AI safety teams whenever someone mentions releasing an AI prior to installing their 5000 censorship hurdles
๐คฃ6๐1๐คฌ1
OpenAI: Weโre launching a classifier trained to distinguish between AI-written and human-written text.
In our evaluations on a โchallenge setโ of English texts, our classifier correctly identifies 26% of AI-written text (true positives) as โlikely AI-written,โ while incorrectly labeling human-written text as AI-written 9% of the time (false positives). Our classifierโs reliability typically improves as the length of the input text increases.
https://platform.openai.com/ai-text-classifier
In our evaluations on a โchallenge setโ of English texts, our classifier correctly identifies 26% of AI-written text (true positives) as โlikely AI-written,โ while incorrectly labeling human-written text as AI-written 9% of the time (false positives). Our classifierโs reliability typically improves as the length of the input text increases.
https://platform.openai.com/ai-text-classifier
๐30๐ฅ6๐ฑ4๐3๐ฅฐ3โ2๐คฌ2๐คก1