Live data from Hacker News

Expanding on what we missed with sycophancy

openai.com

291–297 of 297 posts

Re: Expanding on what we missed with sycophancy

#291
post #157

Earlier quoted context omitted.

Empathy does not lie in its perception on receipt but in its inception as a feeling. It is fundamentally a manifestation of the modalities enabled in shared experience. As such impossible to the extent that our experiences are not compatible with those of an intelligence that does not put emphasis on lived context trying to substitute for it with offline batch learning. Understanding is possible in this relationship,…

> As such impossible to the extent that our experiences are not compatible with those of an intelligence that does not put emphasis on lived context trying to substitute for it with offline batch learning. Conversely that means empathy is possible to the extent that our experiences are compatible with those of an AI. That is precisely what's under consideration here and you have not shown that it is zero. an intellig…

Lived context is to me more than online learning. I admit I am not so versed in the space as to be able to anticipate the nature of context in the case of online learning, so, yes, indeed I may change my tune if it somehow makes learning more of an experience rather than an education. My understanding is it won’t. I have not proven, but argued that experience compatibility is zero, to the extent a Lim does not experience anything. Happy to accept alternative viewpoints and accordingly that someone may perceive something as a sign of empathy whether it is or not.

Re: Expanding on what we missed with sycophancy

#292

Earlier quoted context omitted.

They have already introduced ads btw. “The "Enshittification" has arrived I asked ChatGPT about the impact of the current tarrifs on inventories over the next few months. It returned a long list of links to toiletries I might want to buy. I asked it why it did that. It replied: "As of April 28, 2025, OpenAl introduced new shopping features to ChatGPT, enhancing its capabilities to provide product recommendations comp…

Are you able to link the chat, I would like to see that.

https://www.reddit.com/r/ChatGPT/comments/1kan9c1/the_enshit...

Re: Expanding on what we missed with sycophancy

#293
post #157

Earlier quoted context omitted.

Empathy does not lie in its perception on receipt but in its inception as a feeling. It is fundamentally a manifestation of the modalities enabled in shared experience. As such impossible to the extent that our experiences are not compatible with those of an intelligence that does not put emphasis on lived context trying to substitute for it with offline batch learning. Understanding is possible in this relationship,…

I happen to agree with what you said. (Paraphrasing: A machine cannot have "real empathy" because a machine cannot "feel" in general.) But I think you're arguing a different point from the grandparent's. rurp said: > Someone who sees a person stranded on the side of a road might feel for them and stop to lend a hand. ChatGPT will never do that [...] Now, on the one hand that's because ChatGPT cannot "see a person" no…

Yes, I am not in fact expanding on GPs argument but etymologically attack the premise. Pathos is not learnt. When I clutch my legs at the sight of someone getting kicked in the balls, that’s empathy. When, as now, I write about it, it’s not, even in my case where I have lived experience of it. More sophisticated kinds of empathy build on the foundations of these gut-driven ones. Thank you for reading recommendation, will look for it.

Re: Expanding on what we missed with sycophancy

#294

Earlier quoted context omitted.

As mentioned I’m in the rather unique position to have analyzed the complete browser history for a substantial number of people - I have learned far more than I wished to. The behaviors of very high IQ people are rather distinct from regular IQ people due to IQ being both largely generic and those genes having other comorbidities. Most obviously is the depression, anxiety, and bipolar disorders. This is so obvious th…

What gets me the most about this is how certain you are that because you have some data about people’s browser history you can make these conclusions. I hope age and experience avails you of these notions - there is no master database that will tell you the nature of people. All you have is the data you have - I’m sure there are many interesting insights and confounding results, but it’s foolish to think you can know…

[dead]

Re: Expanding on what we missed with sycophancy

#296

Earlier quoted context omitted.

see my other comments about the trustworthiness about asking a chat system how it's internals work. They have reason to be cagey.

Your personifying a statistical engine. LLMs aren't cagey. They can't be.

They can when there are entire teams dedicated to adding guardrails via hidden system prompts and running all responses through other LLMs trained on flagging and editing certain things before the original output gets relayed to the user.

Re: Expanding on what we missed with sycophancy

#297
This sort of thing highlights for me how what people see as "AI" is not some kind of "raw" AI that is making these decisions for itself about how to respond. There are humans turning knobs to make the AI behave a certain way. The AI will be a sycophant or not based on how the humans tune it.

On the one hand, this could be comforting, since it means humans are still in control on some level and it's not a SkyNet situation. On the other hand, it's horrible, because it means users think they're dealing with some kind of "autonomous" AI, when in fact they're dealing with a product that may have deliberately built-in biases to sell it certain products, push certain viewpoints, or whatever. When you use ChatGPT you're not using "AI"; you're just using an OpenAI product that OpenAI can and will manipulate to get the best outcome for OpenAI.

Post reply on HN