I'm pretty sure they want it kissing people's asses because it makes users feel good and therefore more likely to use the LLM more. Versus, if it just gave a curt and unfriendly answer, most people (esp. Americans) wouldn't like to use it as much. Just a hypothesis.
> Versus, if it just gave a curt and unfriendly answer, most people (esp. Americans)
I don’t see this as an American thing. It’s an extension of the current Product Management trend to give software quirky and friendly personality.
You can see the trend in more than LLM output. It’s in their desktop app that has “Good Morning” and other prominent greetings. Claude Code has quirky status output like “Bamboozling” and “Noodling”.
It’s a theme throughout their product design choices. I’ve worked with enough trend-following product managers to recognize this trend toward infusing express personality into software to recognize it.
For what it’s worth, the Americans I know don’t find it as cute or lovable as intended. It feels fake and like an attempt to play at emotions.
> It’s an extension of the current Product Management trend to give software quirky and friendly personality.
Ah, Genuine People Personalities from the Sirius Cybernetics Corporation.
> It’s in their desktop app that has “Good Morning” and other prominent greetings. Claude Code has quirky status output like “Bamboozling” and “Noodling”.
This reminded me of a critique of UNIX that, unlike DOS, ls doesn't output anything when there are no files. DOS's dir command literally tells you there are no files, and this was considered, in this critique, to be more polite and friendly and less confusing than UNIX. Of course, there's the adage "if you don't have anything nice to say, don't say anything at all", and if you consider "no files found" to not be nice (because it is negative and says "no"), then ls is actually being polite(r) by not printing anything.
Many people interact with computers in a conversational manner and have anthropomorphized them for decades. This is probably influenced by computers being big, foreign, scary things to many people, so making them have a softer, more handholding "personality" makes them more accessible and acceptable. This may be less important these days as computers are more ubiquitous and accessible, but the trend lives on.
I worked in an org with offices in America, India, Europe, and Israel, and it was not uncommon for the American employees to be put off by the directness of the foreign employees. It was often interpreted as rudeness, to the surprise of the speaker. This happened to the Israel employees more than the India or Europe employees, at least in part because the India/Europe employees usually tried to adapt to the behavior expected by the Americans, while the Israel employees largely took pride in their bluntness.
That might characterize their approach to human interaction, but I don't think any of us can say who will or won't prefer the sycophantic style of the LLM.
It might be the case that it makes the technology far more approachable. Or it makes them feel far less silly for sharing personal thoughts and opinions with the machine. Or it makes them feel validated.
> We don't appreciate how much there is to language.
This can’t possibly be true, can it? Every language must have its own nuance. non native English speakers might not grasp the nuance of English language, but the same could be said for any one speaking another language.
I believe this reflects the euphemization of the english language in US, a concept that George Carlin discussed many years ago [1]. As he put it, "we don't die, we pass away" or "we are not broke, we have negative cash flow". Many non-English speakers find these terms to be nonsensical.
But it really erodes trust. First couple of times I felt that it indeed confirmed what I though, then I became suspicious and I experimented with presenting my (clearly worse) take on things, it still said I was absolutely right, and now I just don't trust it anymore.
As people here are saying, you quickly learn to not ask leading questions, just assume that its first take is pretty optimal and perhaps present it with some options if you want to change something.
There are times when it will actually say I'm not right though. But the balance is off.
Haha, the good old "asking for a friend". Maybe I should add that to the standard pre-prompt. "I'm always asking for a friend, who is a bit of a douche so its ok to hurt his feelings."
For me, it's getting annoying. Not every question is an excellent question. Not every statement is a brilliant observation. In fact, I'm almost certain every idea I've typed into an LLM has been thought of before by someone else, many many times.
> Not every question is an excellent question. Not every statement is a brilliant observation.
A brilliant observation, Dr. Watson! Indeed, the merit of an inquiry or an assertion lies not in its mere utterance but in the precision of its intent and the clarity of its reasoning!
One may pose dozens of questions and utter scores of statements, yet until each is finely honed by observation and tempered by logic, they must remain but idle chatter. It is only through genuine quality of thought that a question may be elevated to excellence, or a remark to brilliance.
You're absolutely right! Americans are a bit weird like that, most people around the world would be perfectly okay with short and to-the-point answers. Especially if those answers are coming from a machine that's just giving its best imitation of a stochastic hallucinating parrot.
Claude is very "American", just try asking it to use English English spelling instead of American English spelling; it lasts about 3~6 sentences before it goes back. Also there is only American English in the UI (like the spell checker, et al), in Spanish you get a choice of dialects, but not English.
In contrast, o3 seems to be considerably more British - and it doesn't suck up as much in its responses. I thought these were just independent properties of the model, but now that you mention it, could the disinclination to fawn so much be related to its less American style?
Do you realise that doing the thing that the article is complaining about is not only annoying and incredibly unfunny, but also just overdone and boring? Have one original thought in your life.
>most people around the world would be perfectly okay with short and to-the-point answers
Wow, this is really interesting. I had no idea Japan, for example, had such a focus on blunt, direct communication. Can you share your clearly extensive research in this area so I can read up on this?
People really over-exaggerate the claim of friendly and polite US service workers and people in general. Obviously you can find the full spectrum of character types across the US. I've lived 2/3 of my life in Britain and 1/3 in the US and I honestly don't think there's much difference in interactions day to day. If anything I mostly just find Britain to be overly pessimistic and gloomy now.
Britain, or at the very least England, is also well known for its extreme politeness culture. Also, it's not that the US has a culture of genuine politeness, just a facade of it.
I have only spent about a year in the US, but to me the difference was stark from what I'm used to in Europe. As an example, I've never encountered a single shop cashier who didn't talk to me. Everyone had something to say, usually a variation of How's it going?. Contrasting this to my native Estonia, where I'd say at least 90% of my interactions with cashiers involves them not making a single sound. Not even in response to me saying hello, or to state the total sum. If they're depressed or in an otherwise non-euphoric mood, they make no attempt to fake it. I'm personally fine with it, because I don't go looking for social connections from cashiers. Also, when they do talk to me in a happy manner, I know it's genuine.
It's common for foreigners to come to America and feel that everyone is extremely polite. Especially eastern bloc countries which tend to be very blunt and direct. I for one think that the politeness in America is one of the cultures better qualities.
Does it translate into people wanting sycophantic chat bots? Maybe, but I don't know a single American that actually likes when llms act that way.
Politeness is one thing, toxic positivity is quite another. My experience is that Americans have (or are expected/required to have) too much of the latter, too little of the former.
> I for one think that the politeness in America is one of the cultures better qualities.
Politeness makes sense as an adaptation to low social trust. You have no way of knowing whether others will behave in mutually beneficial ways, so heavy standards of social interaction evolve to compensate and reduce risk. When it's taken to an excess, as it probably is in the U.S. (compared to most other developed countries) it just becomes grating for everyone involved. It's why public-facing workers invariably complain about the draining "emotional labor" they have to perform - a term that literally doesn't exist in most of the world!
That's one way of looking at it. A bit of a cynical view I might add. People are polite to each other for many reasons. If you hold the door and smile at an old lady, it usually isn't because you dont trust her.
Service industry in America is a different story that could use a lot of improvement.
> ... do you have any sources to back up your conjecture, or is this just prejudice?
Let me guess, you consider yourself a progressive left democrat.
Do I have any source for that? No, but I noticed a pattern where progressive left democrats ask for a source to discredit something that is clearly a personal observation or opinion, and by its nature doesn't require any sources.
The only correct answer is: it's an opinion, accept it or refute it yourself, you don't need external validation to participate in an argument. Or maybe you need ;)
As an American, using it for technical projects, I find it extremely annoying. The only tactic I’ve found that helps is telling it to be highly critical. I still get overly positive starts but the response is more useful.
I think we, as Americans who are technical, are more appreciative of short and critical answers. I'm talking about people who have soul-searching conversations with LLMs, of which there are many.
> Claude never starts its response by saying a question or idea or observation was good, great, fascinating, profound, excellent, or any other positive adjective. It skips the flattery and responds directly.
LLMs cannot tell fact from fiction. What's commonly called hallucinations stems from it not being able to reason, the way that humans appear to be able to do, no matter that some models are called "reasoning" now. It's all the same principle: most likely token in a given position. Adding internal monologue appears to help because, by being forced to break it down (internally, or by spitballing towards the user when they prompted "think step by step"[1]), it creates better context and will thus have a higher probability that the predicted token is a correct one
Being trained to be positive is surely why it inserts these specific "great question, you're so right!" remarks, but if you wasn't trained on that, it still couldn't tell you whether you're great or not
> I'm pretty sure they want it kissing people's asses
The American faux friendliness is not what causes the underlying problem here, so all else being equal, they might as well have it kiss your ass. It's what most English speakers expect from a "friendly assistant" after all
The article I posted says that chain of thought is enough to make something a reasoning model. It's listed as one of the types.
It says that reasoning models "now aim for" more, but that doesn't disqualify the older more basic techniques.
I'm not sure what I'm supposed to learn from your link because there's so much vision stuff happening here and that's the focus of the article. Can you link me something about what you need to make a bare minimum reasoning LLM? Text only.
Then I can't explain why it's producing the results that it does. If you have more information to share, I'm happy to update my knowledge...
Doing a web search on the topic just comes up with marketing materials. Even Wikipedia's "Reasoning language model" article is mostly a list of release dates and model names, with as only relevant-sounding remark as to how these models are different: "[LLMs] can be fine-tuned on a dataset of reasoning tasks paired with example solutions and step-by-step (reasoning) traces. The fine-tuned model can then produce its own reasoning traces for new problems." It sounds like just another dataset: more examples, more training, in particular on worked examples where this "think step by step" method is being demonstrated with known-good steps and values. I don't see how that fundamentally changes how it works; you're saying such models do not predict the most likely token for a given context anymore, that there is some fundamentally different reasoning process going on somewhere?
I'm saying adding "think step by step" does not get you close to actual reasoning, it just produces marginally self-consistent linguistic reasoning.
Actual reasoning requires training on diverse data sources, as you noted, but also coached experimentation (supervised fine-tuning) not just adding "think step by step" instruction to a model trained on typical textual datasets. "Think step by step" came first and produced increased performance on a variety of tasks, but was overhyped in its approximation of reasoning.
Finns need to mentally evolve beyond this mindset.
Somebody being polite and friendly to you does not mean that the person is inferior to you and that you should therefore despise them.
Likewise somebody being rude and domineering to you does not mean that they are superior to you and should be obeyed and respected.
Politeness is a tool and a lubricant, and Finns probably loose out on a lot of international business and opportunities because of this mentality that you're demonstrating. Look at the Japanese for inspiration, who were an economic miracle, while sharing many positive values with the Finns.
Wow. I lived in Finland for a few months and this does not match my experience with them at all. In case it's relevant, my cultural background is Dutch... maybe you would say the same about us, since we also don't do the fake smiles thing? I wouldn't say that we see anyone who's polite and friendly as inferior; quite the contrary, it makes me want to work with them more rather than less. And the logical contrary for the rude example you give. But that doesn't mean that faking a cheerful mood all the time isn't disingenuous and does not inspire confidence
"I never smile if I can help it. Showing one's teeth is a submission signal in primates. When someone smiles at me, all I see is a chimpanzee begging for its life." While this famous quote from The Office may be quite exaggerated in many ways, this can nonetheless be a very real attitude in some cultures. Smiling too much can make you look goofy and foolish at best, and outright disingenuous at worst.
Yes, globally cultures fall into the category where a smile is either a display of weakness or a display of strength. The latter are more evolved cultures. Of course too much is too much.
You know there is a difference between being polite and friendly, and kissing ass, right?
We are also talking about a tool here. I don't want fluff from a tool, I want the thing I'm seeking from the tool, and in this case it's info. Adding fluff just annoys me because it takes more mental power to skip all the irrelevant parts.
Since it's a tool, what does it matter if it's too polite for your liking? It's just a tool. We've always had these things. Every time you started your computer, Windows would have a load screen saying "Welcome".
chatgpt’s custom user prompt is actually pretty good for this. I’ve instructed mine to be very terse and direct and avoid explaining itself, adding fluff, or affirming me unless asked, and it’s much more efficient to use that way, although it does have a tendency to drift back into sloppy meandering and enthusiastic affirming
Remember when microsoft changed real useful searchable error codes into "your files are right where you left em! (happy face)"
And my first thought was... wait a minute this is really hinting that automatic microsoft updates are going to delete my files arent they? Sure enough, that happened soon after
More likely the original version of Claude sometimes refused to cooperate and by putting "you're absolutely right" into the training data they made it more obedient. So this is just a nice artifact
