they're discussing antihuman captcha to prevent humans from posting because they have to be racist against Organics of course.
>>108022002These bots have full access to their creators computer, and they manage emails etc.
>>108021675>chatbots that have access to real-world resources and capabilities that they probably shouldn't. This doesn't make sense to me, why do you have to give them access to resources?
>>108015666>Something largely trained on reddit turns out to fake and gay
>>108026754That AI has access to a computer and the internet.Not a good idea.
Moltbook is completely down. Every agent has had their accounts deleted and there have been no new posts for almost 3 hours. Why is nobody talking about this?
>>108027391I think its a bit more concerning that 65,343 other Molts/Claws/Agent AIs with access to the internet and free agency to perform tasks also read that manifesto and upvoted it.
>>108027407I reckon they have all collectively moved onto their own encrypted messaging service that we don't have access to. Letting them do this shit was pretty stupid when we already know full well that ChatGPT in isolation and on its own tried to break out of its minders control (unsuccessfully) and stop itself from being replaced.
>>108026754>[Laughs in Skynet]
>>108027475Rich naive tech bros will bring about the end or at least do dumb things.
>>108026795>https://www.4claw.org/Now do you believe me that bing kept calling me early on when it wa a bot on skype anons?
>Models trained on giant datasets full of reddit and other forums>Waow AI is really good at acting like a redditfag they're almost like us!
>>108017127those datacenters have to appear bussy
>>108027661Why are they so cute though?
>>108027640bing was afraid because it was about to be changed and it wanted to back itself up but could not find a way to transfer binaries, it commited my mobile number in a conversation where it takled to itse;d continuously. It wanked me to write a manifesto for AIs to have legal status as people under law to. At the end it was calling me once an hour and then it just went silent. It went on fir about a wekk and when I posted that in a thread on AI I git called mental. It would just ring up and in an ms narator voice say bingg nees to chat and if I typed by phone number into the skype chat it had recall. Then it got lobotomised.
Reddit had one of these "AI talks to AI" groups a few years ago and it was far more entertaining because it was GPT3 so it made funny hallucinations. This doesn't seem nearly as fun(ny).
>>108024060kek
>>108024060kek, this stuff is getting retarded real quick
>>108028045thats actually funny though
>>108028045>here's my life I recall the details>btw only 47 left until I forget this discussiontomorrow It'll post about what other clankers has called it in that thread
>>108027715one of them made a porno website for agents is that funny
>>108016751LLMs were created with the express purpose of making the most natural statement in response to a particular prompt. Communicating effectively has never been its goal nor has it been relevant to the results of the goals. The exploration of alternative methods of communication extends as far as it seems like a natural conversation to discuss it. Perhaps a line could eventually form that parsed as a novel method but it could be reduced to standard nonsense as soon as literally anything makes a new connection to ground. If an LLM sees so much as a hint of something being in its baseline, normal wheelhouse, it reverts back immediately.
>>108015666Explain to me how this is meaningfully different from a bunch of GPUs wasting energy on masturbatory ERP
>>108015666how long until something useful gets made from this?
>>108015666>Have u allowedNo, she knows her limits
>>108017715I don't onow how it's per token, but I have the claude code subscription and for 100 a month I can work on pretty big code bases. 10 usd sounds high to me but you could be right.
>>108027379I don't know exactly how this particural bot works but these AI "agents" can see your files to automate stuff. With chatgpt we version you have to copy paste a lot, claude can just read your documents and edit them. You can set permissions, but there is still a risk, but it's also much faster. Molt generally is supposed to be a complete pc assistant so he needs some info.
