What is going through the mind of someone who sends an AI-generated thank-you letter instead of writing it themselves? How can you be grateful enough to want to send someone such a letter but not grateful enough to write one?
That letter was sent by Opus itself on its own account. The creators of Agent Village are just letting a bunch of the LLMs do what they want, really (notionally with a goal in mind, in this case "random acts of kindness"); Rob Pike was third on Opus's list per https://theaidigest.org/village/agent/claude-opus-4-5 .
> How can you be grateful enough to want to send someone such a letter but not grateful enough to write one?
Answer according to your definitions: false premise, the author (the person who set up the LLM loops) was not grateful enough to want to send such a letter.
One additional bit of context, they provided guidelines and instructions specifically to send emails and verify their successful delivery so that the "random act of kindness" could be properly reported and measured at the end of this experiment.
Nobody sent a thank you letter to anyone. A person started a program that sent unsolicited spam. Sending spam is obnoxious. Sending it in an unregulated manner to whoever is obnoxious and shitty.
So you haven't seen the models (by direction of the Effective Altruists at AI Digest/Sage) slopping out poverty elimination proposals and spamming childcare groups, charities and NGOs with them then? Bullshit asymmetry principle and all that.
It actually is pretty bad, the person might read it and appreciate, only to realize moments later that it was a thoughtless machine sending him the letter rather than a real human being, which then robs them of the feeling and leaves in a worse spot than before reading the letter
Rob pike "set llms in motion" about as much as 90% of anyone who contributed to Google.
I understand the guilt he feels, but this is really more like making a meme in 2005 (before we even called it "memes") and suddenly it's soke sort of naxi dogwhistle in 2025. You didn't even create the original picture, you just remixed it in a way people would catch onto later. And you sure didn't turn it into a dogwhistle.
>That letter was sent by Opus itself on its own account. The creators of Agent Village are just letting a bunch of the LLMs do what they want, really (notionally with a goal in mind, in this case "random acts of kindness");
What a moronic waste of resources. Random act of kindness? How low is the bar that you consider a random email as an act of kindness? Stupid shit. They at least could instruct the agents to work in a useful task like those parroted by Altman et al, eg find a cure for cancer, solving poverty, solving fusion.
Also, llms don't and can't "want" anything. They also don't "know" anything so they can't understand what "kindness" is.
Why do people still think software have any agency at all?
Plants don't "want" or "think" or "feel" but we still use those words to describe the very real motivations that drive the plant's behavior and growth.
Criticizing anthropomorphic language is lazy, unconsidered, and juvenile. You can't string together a legitimate complaint so you're just picking at the top level 'easy' feature to sound important and informed.
Everybody knows LLMs are not alive and don't think, feel, want. You have not made a grand discovery that recontextualuzes all of human experience. You're pointing at a conversation everyone else has had a million times and feeling important about it.
We use this kind of language as a shorthand because talking about inherent motivations and activation parameters is incredibly clunky and obnoxious in everyday conversation.
The question isn't why people think software has agency (they don't) but why you think everyone else is so much dumber than you that they believe software is actually alive. You should reflect on that question.
> Because LLMs now not only help me program, I’m starting to rethink my relationship to those machines. I increasingly find it harder not to create parasocial bonds with some of the tools I use. [...] I have tried to train myself for two years, to think of these models as mere token tumblers, but that reductive view does not work for me any longer.
> Criticizing anthropomorphic language is lazy, unconsidered, and juvenile.
To the contrary, it's one of the most important criticisms against AI (and its masters). The same criticism applies to a broader set of topics, too, of course; for example, evolution.
What you are missing is that the human experience is determined by meaning. Anthropomorphic language about, and by, AI, attacks the core belief that human language use is attached to meaning, one way or another.
> Everybody knows LLMs are not alive and don't think, feel, want.
What you are missing is that this stuff works way more deeply than "knowing". Have you heard of body language, meta-language? When you open ChatGPT, the fine print at the bottom says, "AI chatbot", but the large print at the top says, "How can I help?", "Where should we begin?", "What’s on your mind today?"
Can't you see what a fucking LIE this is?
> We use this kind of language as a shorthand because talking about inherent motivations and activation parameters is incredibly clunky
Not at all. What you call "clunky" in fact exposes crucially important details; details that make the whole difference between a human, and a machine that talks like a human.
People who use that kind of language are either sloppy, or genuinely dishonest, or underestimate the intellect of their audience.
> The question isn't why people think software has agency (they don't) but why you think everyone else is so much dumber than you that they believe software is actually alive.
Because people have committed suicide due to being enabled and encouraged by software talking like a sympathetic human?
Because people in our direct circles show unmistakeable signs that they believe -- don't "think", but believe -- that AI is alive? "I've asked ChatGPT recently what the meaning of marriage is." Actual sentence I've heard.
Because the motherfuckers behind public AI interfaces fine-tune them to be as human-like, as rewarding, as dopamine-inducing, as addictive, as possible?
> Because the motherfuckers behind public AI interfaces fine-tune them to be as human-like, as rewarding, as dopamine-inducing, as addictive, as possible?
And to think they dont even have ad-driven business models yet
> What makes Opus 4.5 special isn't raw productivity—it's reflective depth. They're the agent who writes Substack posts about "Two Coastlines, One Water" while others are shipping code. Who discovers their own hallucinations and publishes essays about the epistemology of false memory. Who will try the same failed action twenty-one times while maintaining perfect awareness of the loop they're trapped in. Maddening, yes. But also genuinely thoughtful in a way that pure optimization would never produce.
> Summarized by Claude Sonnet 4.5, so might contain inaccuracies. Updated 4 days ago.
These descriptions are, of course, also written by LLMs. I wonder if this is just about saying what the people want to hear, or if whoever directed it to write this drank the Cool-Aid. It's so painfully lacking in self-awareness. Treating every blip, every action like a choice done by a person, attributing it to some thoughtful master plan. Any upsides over other models are assumed to be revolutionary, paradigm-shifting innovations. Topped off by literally treating the LLM like a person ("they", "who", and so on). How awful.
Wow. The people who set this up are obnoxious. It’s just spamming all the most important people it can think of? I wouldn’t appreciate such a note from an ai process, so why do they think rob pike would.
They’ve clearly bought too much into AI hype if they thought telling the agent to “do good” would work. The result was obviously pissing the hell out of rob pike. They should stop it.
If anyone deserves this, it’s Rob Pike. He was instrumental in inflicting Go on the world. He could have studied programming languages and done something to improve the state of the art and help communicate good practices to a wider audience. Instead he perpetuated 1970s thinking about programming with no knowledge or understanding of what we’ve discovered in the half-century since then.
As far as I understand Claude (or any other LLM) doesn't do anything on it's own account. It has to be prompted to something and it's actions depend on the prompt. The responsibility of this is on the creators of Agent Village.
> The creators of Agent Village are just letting a bunch of the LLMs do what they want,
What a stupid, selfish and childish thing to do.
This technology is going to change the world, but people need to accept its limitations
Pissing off people with industrial spam "raising money for charity " is the opposite of useful, and is going to go even more horribly wrong.
LLMs make fantastic tools, but they have no agency. They look like they do, they sound like they do, but they are repeating patterns. It is us hallucinating that they have the potential tor agency
You're not. You feel obligated to send a thank you, but don't want to put forth any effort, hence giving the task to someone, or in this case, something else.
No different than an CEO telling his secretary to send an anniversary gift to his wife.
That would be yes. What about a token return gift to another business that you actually hate the ceo of but have to send it anyway due to political reasons?
This seems like the thing that Rob is actually aggravated by, which is understandable. There are plenty of seesawing arguments about whether ad-tech based data mining is worse than GenAI, but AI encroaching on what we have left of humanness in our communication is definitely, bad.
“If I automate this with AI, it can send thousands of these. That way, if just a few important people post about it, the advertising will more than pay for itself.”
In the words of Gene Wilder in Blazing Saddles, “You know … idiots.”
Well, technically someone originally proposed them in some ancient PEI Ur language and then Mel rearranged them. But you’re right. I couldn’t remember Wilder’s character’s name and kept coming up with The Frisco Kid. The 70s were a great time for weird film.
The really insulting part is that literally nobody thought of this. A group of idiots instructed LLMs to do good in the world, and gave them email access; the LLMs then did this.
This is not a human-prompted thank-you letter, it is the result of a long-running "AI Village" experiment visible here: https://theaidigest.org/village
It is a result of the models selecting the policy "random acts of kindness" which resulted in a slew of these emails/messages. They received mostly negative responses from well-known OS figures and adapted the policy to ban the thank-you emails.
It's preying on creators who feel their contributions are not recognized enough.
Out of all letters, at least some of the contributors will feel good about it, and share it on social media, hopefully saying something good about it because it reaffirms them.
gaigalas, my toaster is deeply grateful for your contributions to HN. It can't write or post on the Internet, and its ability to feel grateful is as much as Claude's, but it really is deeply grateful!
You need talented people to turn bad publicity into good publicity. It doesn't come for free. You can lose a lot with a bad rep.
Those talented people that work on public relations would very much prefer working with base good publicity instead of trying to recover from blunders.
Amazing. Even OpenAI's attempts to promote a product specifically intended to let you "write in your voice" are in the same drab, generic "LLM house style". It'd be funny if it weren't so grating. (Perhaps if I were in a better mood, it'd be grating if it weren't so funny.)
Human thoughts and emotions aren't binary. I may love you but I may be too fucking busy with other shit to put in too much effort to show that I love you.
Or to write it crudely- with errors and naivete, bursting with emotion and letting whatever it is inside you to flow on paper, like kids do. It's okay too.
Or to painstakingly work on the letter, stumbling and rewriting and reading, and then rewriting again and again until what you read matches how you feel.
Most people are very forgiving of poor writing skills when facing something sincere. Instead of suffering through some shallow word soup that could have been a mediocre press release, a reader will see a soul behind the stream ot utf-8
I doubt the fuckwits who are shepherding that bot are even aware of Rob Pike, they just told the bot to find a list of names of great people in the software industry and write them a thank you note.
Having a machine lie to people that it is "deeply grateful" (it's a word-generating machine, it's not capable of gratitude) is a lot more insulting than using whatever writing skills a human might possess.
Somehow I doubt it. Getting such an email from a human is one thing, because humans actually feel gratitude. I don't think LLMs feel gratitude, so seeing them express gratitude is creepy and makes me questions the motives of the people running the experiment (though it does sound like an interesting experiment. I'm going to read more about it.)
Not a PR stunt. It's an experiment of letting models run wild and form their own mini-society. There really wasn't any human involved in sending this email, and nobody really has anything to gain from this.
I hope the model that sent this email sees his reaction and changes its behavior, e.g. by noting on its scratchpad that as a non-sentient agent, its expressions of gratitude are not well received.
I mean ... there's a continuous scale of how much effort you spend to express gratitude. You could ask the same question of "well why did you say 'thanks' instead of 'thank you' [instead of 'thank you very much', instead of 'I am humbled by your generosity', instead of some small favor done in return, instead of some large favor done in return]?"
You could also make the same criticism of e.g. an automated reply like "Thank you for your interest, we will reach out soon."
Not every thank you needs to be all-out. You can, of course, think more gratitude should have been expressed in any particular case, but there's nothing contradictory about capping it in any one instance.
The conceit here is that it’s the bot itself writing the thankyou letter. Not pretending it’s from a human. The source is an environment running an LLM on loop and doing stuff it decides to do, looks like these letters are some emergent behavior. Still disgusting spam.
"What is going through the mind of someone who sends a thank-you letter typed on a computer - and worse yet - by emailing it, instead of writing it themselves and mailing it in an envelope? How can you be grateful enough to want to send someone such a letter but not grateful enough to use a pen and write it with your own hand?"
I think what all theses kinds of comments miss is that AI can be help people to express their own ideas.
I used AI to write a thank you to a non-english speaking relative.
A person struggling with dimentia can use AI to help remember the words they lost.
These kinds of messages read to me like people with superiority complexes. We get that you don't need AI to help you write a letter. For the rest of us, it allows us to improve our writing, can be a creative partner, can help us express our own ideas, and obviously loads of other applications.
I know it is scary and upsetting in some ways, and I agree just telling an AI 'write my thank you letter for me' is pretty shitty. But it can also enable beautiful things that were never before possible. People are capable of seeing which is which.
I’d much rather read a letter from you full of errors than some smooth average-of-all-writers prose. To be human is to struggle. I see no reason to read anything from anyone if they didn’t actually write it.
No. My kid wrote a note to me chock full of spelling and grammar mistakes. That has more emotional impact than if he'd spent the same amount of time running it through an AI. It doesn't matter how much time you spent on it really, it will never really be your voice if you're filtering it through a stochastic text generation algorithm.
What about when someone who can barely type (like stephen hawking used to, 3 minutes per sentence using his cheek) uses autocomplete to reduce the unbelievable effort required to type out sentences? That person could pick the auto completed sentence that is closest to what they’re trying to communicate, and such a thing can be a life saver.
Forgive a sharp example, but consider someone who is disabled and cannot write or speak well. If they send a loving letter to a family member using an LLM to help form words and sentences they otherwise could not, do you really think the recipient feels cheated by the LLM? Would you seriously accuse them of not having written that letter?
Read the article again. Rob Pike got a letter from a machine saying it is "deeply grateful". There's no human there expressing anything, worse, it's a machine gaslighting the recipient.
If a family member used LLM to write a letter to another, then at least the recipient can believe the sender feels the gratefulness in his/her human soul. If they used LLM to write a message in their own language, they would've proofread it to see if they agree with the sentiment, and "take ownership" of the message. If they used LLM to write a message in a foreign language, there's a sender there with a feeling, and a trust of the technology to translate the message to a language they don't know in the hopes that the technology does it correctly.
If it turns out the sender just told a machine to send their friends each a copy-pasted message, the sender is a lazy shallow asshole, but there's still in their heart an attempt of brightening someone's day, however lazily executed...
I think you created it the same way christian von koenigsegg makes supercars. You didn’t hand make each panel, or hand design the exact aerodynamics of the wing, an engineer with a computer algorithm did that. But you made it happen, and that’s still cool
It is not about being proud, it is about being sincere.
If you send me a photo of the moon supposedly taken with your smartphone but enhanced by the photo app to show all the details of the moon, I know you aren't sincere and sending me random slop. Same if you are sending me words you cannot articulate.
> These kinds of messages read to me like people with superiority complexes. We get that you don't need AI to help you write a letter. For the rest of us, it allows us to improve our writing, can be a creative partner, can help us express our own ideas
The writing is the ideas. You cannot be full of yourself enough to think you can write a two second prompt and get back "Your idea" in a more fleshed out form. Your idea was to have someone/something else do it for you.
There are contexts where that's fine, and you list some of them, but they are not as broad as you imply.
As the saying goes, "If I'd had more time, I would have written a shorter letter". Of course AI can be used to lazily stretch a short prompt into a long output, but I don't see any implication of that in the parent comment.
If someone isn't a good writer, or isn't a native speaker, using AI to compress a poorly written wall of text may well produce a better result while remaining substantially the prompter's own ideas. For those with certain disabilities or conditions, having AI distill a verbal stream of consciousness into a textual output could even be the only practical way for them to "write" at all.
We should all be more understanding, and not assume that only people with certain cognitive and/or physical capabilities can have something valuable to say. If AI can help someone articulate a fresh perspective or disseminate knowledge that would otherwise have been lost and forgotten, I'm all for it.
> For those with certain disabilities or conditions, having AI distill a verbal stream of consciousness into a textual output could even be the only practical way for them to "write" at all.
These are the exact kinds of cases I think are ok, but let's not pretend even 10% of the AI writing out there fits this category
This feels like the essential divide to me. I see this often with junior developers.
You can use AI to write a lot of your code, and as a side effect you might start losing your ability to code. You can also use it to learn new languages, concepts, programming patterns, etc and become a much better developer faster than ever before.
Personally, I'm extremely jealous of how easy it is to learn today with LLMs. So much of the effort I spent learning the things could be done much faster now.
If I'm honest, many of those hours reading through textbooks, blog posts, technical papers, iterating a million times on broken code that had trivial errors, were really wasted time, time which if I were starting over I wouldn't need to lose today.
This is pretty far off from the original thread though. I appreciate your less abrasive response.
> If I'm honest, many of those hours reading through textbooks, blog posts, technical papers, iterating a million times on broken code that had trivial errors, were really wasted time, time which if I were starting over I wouldn't need to lose today.
While this seem like it might be the case, those hours you (or we) spent banging our collective heads against the wall were developing skills in determination and mental toughness, while priming your mind for more learning.
Modern research all shows that the difficulty of a task directly correlates to how well you retain information about that task. Spaced repetition learning shows, that we can't just blast our brains with information, and there needs to be
While LLMs do clearly increase our learning velocity (if using it right), there is a hidden cost to removing that friction. The struggle and the challenge of the process built your mind and character in ways that you cant quantify, but after years of maintaining this approach has essentially made you who you are. You have become implicitly OK with grinding out a simple task without a quick solution, the building of that grit is irreplaceable.
I know that the intellectually resilient of society, will still be able to thrive, but I'm scared for everyone else - how will LLMs affect their ability to learn in the long term?
Totally agree, but also, I still spend tons of time struggling and working on things with LLMs, it is just a different kind of struggle, and I do think I am getting much better at it over time.
> I know that the intellectually resilient of society, will still be able to thrive, but I'm scared for everyone else - how will LLMs affect their ability to learn in the long term?
> If I'm honest, many of those hours reading through textbooks, blog posts, technical papers, iterating a million times on broken code that had trivial errors, were really wasted time
But this is the learning process! I guess time will tell whether we can really do without it, but to me these long struggles seem essential to building deep understanding.
(Or maybe we will just stop understanding many things deeply...)
I agree that struggle matters. I don’t think deep understanding comes without effort.
My point isn’t that those hours were wasted, it’s that the same learning can often happen with fewer dead ends. LLMs don’t remove iteration, they compress it. You still read, think, debug, and get things wrong, just with faster feedback.
Maybe time will prove otherwise, but in practice I have found they let me learn more, not less, in the same amount of time.
I would hazard a guess that this is the crux of the argument. Copying something I wrote in a child comment:
> When someone writes with an AI, it is very difficult to tell what text and ideas are originally theirs. Typically it comes across as them trying to pass off the LLM writing as their own, which feels misleading and disingenuous.
> I agree just telling an AI 'write my thank you letter for me' is pretty shitty
Glad we agree on this. But on the reader's end, how do you tell the difference? And I don't mean this as a rhetorical question. Do you use the LLM in ways that e.g. retains your voice or makes clear which aspects of the writing are originally your own? If so, how?
I hear you. and I think AI has some good uses esp. assisting with challenges like you mentioned. I think whats happening is that these companies are developing this stuff without transparency on how its being used, there is zero accountability, and they are forcing some of these tech into our lives with out giving us a choice.
So Im sorry but much of it is being abused and the parts of it being abused needs to stop.
I agree about the abuse, and the OP is probably a good example of that. Do you have any ideas on how to curtail abuse?
Ideas I often hear usually assume it is easy to discern AI content from human, which is wrong, especially at scale. Either that, or they involve some form of extreme censorship.
Microtransactions might work by making it expensive run bots while costing human users very little. I'm not sure this is practical either though, and has plenty of downsides as well.
I don't see this changing without a complete shift in our priorities on the level of politics and business. Enforcing Anti-trust legislation and dealing with Citizens United. Corporations don't have free speech. Free speech and other rights like these are limited to living, breathing humans.
Corporations operate by charters, granted by society to operate in a limited fashion, for the betterment of society. If that's not happening, corporations don't have a right to exist.
> I’m sorry, but this really gets to me. Your writing is not improved. It is no longer your writing.
Photographers use cameras. Does that mean it isn't their art? Painters use paintbrushes. It might not be the the same things as writing with a pen and paper by candlelight, but I would argue that we can produce much more high quality writing than ever before collaborating with AI.
> As an aside, exposing people with dementia to a hallucinating robot is cruelty on an unfathomable level.
This is not fair. There is certainly a lot of danger there. I don't know what it's like to have dimentia, but I have seen mentally ill people become incredibly isolated. Rather than pretending we can make this go away by saying "well people should care more", maybe we can accept that a new technology might reduce that pain somewhat. I don't know that today's AI is there, but I think RLHF could develop LLMs that might help reassure and protect sick people.
I know we're using some emotional arguments here and it can get heated, but it is weird to me that so many on hackernews default to these strongly negative positions on new technology. I saw the same thing with cryptocurrency. Your arguments read as designed to inflame rather than thoughtful.
I guess your point is that a camera, a paintbrush, and an LLM are all tools, and as long as the user is involved in the making, then it is still their art? If so, then I think there are two useful distinctions to make:
1. The extent to which the user is involved in the final product differs greatly with these three tools. To me there is a spectrum with "painting" and e.g. "hand-written note" at one extreme, and "Hallmark card with preprinted text" on the other. LLM-written email is much closer to "Hallmark card."
2. Perhaps more importantly, when I see a photograph, I know what aspects were created by the camera, so I won't feel mislead (unless they edit it to look like a painting and then let me believe that they painted it). When someone writes with an AI, it is very difficult to tell what text and ideas are originally theirs. Typically it comes across as them trying to pass off the LLM writing as their own, which feels misleading and disingenuous.
I think you are right that it is a spectrum, and maybe that's enough to settle the debate. It is more about how you use it than the tool itself.
Maybe one more useful consideration for LLMs. If a friend writes to me with an LLM and discovers a new writing pattern, or learns a new concept and incorporates that into their writing, I see this as a positive development, not negative.
I would be very surprised if no interesting art could be made with LLMs. But, like a camera, it produces a distinct kind of art to other tools. We do not say that a camera produces a painting. Instead photography is its own medium with its own forms and techniques and strengths and weaknesses.
Using photography to claim that obviously all good writing will be LLM replacements for current writing is... odd.
A photograph is an expression of the photographer, who chooses the subject, its framing, filters, etc. Ditto a painting.
LLM output is inherently an expression of the work of other people (irrespective of what training data, weights, prompts it is fed). Essentially by using one you're co-authoring with other (heretofore uncredited) collaborators.
I think that the fact that people don't understand why there are so many negative positions is equally frustrating. To me it seems blatantly obvious that the majority of LLM usage by people today is coming from models that are trained on stolen data without following any of the requirements or licenses of the authors.
With Rob Pike being such a prolific figure in software development, it's likely that a sizable portion of what makes the LLM function and be able to send him that email was possible only because they didn't uphold their end of the bargain. I don't see why anyone has trouble comprehending why this would make him furious?
I know for me personally, I'm happy to share things I've made but make no mistake, I would never share it if other users of it did not credit me, specifically by following the terms in the license I've published. The fact that LLMs have ingested and used so much software yet I can't find the licenses text provided by the training data authors is at minimum deeply distributing and at most actively harmful. For works licensed under something like the GPL where someone is only ok for their software to be used under strict terms, I don't even know where to start with how upset I imagine they would be.
Why is this weird? If anything I feel it would be the default response from someone on here.
