Rohan Paul· @rohanpaul_ai · X·· 5 小时前AI 评分51
AI 导读
Rohan Paul 转述了一项名为 RL with Neural Feedback（RLNF）的方法，称研究者找到了仅凭脑信号训练 AI 模型响应人类偏好的可靠方式。示例中，一张人脸图片会在受试者注视时自动变成对其更有吸引力的版本。作者称，图像模型通过读取受试者观看时的脑反应来学习其个人审美偏好，再据此修改图片以获得更高评分，全程无需评分、点击或提示词。
正文
this is truly next level stuff.
RL with Neural Feedback (RLNF)
An AI image model learns what you personally find attractive by reading your brain’s reaction as you look, then changes the picture to score higher, with no ratings, clicks or prompts from you.
Introducing RL with Neural Feedback (RLNF). We have found reliable ways to train AI models to respond to human preferences purely from brain signals. The example below is a picture of a face that self-transforms into a more attractive version for the subject as they are looking at it.在 X 查看被引用的帖子
来源：Rohan Paul · x.com