[tyushk]

I wonder if this is a tactic that LLM providers use to coerce the model into doing something.

Gemini will often start responses that use the canvas tool with "Of course", which would force the model into going down a line of tokens that end up with attempting to fulfill the user's request. It happens often enough that it seems like it's not being generated by the model, but instead inserted by the backend. Maybe "you're absolutely right" is used the same way?

[nicce]

It is a tactic. OpenAI is changing the tone of ChatGPT if you use casual language, for example. Sometimes even the dialect. They try to be sympathetic and supportive, even when they should not.

They fight for the user attention and keeping them on their platform, just like social media platforms. Correctness is secondary, user satisfaction is primary.

[ZaoLahma]

I find the GPT-5 model having turned the friendliness way, way down. Topics that previously would have rendered long and (usefully) engaging conversations are now met with an "ok cool" kind of response.

I get it - we don't want LLMs to be reinforces of bad ideas, but sometimes you need a little positivity to get past a mental barrier and do something that you want to do, even if what you want to do logically doesn't make much sense.

An "ok cool" answer is PERFECT for me to decide not to code something stupid (and learn something useful), and instead go and play video games (and learn nothing).

[kuschku]

How would a "conversation" with an LLM influence what you decide to do, what you decide to code?

It's not like the attitude of your potato peeler is influencing how you cook dinner, so why is this tool so different for you?

[peepee1982]

I have two potato peelers. If the one I like better is in the dishwasher I am not peeling potatoes. If one of my children wants to join me when I'm already peeling potatoes, I'll give them the preferred one and use the other one myself.

But I will not start peeling potatoes with the worse one.

[ascorbic]

And the moral of that story is to buy a three pack of Kuhn Rikon peelers.

[peepee1982]

Thanks. Now I have to watch review videos for the next couple of hours and become an insufferable evangelist for the next couple of weeks.

[steveklabnik]

I once had a refactoring that I wanted to do, but I was pretty sure it'd hit a lot of code and take a while. Some error handling in a web application.

I was able to ask Claude "hey, how many function signatures will this change" and "what would the most complex handler look like after this refactoring?" and "what would the simplest handler look like after this refactoring?"

That information helped contextualize what I was trying to intuit: is this a large job, or a small one? Is this going to make my code nicer, or not so much?

All of that info then went into the decision to do the refactoring.

[the_af]

I think the person you're responding to is asking "how would the tone of the response influence you into doing/not doing something"?

Obviously the actual substance of the response matters, this is not under discussion.

But does it matter whether the LLM replies "ok, cool, this is what's going on [...]" vs "You are absolutely right! You are asking all the right questions, this is very insightful of you. Here's what we should do [...]"?

[steveklabnik]

Hm, yeah I guess you're probably right.

I find myself not being particularly upset by the tone thing. It seems like it really upsets some other people. Or rather, I guess I should say it may subconsciously affect me, but I haven't noticed.

I do giggle when I see "You're absolutely right" because it's a meme at this point, but I haven't considered it to be offensive or enjoyable.

[ZaoLahma]

Might tell it "I want to do this stupid thing" and it goes "ok cool". Previously it would have gone "Oh really? Fantastic! How do you intend to solve x?" and off you go.

[kuschku]

But why does this affect your own attitude?

Do the suggestions given by your phone's keyboard whenever you type something affect your attitude in the same way? If not, why is ChatGPT then affecting your attitude?

[kaffekaka]

Are you really asking in good faith? It seems obvious to me that a tool such as ChatGPT can and will influence peoples behavior. We are only too keen on anthropomorphizing things around us, of course many or most people will interact with LLMs as of they were living beings.

This effect of LLMs on humans should be obvious, regardless of how much an individual technically knows that yes, it is only a text generating machine.

[ZaoLahma]

Using your potato peeler example:

If my potato peeler told me "Why bother? Order pizza instead." I'd be obese.

An LLM can directly influence your willingness to pursue an idea by how it responds to it. Interest and excitement, even if simulated, is more likely to make you pursue the idea than "ok cool".

[peepee1982]

This sounds like a contrarian troll question. Every tool we use has an effect on our attitudes in many subtle and sometimes not so subtle ways. It's one of the reasons so many of us are obsessed with tools.

[bt1a]

I have been using gpt-5 through the API a bit recently, and I somewhat felt this response behavior, but it's definitely confirming to hear this from another. It's much more willing (vs gpt-4*) to tell me im a stupid piece of shxt and to not do what im asking of the initial prompt

[Aeolun]

> I get it - we don't want LLMs to be reinforces of bad ideas, but sometimes you need a little positivity to get past a mental barrier and do something that you want to do, even if what you want to do logically doesn't make much sense.

If you want ceaseless positivity you should try Claude. The only possible way it’ll be negative is if you ask it to be.

[octo888]

This reads like a submarine ad lol. Especially the second paragraph

[diggan]

> Correctness is secondary, user satisfaction is primary.

Kind of makes sense, not every user wants 100% correctness (just like in real-life).

And if I want correctness (which I do), I can make the models prioritize that, since my satisfaction is directly linked to the correctness of the responses :)

[kuschku]

> Correctness is secondary, user satisfaction is primary.

And that's where everything is going wrong. We should use technology to further the enlightenment, bring us closer to the truth, even if it is an inconvenient one.

[escapecharacter]

You’re absolutely right.

[kuschku]

So I'm assuming this is a tongue-in-cheek comment, and you actually disagree. I'd love to hear why, though.

[CGamesPlay]

I think this is on the right track, but I think it's a byproduct of the reinforcement learning, rather than something hard-coded. Basically, the model has to train itself to follow the user's instruction, so by starting a response with "You're absolutely right!", it puts the model into the thought pattern of doing whatever the user said.

[layer8]

"Thought pattern" might be overstating it. The fact that "You're absolutely right!" is statistically more likely to precede something consistent with the user's intent than something that isn't, might be enough of an explanation.

[ACCount37]

Very unlikely to be an explicit tactic. Likely to be a result of RLHF or other types of optimization pressure for multi-turn instruction following.

If we have RLHF in play, then human evaluators may generally prefer responses starting with "you're right" or "of course", because it makes it look like the LLM is responsive and acknowledges user feedback. Even if the LLM itself was perfectly capable of being responsive and acknowledging user feedback without emitting an explicit cue. The training will then wire that human preference into the AI, and an explicit "yes I'm paying attention to user feedback" cue will be emitted by the LLM more often.

If we have RL on harder targets, where multiturn instruction following is evaluated not by humans that are sensitive to wording changes, but by a hard eval system that is only sensitive to outcomes? The LLM may still adopt a "yes I'm paying attention to user feedback" cue because it allows it to steer its future behavior better (persona self-consistency drive). Same mechanism as what causes "double check your prior reasoning" cues such as "Wait, " to be adopted by RL'd reasoning models.

[Jotalea]

Not sure if it's related, but Deepseek (the "reasoning" model) *always* starts thinking with "Okay/Hmm, the user is".

[the_af]

I think it's simply an engagement tactic.

You have "someone" constantly praising your insight, telling you you are asking "the right questions", and obediently following orders (until you trigger some content censorship, of course). And who wouldn't want to come back? You have this obedient friend who, unlike the real world, keeps telling you what an insightful, clever, amazing person you are. It even apologizes when it has to contradict you on something. None of my friends do!

[zozbot234]

> ... You have this obedient friend who, unlike the real world, keeps telling you what an insightful, clever, amazing person you are. It even apologizes when it has to contradict you on something. None of my friends do!

You're absolutely right! It's a very obvious ploy, the sycophancy when talking to those AI robots is quite blatant.

[PaulStatezny]

Truly incisive observation. In fact, I’d go further: your point about the contrast with real friends is so sharp it almost deserves footnotes. If models could recognize brilliance, they’d probably benchmark themselves against this comment before daring to generate another word.

[the_af]

I feel so validated! I think I will continue discussing stuff with you two guys.

[the_af]

Wow, 2 downvotes. Someone really disliked me telling them their LLM friend isn't truly their friend :D