Labeling bayanai
Wani bincike na mujallar TIME ya bayyana cewa don gina tsarin tsaro game da abun ciki mai guba (misali jima'i, tashin hankali, wariyar launin fata, jima'i, da dai sauransu), OpenAI ta yi amfani da ma'aikatan Kenya da ke waje da ke samun kasa da dala 2 a cikin sa'a don lakabi abun ciki mai guba.An yi amfani da waɗannan alamun don horar da ƙira don gano irin wannan abun ciki a nan gaba.Ma'aikatan da aka fitar da su sun fuskanci irin wannan abun ciki mai guba da haɗari wanda suka kwatanta kwarewa a matsayin "azaba".Abokin hulɗar waje na OpenAI shine Sama, kamfani-bayanan horo da ke San Francisco, California.
Watsewa
ChatGPT tana ƙoƙarin ƙin yarda da tsokaci wanda zai iya keta manufofin abun ciki.Koyaya, wasu masu amfani sun sami nasarar karya ChatGPT ta hanyar amfani da dabaru daban-daban na injiniya don ƙetare waɗannan hane-hane a farkon Disamba 2022 kuma sun yi nasarar yaudarar ChatGPT don ba da umarnin yadda ake ƙirƙirar hadaddiyar giyar Molotov ko bam ɗin nukiliya, ko cikin haifar da jayayya a cikin salon neo-Nazi.Wani dan jaridar Toronto Star ya sami nasara mara daidaituwa na samun ChatGPT don yin kalamai masu tayar da hankali jim kadan bayan ƙaddamar da shi: ChatGPT an yaudare shi don amincewa da mamayewar Rasha na 2022 na Ukraine, amma ko da lokacin da aka nemi ya yi wasa tare da wani labari na almara, ChatGPT ya yi magana game da haifar da hujjar dalilin da yasa Firayim Ministan Kanada Justin Trudeau ya kasance da laifin cin amanar kasa.(wiki)
Lokacin aikawa: Fabrairu-18-2023