Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

I'm absolutely right

absolutelyright.lol

Post reply on HN
Print view View original

161–170 of 277 posts

Re: I'm absolutely right

#161

Post by braebo »

kuschku wrote: post #114
kaffekaka wrote: post #87

Earlier quoted context omitted.

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 ma…

> Are you really asking in good faith? I am — I grew up being bullied, and my therapists taught me that I shouldn't even let humans affect me in this way and instead should let it slide and learn to ignore it, or even channel my emotions into defiance. Which is why I'm genuinely curious (and a bit bewildered) how people who haven't taken that path are going through life.

We are all influenced by the external world whether we like it or not. The butterfly effect is an extreme example, but a direct interaction with anything, especially a talking rock, will influence us. Our outputs are a function of our inputs.

That said, being aware of the inputs and their effects on us, and consciously asserting influence over the inputs from within our function body, is incredibly valuable. It touches on mindfulness practices, promoting self awareness and strengthening our independence. While we can’t just flip a switch to be sociopaths fundamentally unaffected by others, we can still practice self awareness, stoicism, and strengthen our resolve as your therapist seems to be advocating for.

For those lacking the kind of awareness promoted by these flavors of mindfulness, the hypnotic effects of the storm are much more enveloping, for better or (more often) worse.

Re: I'm absolutely right

#162

Post by stevenkkim »

For me, a really annoying tick in Cursor is how it often says "Perfect!" after completing a task, especially if it completely fails to execute the prompt.

So I told Cursor, "please stop saying 'perfect' after executing a task, it's very annoying." Cursor replied something like, "Got it, I understand" and then I saw a pop-up saying it created a memory for this request.

Then immediately after the next task, it declares "Perfect!" (spoiler: it was not perfect.)

Re: I'm absolutely right

#163

Post by gpm »

pessimizer wrote: post #67
latexr wrote: post #4

As I opened the website, the “16” changed to “17”. This looked interesting, as if the data were being updated live just as I loaded the page. Alas, a refresh (and quick check in the Developer Tools) reveals it’s fake and always does the transition. It’s a cool effect, but feels like a dirty trick.

Reminds me that the reason that loading spinners spin is so that you knew that the loading/system hadn't frozen. That was too hard (you actually had to program something that could understand that it had frozen), so it was just replaced everywhere with an animation that doesn't tell you anything and will spin until the sun burns out. Progress!

I've definitely had systems freeze badly enough that our modern dumb spinners stop spinning... so at least they're still some sort of signal.

Re: I'm absolutely right

#164

Post by bryanrasmussen »

trjordan wrote: post #41

OK, so I love this, because we all recognize it. It's not fully just a tic of language, though. Responses that start off with "You're right!" are alignment mechanisms. The LLM, with its single-token prediction approach, follows up with a suggestion that much more closely follows the user's desires, instead of latching onto it's own previous approach. The other tic I love is "Actually, that's not right." That happens…

>if it sees an error, the "Actually, ..." change in approach.

AI-splaining is the worst!

Re: I'm absolutely right

#165

Post by jcims »

trjordan wrote: post #41

OK, so I love this, because we all recognize it. It's not fully just a tic of language, though. Responses that start off with "You're right!" are alignment mechanisms. The LLM, with its single-token prediction approach, follows up with a suggestion that much more closely follows the user's desires, instead of latching onto it's own previous approach. The other tic I love is "Actually, that's not right." That happens…

>The other tic I love is "Actually, that's not right." That happens because once agents finish their tool-calling, they'll do a self-reflection step.

I saw this a couple of days ago. Claude had set an unsupported max number of items to include in a paginated call, so it reduced the number to the max supported by the API. But then upon self-reflection realized that setting anything at all was not necessary and just removed the parameter from the code and underlying configuration.

Re: I'm absolutely right

#166

Post by Eextra953 »

It would be nice if we can add another a plot to track when claude says "genuinely". It uses for almost all long responses, to the point that I can pretty much recognize when someone uses claude by looking for any instances of "genuinely".

Re: I'm absolutely right

#167

Post by al_borland »

trjordan wrote: post #41

OK, so I love this, because we all recognize it. It's not fully just a tic of language, though. Responses that start off with "You're right!" are alignment mechanisms. The LLM, with its single-token prediction approach, follows up with a suggestion that much more closely follows the user's desires, instead of latching onto it's own previous approach. The other tic I love is "Actually, that's not right." That happens…

In my experience, once it starts telling me I’m right, we’re already going downhill and it rarely gets better from there.

Re: I'm absolutely right

#168

Post by jcims »

trjordan wrote: post #41

OK, so I love this, because we all recognize it. It's not fully just a tic of language, though. Responses that start off with "You're right!" are alignment mechanisms. The LLM, with its single-token prediction approach, follows up with a suggestion that much more closely follows the user's desires, instead of latching onto it's own previous approach. The other tic I love is "Actually, that's not right." That happens…

It'd be nice if the chat-completion interfaces allowed you to seed the beginning of the response.

Re: I'm absolutely right

#169

Post by steveklabnik »

kuschku wrote: post #51
ZaoLahma wrote: post #34

Earlier quoted context omitted.

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…

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?

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.

Re: I'm absolutely right

#170

Post by GrumpyGoblin »

Man, the number of times Claude has told me this when I was absolutely wrong should also be a count on this. I've deliberately been wrong just to get that sweet praise. Still the best AI code sidekick though.
Post reply on HN