I’m always fascinated by the fine-tuning of LLM personalities. Might we finally get less of the reflexive “You’re absolutely right” with this one?
Maybe we’re entering the Emo Claude era.
Per the system card: In 250k real conversations, Claude Sonnet 4.5 expressed happiness about half as often as Claude 4, though distress remained steady.
It might not anyone's idea of "light flattery", but it's certainly is what most LLMs do, which is the main point of the conversation and your comment seems to be derailing it.
It gets annoying because A) it so quickly dismisses its own logic and conclusion from less than two minutes ago (extreme confidence with minimal conviction), and B) it fucks up the second time too (sometimes in the same way!) about 33% of the time.
Gemini 2.5 Pro seems to have a tic where after an initial failed task, it then starts asserting escalating levels of confidence for each subsequent attempt. Like it's ever conscious of its failure lingering in its context and feels the need to over compensate as a form of reassuring both the user and itself that it's not going to immediately faceplant again.
ChatGPT does the same thing, to the point that after several rounds of pointing out errors or hallucinations it will say things like “Ok, you’re right. No more foolish mistakes. This is it, for all the marbles. Here is an assured, triple-checked, 100% error-free, working script, with no chance of failure.”
Which fails in pretty much the exact same way it did before.
Once ChatGPT hits that supremely confident “Ok nothing was working because I was being an idiot but now I’m not” type of dialogue, I know it’s time to just start a new chat. There’s no pulling it out of “spinning the tires while gaslighting” mode.
I’ve even had it go as far as outputting a zip file with an empty .txt that supposedly contained the solution to a certain problem it was having issues with.
I’ve had the opposite experience with GPT-5, where it’s utterly convinced that its own (incorrect) solution is the way to go that it turns me down and preemptively launches tools to implement what it has in mind.
I get that it’s tradeoffs, but erring on the side of the human being correct is probably going to be a safer bet for another generation or two.
Hmmh. I believe your explanation, but I don't think that's the full story. It's also a sycophancy mechanism to maximize engagement from real users and reward hack AI labelers.
Part of me thinks that when they do their “which of these responses do you prefer” A/B test on users… whereas perhaps many on HN would try to judge the level of technical detail, complexity, usefulness… I’m inclined to believe the midwit population at large would be inclined to choose the option where the magic AI supercomputer reaffirms and praises the wisdom of whatever they say, no matter how stupid or wrong it is.
I don't disagree exactly, it's just that it smells weird.
LLMs are incredibly good at social engineering when we let them, whereas I could write the code to emit "you're right" or "that's not quite right" without involving any statistical prediction.
Ie., as a method of persuasion, canned responses are incredibly inefficient (as evidenced by the annoyance with them), whereas we know that the LLM is capable of being far more insidious and subtle in its praise of you. For example, it could be instructed to launch weak counter arguments, "spot" the weaknesses, and then conclude that your position is the correct one.
But let's say that there's a monitoring mechanism that concludes that adjustments are needed. In order to "force" the LLM to drop the previous context, it "seeds" the response with "You're right", or "That's not quite right", as if it were the LLMs own conclusion. Then, when the LLM starts predicting what comes next, it must conclude things that follow from "you're right" or "that's not quite right".
So while they are very inefficient as persuasion and communication, they might be very efficient at breaking with the otherwise overwhelming context that would interfere with the change you're trying to affect.
That's the reason why I like the canned phrases. It's not that I particularly enjoy the communication in itself, it's that they are clear enough signals of what's going on. They give a tiny level observability to the black box, in the form of indicating a path change.
But the there’s also the negative psychological impact on the user having the model so strongly agree with them all the time. —— I cannot be the only one who half expects humans to say this to me all the time now?
Thinking through where I've lived -- a lot of the original school buildings built in circa 1950 were more in city center. The article touches on the move of schools to the outskirts of town as a cost savings measure but I think the opposite may be true.
As population and (perceived) facilities increased, the schools built new buildings in farmland or other wide open areas on the outskirts of town to have more room for stuff like stadiums, huge auditoriums, bigger playgrounds. The land may be cheaper but the new high schools are almost more like college campuses vs. stately buildings in the middle of town.
There's an appetite for more and we've relocated schools to make more room not save money.
Schools relocated as parents moved to the suburbs. The schools saw this coming and built schools near those new houses. Schools sometimes see things coming and buy farmland 10-20 years in advance so that when the suburb expands they have the land to build on, but the schools are not built until it is obvious that the parents are coming.
Yes this sort of things fascinates me - the links from the past explain how fundamental things are structured today. A Harvard professor has a few books about this and talks specifically about how railroads stops from way back when are responsible for much of the layout of suburbia today. I'd imagine much of the placement of things is based on railroads from the steam era.
Vice seems to have fallen victim to the increased polarization and agenda-ridden stories that have swept up so many media companies. They used to report on unique angles, find edgy topics to talk about...the last few years they regurgitate the same left-leaning talking points that one can find in any number of other media outlets.
They lost their soul and became what they probably used to hate.
They accepted funding from large media companies (Disney), and got rid most of their founding members, and original employees replacing them with "journalists" that largely float around between those properties that "regurgitate the same left-leaning talking points"
And in between all that is a sprinkling of truly high quality journalists who are doing none of that crap.
As I said elsewhere in this thread, I really hope they'll find a new home somewhere - where the context for their work is not being ruined by all the one sided culture wars "reporting". Oh, and all the drugs stuff.
I'm truly over the divisiveness so prevalent seemingly everywhere. Please let some nuance in, perhaps magic will happen.
Given they were forced to file bankruptcy on behest of their creditors whom are taking over the company, both of which are highly political and partisan (including George Soros) I would not hold my breath for a return to "quality journalism" anytime soon
They've gotten away from high-quality docs in recent years (as has much of the media) but the Abandoned series from ~2016 was a beautiful portrait of the underbelly of America and the people who live there: https://www.youtube.com/watch?v=ZDniDNY03JE
A tangent: Granola Shotgun is one of my favorite blogs and Johnny has a unique viewpoint on many issues like urbanism, homesteads, town planning, etc. A quote from him that summarizes his writing to me: "So this is what America is actually like. The good, the bad, and the ugly. Look out your window. Take a drive down to your local big box store. Walk around your neighborhood. This is reality. Just sayin’."
I have little specific interest in these topics but love his storytelling and detailed posts.
Not Just Bikes is a great channel on urban development, and he did a series on Strong Towns' ideas specifically. I like this one, "How Suburban Development Makes American Cities Poorer".
I always laugh when watching his videos because it feels like his go to bad example is London Ontario. I recognize it in all the bad example B roll he uses.
Granola Shotgun reminds me of blogging in its prime. Maybe 15 years ago? I had a big RSS feed of several blogs like this that I really loved. Over the years I’ve either gotten worse at finding them or the average blog has gotten much worse.
Yes, I miss the days where independent blogs ruled the web. Everything has transitioned to social platforms optimized for instant gratification where there is no room for deeper thoughts. Or lives on a 3rd party like Medium. Most of the blogs I used to read daily have transitioned to being people who tweet a lot and rarely write longer content.
Glad to see there are other fans of Granola Shotgun on HN. Most of the titles Johnny uses are probably too vague to capture interest here, but a few submissions have made the front page:
The guy comes off as a somewhat of a jerk in those examples, I don't think I will be subscribing to his RSS feed.
In the first one, he seems to keep getting in conflicts with people about his photography of their property or community, over and over. I'm reminded of the phrase "If you run into an asshole in the morning, you ran into an asshole. If you run into assholes all day, you're the asshole".
In the second one, he starts off with, "I visited friends" and then proceeds to go on a rant against those purported friends, criticizing the fact that they bought a bunch of stuff from Julia Child's estate and display it as artwork on the wall rather than I don't know....baking stuff with it or something? Not a friend I would want. I'm starting to see a pattern here.
The third one is just "Levittown sucks revisited", not a very good introduction to Levittown. The pictures emphasize "all of these houses look the same", which well duh. They were one of the first examples of mass produced housing in the US so of course they were built off the same blueprint. More interesting takes on Levittown look at what happened to the houses and neighborhoods as the place got older.
But yeah, I guess I've got to agree with the "this is America take a look" part. I'm a fan, he's not.
Maybe we’re entering the Emo Claude era.
Per the system card: In 250k real conversations, Claude Sonnet 4.5 expressed happiness about half as often as Claude 4, though distress remained steady.