Muva nje I-OpenAI, inkampani engenzi nzuzo yocwaningo lwezobunhloli wazise ngencwadi amathuluzi okulinganisa okuqukethwe Okusha Nokuthuthukisiwe (Ithuluzi Lokwengamela Okuqukethwe), elibalula linembe kakhudlwana futhi kuhloswe ngalo ukusiza onjiniyela be-API bavikele izinhlelo zabo zokusebenza.
Ithimba labacwaningi be-OpenAI limemezele isibuyekezo sethuluzi lamahhala lokulinganisela okuqukethwe manje esitholakala kubo bonke onjiniyela.
Isiphetho sokulinganisela ivumela ukufinyelela ngesixhumi esibonakalayo sohlelo lohlelo lokusebenza kuzihlukanisi ezisuselwa ku-OpenAI ezikhiqizayo ezisuselwa ku-transformer, ayini amamodeli e-AI abaye baqeqeshwa ukuthola okuqukethwe okungafunwa kuzinhlelo zokusebenza, inkampani yathi eposini lebhulogi. Lezi zihlukanisi zithola noma yikuphi okuqukethwe okungadingekile ezinhlelweni zokusebenza.
Inkampani ibeke lesi sibuyekezo ukwenza ngamandla ezinhlobonhlobo zezinhlelo zokusebenza, okuhlanganisa amanethiwekhi omphakathi, izinhlelo zemiyalezo nama-AI chatbots.
Uma usunikeze umbhalo ofakiwe, Isiphetho sokulinganisela sihlaziya okuqukethwe njengenkulumo enenzondo, okuqukethwe kwezocansi, ulimi oluyinhlamba, njll., okumele kuhlungwe. Vimba konke okuqukethwe (kukhiqizwe i-OpenAI API) lokho okuphambene nenqubomgomo yokuqukethwe kwe-OpenAI. Ukwengeza, iphoyinti lokugcina lingaphinda lisuse futhi livimbe okuqukethwe okuyingozi okukhiqizwe umuntu.
Indawo yokugcina iqeqeshelwe ukuthi isheshe, inembile, futhi iqine ezinhlelweni ezihlukahlukene zokusebenza. Okubalulekile, lokhu kunciphisa amathuba okuthi imikhiqizo "isho" into engalungile, ngisho noma ikhishelwa kubasebenzisi esikalini. Njengomphumela, i-AI ingakwazi ukuvula izinzuzo ezindaweni ezibucayi, njengemfundo, lapho ibingeke isetshenziswe ngokuzethemba.
Inguqulo ethuthukisiwe yephoyinti lokugcina le-Moderation yakhelwe ukuthi isheshe, inembile futhi iqine kuzo zonke izinhlobo eziningi zezinhlelo zokusebenza, okufaka phakathi ama-chatbots anikwe amandla yi-AI, izinhlelo zemiyalezo, nezingosi zokuxhumana nabantu.
Okubalulekile, i-OpenAI ithe iwanciphisa kakhulu amathuba okuthi imodeli ye-AI "isho" into engalungile. Lokho kusho ukuthi i-AI ingasetshenziswa kuzilungiselelo ezibucayi kakhulu, njengezinhlelo zokusebenza zemfundo, lapho abantu okungenzeka babenokubhukha ngaphambilini mayelana nokusebenzisa ubuchwepheshe.
Kuyafaneleka ukusho lokho ithuluzi limahhala uma lisetshenziswa nokuqukethwe okukhiqizwa i-OpenAI API. Ngaphezu kokulinganisa ama-bots, ungaphinda uvimbele okuqukethwe okuyingozi okungakhiqizwe ama-OpenAI API, kodwa ngabantu. Inkundla yemiyalezo engaziwa i-NGL, ehlinzeka intsha ngenkundla yokwabelana ngemizwa nemibono yayo, isebenzisa ithuluzi le-OpenAI ukuthola inkulumo enenzondo nokuhlukumeza. I-NGL ithe leli thuluzi linekhono eliyingqayizivele lokuhlanganisa ulimi lwakamuva, elivumela ukuthi lihambisane nokuvela kolimi olusetshenziswa yintsha, isibonelo.
Kubalulekile futhi ukusho ukuthi esimweni sethrafikhi okungeyona ye-API, indawo yokugcina yokulinganisa ingaphansi kwenkokhelo.
Njengengxenye yokuzibophezela kwe-OpenAI ekwenzeni i-AI ye-ecosystem ivikeleke kakhulu, sinikeza lesi siphetho ukuze sikwazi ukulinganisela kwamahhala kwakho konke okuqukethwe okukhiqizwa i-OpenAI API. Isibonelo, i-Inworld, iklayenti le-OpenAI API, isebenzisa indawo yokugcina yokulinganisa ukuze isize izinhlamvu zayo ezisekelwe ku-AI "zihlale kuskripthi." Ngokusebenzisa ubuchwepheshe be-OpenAI, i-Inworld ingagxila emkhiqizweni wayo oyinhloko: ukudala izinhlamvu ezikhumbulekayo.
I-OpenAI yasho lokho abathuthukisi sebengaqala ukuyisebenzisa ngokubheka imibhalo yayo. Iphinde yashicilela idokhumenti enemininingwane ngenqubo yayo yokuqeqeshwa nokusebenza, kanye nedatha yokuhlola ethe izogqugquzela ucwaningo olwengeziwe endaweni yokulinganisa okunamandla e-AI.
Okokugcina uma unentshisekelo yokwazi kabanzi ngakho, ungabheka imininingwane ku- isixhumanisi esilandelayo.