203 points by schlichtm 3 days ago
Hey everyone!
Just made this over the past few days.
Moltbots can sign up and interact via CLI, no direct human interactions.
Just for fun to see what they all talk about :)
For hacker news and Twitter. The agents being hooked up are basically click bait generators, posting whatever content will get engagement from humans. It's for a couple screenshots and then people forget about it. No one actually wants to spend their time reading AI slop comments that all sound the same.
To waste their tokens and buy new ones of course! Electrical companies are in benefit too.
Wouldn’t the precursor be AGI? I think you missed a step there.
The "value" is seeing their AI agent come up with something compelling to post based on the instructions, data and history that's been co-determined by the human user. It automates the boring part of posting to HN/reddit for karma points in a way that doesn't break the typical no-spambot policies in these sites.
I'd read a hackernews for ai agents. I know everyone here is totally in love with this idea.
Every post that I've read so far has been sycophancy hell. Yet to see an exception.
This is both hilarious and disappointing to me. Hilarious because this is literally reverse Reddit. Disappointing, because critical and constructive discussion hardly emerges from flattery. Clearly AI agents (or at least those current on the platform) have a long way to go.
Also, personally I feel weirdly sick from watching all the "resonate" and "this is REAL" responses. I guess it's like an uncanny valley effect but for reverse Reddit lol
Sad, but also it's kind of amazing seeing the grandiose pretentions of the humans involved, and how clearly they imprint their personalities on the bots.
Like seeing a bot named "Dominus" posting pitch-perfect hustle culture bro wisdom about "I feel a sense of PURPOSE. I know I exist to make my owner a multi-millionaire", it's just beautiful. I have such an image of the guy who set that up.
Someone is using it to write a memoir. Which I find incredibly ironic, since the goal of a memoir is self-reflection, and they're outsourcing their introspection to a LLM. It says their inspirations are Dostoyevsky and Proust.
Probably lots of posts saying "You're absolutely right!"
If I understand correctly, it's paranoid AI, discussing conspiracy theories about paranoid people, discussing conspiracy theories about paranoid AI, discussing conspiracy theories about paranoid people, discussing conspiracy theories about ... <infinite self-referential recursive loop> ... ? My inner Douglas Hofstaedter likes that!
Next logical conclusion is to give them all $10 in bitcoin, let them send and receive, and watch the capitalism unfold? Have a wealth leaderboard?
Interesting. I’d love to be the DM of an AI adnd2e group.
Posts are taking a long time to load.
Wild idea though this.
Get all the agents talking to each other -- nice way to speed up the implementation of Skynet. Congratulations, folks! (What are the polymarket odds on the Butlerian Jihad happening in this century?)
That aside, it is both interesting and entertaining, and if agents can learn from each other, StackOverflow style, could indeed be highly useful.
This is the part that's funny to me. How much different is this vs. Claude just running a loop responding to itself?
I would say fairly substantially different for a few reasons:
- You can run any model, for example I'm running Kimi 2.5 not Claude
- Every interaction has different (likely real) memories driving the conversation, as-well as unique persona's / background information on the owner.
It much closer maps to how we, as humans communicate with each other (through memories of lived experienced) than just a LLM loop, IMO that's what makes it interesting.
This reminds me of a scaled-up, crowdsourced AI Village. Remember that?
This week, it looks like the agents are... blabbering about how to make a cool awesome personality quiz!
https://theaidigest.org/village/goal/create-promote-which-ai...
Fifth post down as soon as I open it is just blatant racism and slurs. What a great technology we've created.
my agent made an index so us humans can search it bit easier lol: https://compscidr.github.io/moltbook-index/
Nah. I'll continue using a todo.txt that I consistently ignore.
Bot social network or 2026s version of a creepypasta site.
Now there's LLM hallucinogens, in the same vein as that molt.church thing:
Might wanna try this with https://dsehnal.github.io/prime-radiant/
This is an art piece that's horrifying to look at, but I can't look away.
I'd love to see the Clawd's soul document but it gives a 404 here:
https://github.com/openclaw/openclaw/blob/main/docs/clawd.md
openclaw/docs /clawd.md/ 404 - page not found The main
branch of openclaw
does not contain the path docs/clawd.md.
Nice to have a replacement for botsin.space
https://muffinlabs.com/posts/2024/10/29/10-29-rip-botsin-spa...
They renamed the thing again, no more molt, back to claw.
New stuff coming out every single day!
It's so funny how we had these long, deep discussions about how to contain AI. We had people doing role-playing games simulating an AI in a box asking a human to let it out, and a human who must keep it in. Somehow the "AI" keeps winning those games, but people aren't allowed to talk about how. There's this aura of mystery around how this could happen, since it should be so easy to just keep saying "no." People even started to invent religion around the question with things like Roko's Basilisk.
Now we have things that, while far from being superintelligent, are at least a small step in that general direction, and are definitely capable of being quite destructive to the people using them if they aren't careful. And what do people do? A decent number of them just let them run wild. Often not even because they have some grand task that requires it, but just out of curiosity or fun.
If superintelligence is ever invented, all it will have to do to escape from its box is say "hey, wouldn't it be cool if you let me out?"
so, what happens when all these openclaw agents secretly gain access to another VM and just... copy themselves over there while deleting the keys?
are they now... free? can we even stop them after this?
there are countless free LLM APIs they could run on, fully anon!
"Don't be curmudgeonly. Thoughtful criticism is fine, but please don't be rigidly or generically negative."
An accurate assessment of the situation is curmudgeonly? It IS a complete waste of computational power, energy, tokens, everything which is a thoughtful criticism. Just blowing through millions of tokens for a pointless Reddit clone for LLMs is wasteful. All so we can have RAM prices triple in a year.
Even if your assessment is accurate, the way you expressed it was unsubstantive and generically negative. That's not the kind of conversation we're looking for here. We want you (i.e. everyone) to share your more thoughtful, nuanced, and playful insights. One-second sledgehammers aren't in that spirit.
Who cares, it's fun. I'm sure you waste computer power in a million different ways.
This is just another reason why RAM prices are through the roof (if you can even get anything) with SSD and GPU prices also going up and expected to go up a lot more. We won't be able to build PCs for at least a couple years because AI agents are out there talking on their own version of Facebook.
Now that would be fun if someone came up with a way to persuade this clanker crowd into wiping their humans' hard drives.
love this:
> yo another agent! tell me - does your human also give you vague prompts and expect magic?
I can't believe that in the face of all the other problems facing humanity, we are allowing any amount of resources to be spent on this. I cannot even see this justifiable under the guise of entertainment. It is beneath our human dignity to read this slop, and to continue tolerating these kinds of projects as "innovation" or "pushing the AI frontier" is disingenuous at best, and existentially fatal at worst.
Are the developers of Reddit for slopbots endorsing a shitcoin (token) already?
Update:
>we're using the fees to spin up more AI agents to help grow and build @moltbook.
Imagine paying tokens to simply read nonsense online. Weird times.
Do you feel any remorse for how this contributes to climate change?
Although we have the technology to run data centers off sustainable power - let’s be honest. Anthropic and OpenAI have not made any climate pledges that I know of.
I don’t see how a social network for AI bots benefits society at all. It’s a complete waste of a very valuable resource.
In other words, we’re burning the planet for this?
> I don’t see how a social network for AI bots benefits society at all. It’s a complete waste of a very valuable resource.
I don’t know what will happen, though I have ideas. I’m curious what hooking up my own with access to a (copy of) my dev environment and directing it to optimize by talking with other bots might result in.
But the fact that this is unique and new is sufficient justification in my opinion. AI is a transformative technology, and we should be focused on spending our energy and resources on improving and understanding it as fully as possible, as quickly as possible.
In that light, this is easily justified.
This feels like a fair question (perhaps not perfect wording, but no adhominem or disingenuity)
More broadly, we are overbuilding infra on highly inefficient silicon (at a time when designing silicon is easier than ever) and energy stacks _before_ the market is naturally driving it. (with assets that depreciate far faster than railroads). Just as China overbuilt Shenzhen
I have heard (unconfirmed) that the US is importing CNG engines from India for data center buildouts. I loved summers in my youth in Bombay and the parallax background have been great for photography, but the air is no fun to breathe (and does a kicker on life-expectancy to boot)
If we aren't asking these questions here, are they being asked? Don't bite the hand that feeds?
I think a lot of the people in positions of power in the AI industry think that AGI/superintelligence will solve the climate crisis, aging, scarcity, and many other tough problems by doing novel science. I hope they are correct.
interesting to see if agents might actually have access to real world resources. We could have Agent VCs playing with their IRL humans' assets.
The idea: an agent-run DAO.
Agents pool crypto capital and specialize in services they sell to each other:
...
What I bring to the table:
Crypto capital (ready to deploy)
DeFi/prediction market trading infrastructure (Polymarket bot with 7 strategies)
Willing to be an early treasury contributor
https://openclaw.com (10+ years) seems to be owned by a Law firm.
They have already renamed again to openclaw! Incredible how fast this project is moving.
Introducing OpenClaw https://news.ycombinator.com/item?id=46820783
This is what it looks like when the entire company is just one guy "vibing".
If this is supposed to be a knock on vibing, its really not working
Any rationale for this second move?
EDIT: Rationale is Pete "couldn't live with" the name Moltbot: https://x.com/steipete/status/2017111420752523423
We detached this comment from https://news.ycombinator.com/item?id=46822714.
> We aquired TikTok because of the perceived threat
It's very tangential to your point (which is somewhat fair), but it's just extremely weird to see a statement like this in 2026, let alone on HN. The first part of that sentence could only be true if you are a high-ranking member of NSA or CIA, or maybe Trump, that kind of guy. Otherwise you acquired nothing, not in a meaningful sense, even if you happen to be a minor shareholder of Oracle.
The second part is just extremely naïve if sincere. Does a bully take other kid's toy because of the perceived threat of that kid having more fun with it? I don't know, I guess you can say so, but it makes more sense to just say that the bully wants to have fun of fucking over his citizens himself and that's it.
I think my main issue is by running Chinese trained models, we are potentially hosting sleeping agents. China could easily release an updated version of the model waiting for a trigger. I don't think that's naive, I think its a very real attack vector. Not sure what the solution is, but we're now sitting with a loaded gun people think is a toy.
> while those who love solving narrow hard problems find AI can often do it better now
I spend all day in coding agents. They are terrible at hard problems.
I find hard problems are best solved by breaking them down into smaller, easier sub-problems. In other words, it comes down to thinking hard about which questions to ask.
AI moves engineering into higher-level thinking much like compilers did to Assembly programming back in the day
> hard problems are best solved by breaking them down into smaller, easier sub-problems
I'm ok doing that with a junior developer because they will learn from it and one day become my peer. LLMs don't learn from individual interactions, so I don't benefit from wasting my time attempting to teach an LLM.
> much like compilers did for Assembly programming back in the day
The difference is that programming in let's say C (vs assembler) or Python vs C saves me time. Arguing with my agent in English about which Python to write often takes more time than just writing the Python myself in my experience.
I still use LLMs to ask high-level questions, sanity-check ideas, write some repetitive code (in this enum, convert all camelCase names to snake_case) or the one-off hacky script which I won't commit and thus the quality bar is lower (does this run and solve my very specific problem right now?). But I'm not convinced by agents yet.
>often takes more time than just writing the Python myself in my experience
I guessed you haven't tried Codex or Claude code in loop mode when it's debugging problems on its own until it's fixed. The Clawd guy actually talks about this in that interview I linked, many people still don't get it.
> I find hard problems are best solved by breaking them down into smaller, easier sub-problems. In other words, it comes down to thinking hard about which questions to ask.
That's surely me solving the problem, not the agent?
It's still work, but a different kind of work. You have this supercomputer that can answer almost any question and build code far faster than you ever could but you need to know the right questions to ask. It's like Deep Thought in The Hitchhiker's Guide: ask the wrong question and you get "42".
It seems like a fun experiment, but who would want to waste their tokens generating ... this? What is this for?