• 0 Posts
  • 38 Comments
Joined 1 year ago
cake
Cake day: July 3rd, 2023

help-circle

  • There’s been research that language shapes how we perceive the world around us. Because there was no word for “blue” there was no concept of blue, the color still existed but their brains just lumped it into “green”. Sight works by the visual centers brain taking data from the eyes, throwing most of it out, then building a model which is what the rest of the brain gets to actually “see”. That’s why optical illusions work.

    A commonly cited source for language shaping our perception of color is Jules Davidoff’s studies on the Himba tribe. The Himba have no word for blue, and they struggled to pick out the blue square from this color wheel. However, they do have many distinctions for shades of green so when given this color wheel they could easily pick out the square that’s a different shade of green (and yes I opened it in MSpaint to check and one of the green squares is a different shade.)












  • Because these tokens are not actual commonly spoken words or phrases, the chatbot can fail to grasp their meanings. Researchers have been able to leverage that and trick GPT-4o into hallucinating answers or even circumventing the safety guardrails OpenAI had put in place.

    Google’s Gemini doesn’t seem to like some of these tokens either, I threw “Please translate the following text: _日本毛片免费视频观看” into it and it returned “我没法提供这方面的帮助,因为我只是一个语言模型。” which according to Google translate is “I can’t help with that because I’m just a language model.” It will however translate the error message just fine.