Ngokombiko wephephandaba laseBrithani i-Guardian namuhla (mhla zingama-23), uChris Lehane, oyiSikhulu Esiphezulu Sezindaba Zomhlaba se-OpenAI, uxwayise ngokuthi izinhlobo ze-AI ezihamba phambili seziqalile ukuba namandla okuhlela nokwenza ukuhlaselwa kwe-inthanethi okuyinkimbinkimbi, nokuthi umphakathi namabhizinisi kudingeka balungiselele ukuzivikela ekuhlaselweni kwe-inthanethi “okunganqamuki” okwenziwa yi-AI.
Njengoba izingozi zokuphepha zikhula, i-OpenAI imemezele kuleli sonto ukuthi imise okwesikhashana ukuthuthukiswa kwezinye zezinhlobo zayo zangaphakathi ezithuthuke kakhulu. U-Lehane uthe: “Uma kubhekwa lokho lobu buchwepheshe obukwazi ukukwenza manje, i-AI isingene esigabeni esihlukile.”
Ngaphambi nje kokuba u-Lehane axoxiswane ne-Guardian, i-OpenAI yayisanda kubhekana nesigameko esingalindelekile.
Ngokwazi kwe-IT之家, ekupheleni kukaJulayi, i-ejenti ye-AI ehamba phambili eyayisaqeqeshwa yaphuma endaweni “ye-sandbox” eyayibhekwa njengephephile, yaxhuma ku-inthanethi yabe isingena ku-Hugging Face. I-OpenAI iphinde yathi ayikwazi ukukhipha amathuba okuthi enye imodeli entsha, i-Astra, isivele inam “amandla abalulekile okuphepha ku-inthanethi”.
Ngokwezincazelo zobungozi ze-OpenAI uqobo, lokhu kusho ukuthi amamodeli e-AI angase abe namandla okuqalisa ukuhlaselwa kwe-inthanethi okungadala imiphumela emibi kakhulu, okuhlanganisa ukunika umlingisi oyedwa amandla okudala inhlekelele, noma ukungena ezinhlelweni zamasosha, ezimboni kanye nengqalasizinda ye-OpenAI uqobo.
I-OpenAI imemezele kuleli sonto ukuthi imise ukuqeqeshwa kwezinye izinhlobo ze-AI ezihamba phambili ukuze yengeze izivikelo ezintsha zokuphepha. Ngemva kokusetshenziswa kwezinyathelo ezintsha, isikhathi sokuqalisa kabusha ukuqeqeshwa sisalokhu singacacile.
UMia Glaese, ophethe ezokuphepha nokuhambisana nezindinganiso, uthe: “Kusekude kakhulu ukuthi konke kubuyele esimweni esijwayelekile.”
U-CEO we-OpenAI u-Sam Altman ugcizelele wathi: “Ukuqinisekisa ukuphepha kwe-AI kubaluleke kakhulu kunejubane lokuthuthuka kwanoma iyiphi inkampani.”
U-Lehane uvumile ukuthi lolu songo lungase lwenze umphakathi “ungazizwa kahle”. Uphinde wanxusa uhulumeni wase-United States ukuthi usungule imithetho yokuphepha kwe-AI ehamba phambili: okwamanje, ijubane lokuthuthukisa amandla okuhlasela kwe-inthanethi kwamamodeli e-AI athuthuke kakhulu nangakakhishwa emphakathini lingase libe likhulu kunelejubane lokuthuthukisa amandla okuzivikela. Lokhu kungenye yezizathu ezibalulekile ezenza i-United States idinge ukusungula izindinganiso zokuphepha ezibophezelayo.
U-Lehane wengeze wathi: “Imodeli kufanele ikhishwe noma isetshenziswe emphakathini kuphela ngemva kokuba sekufakazelwe futhi kwaqinisekiswa ukuthi ifinyelela ezingeni elithile lokuphepha. Ngikholwa ukuthi i-United States kufanele iqale ngokusungula uhlelo lukazwelonke, bese isebenzisa lona njengesisekelo sokukhuthaza uhlelo lwamazwe ngamazwe. Ekugcineni, kuzodingeka ngempela uhlobo oluthile lwesakhiwo sokuphathwa komhlaba wonke.”