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

Memory and new controls for ChatGPT

openai.com

Post reply on HN
Print view View original

71–80 of 270 posts

Re: Memory and new controls for ChatGPT

#71

Post by danShumway »

isaacisaac wrote: post #42
__loam wrote: post #40

Earlier quoted context omitted.

I love when people express frustration with this shitty stochastic system and others respond with things like "no no, you need to whisper the prompt into its ear and do so lovingly or it won't give you the output it wants"

People skills are transferrable to prompt engineering

For example, my coworkers have also been instructed to never talk to me except via code comments.

Come to think of that, HR keeps trying to contact me about something I assume is related, but if they want me to read whatever they're trying to say, it should be in a comment on a pull request.

Re: Memory and new controls for ChatGPT

#72

Post by BonoboIO »

denysvitali wrote: post #57
minimaxir wrote: post #45

Earlier quoted context omitted.

Did you try promising it a $500 tip for behaving correctly? (not a shitpost: I'm working on a more academic analysis of this phenomenon)

Did the tipping trend move to LLMs now? I thought there wasn't anything worse than tipping an automated checkout machine, but now I realize I couldn't be more wrong

Wow, you are right, never occurred to me, but yes LLM tipping is a thing now.

I have tried to bribe it with tips to ngos and it worked. More often I get full code answers instead of just parts.

Re: Memory and new controls for ChatGPT

#73

Post by bluish29 »

minimaxir wrote: post #45
bluish29 wrote: post #28

It is already ignoring your prompt and custom instructions. For example, If I explicity ask it to provide a code instead of an overview it will respond by apologizing and then provide the same overview answer with minimal if no code. Will memory provide a solution to that or will be a different thing to ignore?

Did you try promising it a $500 tip for behaving correctly? (not a shitpost: I'm working on a more academic analysis of this phenomenon)

Great, I would be interesting to read your findings. I will tell you what I tried to do.

1- Telling it that this is important, and I will reward it if its successes.

2- Telling it is important and urgent, and I'm stressed out.

3- Telling it that they're someone future and career on the edge.

4- Trying to be aggressive and express disappointment.

5- Tell that this is a challenge and that we need to prove that you're smart.

6- Telling that I'm from a protected group (was testing what someone here suggested before).

7- Finally, I tried your suggestion ($500 tip).

All of these did not help but actually gave different output of overview and apologies.

To be honest, most of my coding questions are about using CUDA and C, so I would understand that even a human will be lazy /s

Re: Memory and new controls for ChatGPT

#75

Post by mhink »

Crespyl wrote: post #3
cl42 wrote: post #2

I love this idea and it leads me to a question for everyone here. I've done a bunch of user interviews of ChatGPT, Pi, Gemini, etc. users and find there are two common usage patterns: 1. "Transactional" where every chat is a separate question, sort of like a Google search... People don't expect memory or any continuity between chats. 2. "Relationship-driven" where people chat with the LLM as if it's a friend or colle…

Personally, I always expect every "conversation" to be starting from a blank slate, and I'm not sure I'd want it any other way unless I can self-host the whole thing. Starting clean also has the benefit of knowing the prompt/history is in a clean/"known-good" state, and that there's nothing in the memory that's going to cause the LLM to get weird on me.

Looks like you'll be able to turn the feature off:

> You can turn off memory at any time (Settings > Personalization > Memory). While memory is off, you won't create or use memories.

Re: Memory and new controls for ChatGPT

#76

Post by dylan604 »

danielbln wrote: post #8
hackerlight wrote: post #6

Earlier quoted context omitted.

OpenAI says they don't train on data from enterprise customers

They say they don't train on: - Any API requests - ChatGPT Enterprise - ChatGPT Teams - ChatGPT with history turned off

As long as it runs in the cloud, there is no way of knowing that is true. As you mentioned, "they say" requires a lot of faith to me.

Re: Memory and new controls for ChatGPT

#77

Post by kibwen »

henry2023 wrote: post #58
sorokod wrote: post #52

Earlier quoted context omitted.

Interesting, promising sexual services doesn't work anymore?

Gpt will now remember your promises and ignore any further questions until settlement

Contractor invoices in 2024:

Plying ChatGPT for code: 1 hour

Providing cybersex to ChatGPT in exchange for aforementioned code: 7 hours

Re: Memory and new controls for ChatGPT

#78

Post by BigParm »

Often I’ll play dumb and withhold ideas from ChatGPT because I want to know what it thinks. If I give it too many thoughts of mine, it gets stuck in a rut towards my tentative solution. I worry that the memory will bake this problem in.

Re: Memory and new controls for ChatGPT

#79

Post by phkahler »

BonoboIO wrote: post #72
denysvitali wrote: post #57

Earlier quoted context omitted.

Did the tipping trend move to LLMs now? I thought there wasn't anything worse than tipping an automated checkout machine, but now I realize I couldn't be more wrong

Wow, you are right, never occurred to me, but yes LLM tipping is a thing now. I have tried to bribe it with tips to ngos and it worked. More often I get full code answers instead of just parts.

>> I have tried to bribe it with tips to ngos and it worked.

Am I still in the same universe I grew up in? This feels like some kind of Twilight Zone episode.

Re: Memory and new controls for ChatGPT

#80

Post by madamelic »

BigParm wrote: post #78

Often I’ll play dumb and withhold ideas from ChatGPT because I want to know what it thinks. If I give it too many thoughts of mine, it gets stuck in a rut towards my tentative solution. I worry that the memory will bake this problem in.

Yep.

Hopefully they'll make it easy to go into a temporary chat because it gets stuck in ruts occasionally so another chat frequently helps get it unstuck.

Post reply on HN