4 ms·
I’ve always liked to speculate that using polite language is rewarded by openAI either by elevating the customers client score elevating the resources available
by ionwake 3y ago
I’ve always liked to speculate that using polite language is rewarded by openAI either by elevating the customers client score elevating the resources available to them if there uneven is such a thing.
I have no idea why but maybe it’s the kind of thing I’d push for if I was CEO to try and help make the world a more civil place maybe even as some form of Easter egg or light hearted prank.
Just my 2 cents
- yccs27 3y agoHumans often also reward politeness by keeping a "sympathy" score and elevating the resources (time, thought and effort) used to be helpful ;) Since ChatGPT training data includes human conversations, it might be that it's just reproducing this.
- J_Shelby_J 3y agoFascinating thought that motivation could accidentally be added to a model.
- J_Shelby_J 3y agoThis is a ray of sunshine in my day. I hope it's true!
- ionwake 3y agoWhat a lovely comment have a great day bro
- hmage 3y agoYou have two documents on internet: First document is an forum thread full of "go fuck yourself fucking do it", and in this kind of scenario, people are not cooperative. Second document is a forum thread full of "Please, take a look at X", and in this kind of scenario, people are more cooperative. By adding "Please" and other politness, you are sampling from dataset containing second document style, while avoiding latent space of first document style - this leads to a model response that is more accurate and cooperative. Hope that explains.
- ionwake 3y agoIt really does and now it makes sense. Thank you for the clarification.