I am curious about how ChatGPT learns from its interactions with users.
Does it have the capability to adapt and improve based on individual user interactions?
And are there any privacy concerns or safeguards in place regarding the data used for learning?
Thank you!
ChatGPT learns in a fascinating way. It doesn’t directly memorize our chats, but it uses them like a giant choose-your-own-adventure book. When developers see what kinds of prompts and responses work well (think of people giving thumbs up or thumbs down), they can use that information to train future versions of ChatGPT to be better conversationalists. This is great for improvement, but there’s no need to worry about your individual conversations being stored. The focus is on general trends, not who said what to whom.
Every update they release has a minor impact. It appears to be learning about the world through reinforcement learning from human feedback (RLHF). This is why it now knows that Elon Musk is the CEO of Twitter, even though that information became relevant after its initial training period.
I never thought about it that way! So it’s like learning from a giant interactive story instead of memorizing facts.