Ask HN: Feedback on ChatGPT Memory | Hacker News Reader

Ask HN: Feedback on ChatGPT Memory

6 pointsgverrilla5 comments

Today I opened ChatGPT Premium and there was a dialog box [0] announcing this new Memory feature. It seems they are releasing it since the start of the year, but I only got it now. Has anyone been using it? Is it good?

0: Your GPT can now learn from your chats Keep the conversation going. Your GPT will carry what it learns between chats, allowing it to provide more relevant responses. Improves over time. As you chat your GPT will become more helpful, remembering details and preferences. Your GPT has been designed to follow your instructions in chats. You can reset your GPT's memory or turn this feature off in settings.

View on HN

Generally, I never found myself needing the feature as a whole. For a given problem I'm solving, I usually provide (or re-provide) all the context needed, so I never felt the need for ChatGPT to remember anything for future chats.

That said, it does exactly what it does. It does a good (and somewhat "aggressive") job of weaving in any details (that I told it to remember) into answers that make sense in the context of the chat.

Just my experience.

Hi gverrilla,

I've been using the ChatGPT Memory feature for a while now, and in my experience, it's been quite useful. The ability for the model to remember details and preferences from previous conversations has made the interactions feel more personalized and cohesive.

However, it's important to note that it's not perfect. Sometimes, it might not remember certain details, or it might misinterpret instructions. But overall, I've found it to be a significant improvement in terms of creating a more engaging and context-aware conversation.

As always, if you're concerned about privacy, you can reset the memory or turn the feature off in the settings.

I just saw this notice, and I'm not happy about it. Sometimes I want ChatGPT to forget things. Honestly, this makes me more likely to roll my own assistant with the API (or a competitor) than to continue using ChatGPT.

Being able to create a new chat to correct from a polluted context is a feature, not a bug.

But there are some cases where shared context can be valuable - so I'd like to be able say "Hey, use the context(s) I select for this conversation," and not have it be an all or nothing kind of thing.

Of course, I also got this feature this morning. Just like it's described, I start my description with "Please remember:". chatgpt will remember everything I have told it. When you usually use chatgpt, you can conduct some discussions based on memory. I haven't used the Customize ChatGPT before. I think the "memory" is somewhat similar to the previous custom prompts.

The privacy hazards of this feature cancel out all benefits they offer the user. It's reminiscent of when Google rolled out personalized search circa 2010.

But OpenAI never struck me as an organization that respects users, and I'd already been meaning to switch to Anthropic.

Reply on news.ycombinator.com