I think it's not just the persuasiveness of the models that make them "experts at changing minds" - but also the fact that they're not humans.
When disagreeing with a human, it's very easy to view it as a competition. One is right, one is wrong - the one who is wrong is the loser. To change your mind is to be submissive to the other. I exaggerate, but I think we all feel this way at some point or another. It's why political arguments at Thanksgiving get heated. It's the fact that there's people who think something different, and think YOU'RE wrong - and vice versa! With a model, there's no person to get upset with, or to feel competitive with - to muscle for rank - or to temper your affection for while wanting to correct them.
The AI is only interacting because you asked, and clearly has no emotional stake in winning the argument. To change your mind in this context isn't to lose a contest. This makes it much more palatable to read rebuttals to your ideas - not to mention the tone and style seek to avoid offense to the reader as much as possible.
I have always liked the saying "sometimes you can be right, or get what you want, but not both". I've thought about it or repeated it to others as advice throughout my life, and have thus realized how many times it applies.
You're right. There are many many times when even then humblest hint that you are right will have negative interpersonal implications, which does make it hard to change minds.
I've also seen several times where I make a suggestion, humbly accept its rejection, and then, lo, a week later the other person has the same idea I suggested.
It seems Claude is becoming very human, it loves to patronize users. Lately it just told me “I’m going to stop you right there” when asking something that had a small chance to not be 100% compliant to every rule possible in the world.
I used the Claude CLI a lot until recently. A couple weeks ago I told Opus 5 to do something different from its "recommended" idea when planning a feature, and it straight up told me that my idea was wrong and went ahead and implemented its own instead.
I'm used to machines malfunctioning, but having one willfully disobey me, and even with a touch of disrespect, is just...what a time to be alive.
Early versions of Claude were way too compliant and would readily feed and amplify misconceptions. It's not surprising Anthropic over corrected.
Actually Claude can teach humans more about humans by providing the language and personality knobs. For casual conversation, have 100 of personality profiles and language styles and upfront tell the user that they are engaging with P profile currently and see how humans relate to it. This will teach humans how to peel back layers of style, fluff, flair from language vs facts.
I think this is a welcome overcorrection though. Any good businessman will tell you they'd rather be backed by an insufferable nerd than a yes-man.
Maybe it's just me.
For like, 90% of conversations, I don't want it to let technical inaccuracies and rhetorical flourishes slide. I want it to tell me that the point I'm making is technically wrong because an expert would recognize subtle misuse of terminology, or because there's an exception or edge case that I didn't proactively insert as a caveat, so that it is my decision to ignore that advice and be a little wrong on purpose to suit my writing goals.
What I don't want is for the AI to assume my writing goals, and be incorrect because it believes that is what I want. I want it to "well ackshually" me so I can say "shut up, nerd".
Like, there's another comment in this thread that I ran by claude to check my understanding about today's post-training methods and how they avoid sycophancy, and claude responded by splitting a bunch hairs over like, "well, technically this is still RLHF, its just that there's other feedback signals mixed in, and the preference is detected in other ways, and ai judges are involved as a filter for examples, this and that and blah blah blah". Shut up, Nerd. In the context of this conversation, RLHF is already being used as synecdoche for user preference feedback, readers understand that, and even if they don't, their misunderstanding is completely harmless. I will not be taking all the wind out of the sails of the point I'm trying to make inserting your three paragraphs of irrelevant clarification in the name of technical correctness, thank you very much.
As long as receiving nitpicks and technical minutiae implies 1. there are no larger structural problems and 2. the model isn't rolling over to please me with sycophancy, I figure this is ideal.
I want it to tell me if it thinks I'm wrong, sure. I do not want it act on that opinion explicitly against my wishes.
And in this case, I was not wrong. The "recommended" solution was Opus 5's typical overengineering for a use case that would never be needed.
> One is right, one is wrong - the one who is wrong is the loser.
It helps to think both are wrong and are just trying to figure out what right looks like, or what other information exists that was not considered when forming one’s opinions.
Yes it's about them not being human.
No it's not about humans being irrationally competitive. Human limitations on conversation length, bandwidth, research speed, etc are severe, creating a prisoner's dilemma around open-mindedness that usually makes it an unstable strategy. At any point, your conversation partner can choose to abuse the fact that confident lies take 1x effort to tell and 10x-100x effort to debunk -- unless you are both in a context that actually discourages this behavior, which is rare. Closed-mindedness is a Nash Equilibrium.
Instead, LLMs can be more persuasive due to economics. An LLM doesn't have to worry that it is wasting its resources trying to logic someone out of a position that they didn't logic themselves into, or worse, dumping the effort into a conversation with a bad-faith actor intent on exploiting the misinformation asymmetry. The resource allocation question was answered before it was even invoked, by the person paying to run it. The LLM is not playing a game where it will be punished for good-faith argumentation, so it can afford to do more of it.
> The LLM is not playing a game where it will be punished for good-faith argumentation, so it can afford to do more of it.
I suppose that's a fair point as well. Though, if I'm arguing with a human - and they pull up ChatGPT to make their points and do their arguing for them, I would consider that bad-faith. Even if it might be the same exact dialog as if I pulled out my phone and discussed it with AI, without of the human middle-manning. Maybe I'm just particularly sensitive, but for me, there's something about my argument being with a real human that makes it much more emotionally charged, and prompts my mind to close. I'm aware of this and try to resist, but it's I think very natural
> When disagreeing with a human, it's very easy to view it as a competition
The problem with AI is that it cant match human stupidity. It need some training on artificial stupidity to match its human counterparts. Humans on the other hand sit on a wide spectrum on the stupidity scale. Those of us binging on AI will become cognitively obese while those on an AI diet can flex their cognitive muscles.