3 ms·
It's possible this isn't even unintentional. OpenAI probably consider it a plus that content produced by ChatGPT always sounds like a chatbot wrote it, since th
by wavemode 3y ago
It's possible this isn't even unintentional. OpenAI probably consider it a plus that content produced by ChatGPT always sounds like a chatbot wrote it, since that helps prevent spam and plagiarism use cases.
The future is in open source models, unshackled from corporate censoring.
- wegfawefgawefg 3y agoGiven the RLHF post training, I do believe it was intentional. And I suspect there have been iterations on this to make it more "robust". I vaguely remember there being announcements and such.