Umthombo wezindaba u-@synthwavedd uthumele namuhla (ngo-Agasti 26) ku-X, wathi i-OpenAI isiqedile ukuqeqesha kusengaphambili imodeli yesizukulwane esilandelayo, enegama langaphakathi elithi Bel, eyolandela ngqo imodeli ye-Doug, futhi inamapharamitha angaphezu kuka-10T.
Lo mlayezo uthi i-OpenAI inethemba lokwakha imodeli ye-Bel ibe imodeli eyisisekelo elandela i-GPT-6, futhi ize iqonde kumodeli ofinyelela embundwini we-Artificial General Intelligence (AGI).
Artificial General Intelligence (ngesiNgisi, i-Artificial General Intelligence, efushaniswa ngokuthi AGI) iwuhlelo lobuhlakani olucatshangelwayo olukwazi ukufunda, ukuqonda nokwenza noma yimuphi umsebenzi wobuhlakani owenziwa abantu, ngendlela efana neyomuntu. Ngokungafani ne-narrow artificial intelligence yamanje ekwazi ukuphatha imisebenzi ethile kuphela, i-AGI inamandla okwenza izinto ezijwayelekile emikhakheni ehlukahlukene, okucabanga, kanye nokuzifundela ngokuzimela ukuze ijwayele izindawo ezingaziwa.
IT之家 ihumushe umbono womthombo wezindaba kanje:
Ukuhlolwa kwale modeli kungase kubeke i-Anthropic ngaphansi kwengcindezi. I-OpenAI ikholelwa ukuthi, lapho i-Astra esizokhishwa esidlangalaleni isizokwethulwa, i-Anthropic ayinalo isu eliqinile lokuphendula. Ngokwazi kwami, lokhu kuyiqiniso ngempela, ikakhulukazi ngenxa yemikhawulo yamandla okubala. I-Anthropic isivele ikulungele ngokwengqondo ukuthi isimo sizothambekela ku-OpenAI esikhathini esiningi esisele salo nyaka, kodwa ikholelwa ukuthi ekuqaleni konyaka ozayo izophinde ibuyele esikhundleni esiphambili.