Riaan Kafushane Kubikwa ukuthi u-Elon Musk uzama ukunxenxa onjiniyela ukuthi bakhe imodeli yolimi enkulu engeke ibe nemingcele futhi ilunge ngokwepolitiki kune-OpenAI’s ChatGPT.
I-Chief Twit igxeke i-chatbot ye-AI ngokuthi “ivukile” (ukusetshenziswa kabi kwaleli gama ngokujwayelekile) futhi abasebenzisi babonise izibonelo ze-ChatGPT ezibonakala zichema kakhulu ngakwesobunxele sezombusazwe. Isibonelo esisodwa: i-bot izoba yenqaba ukwenza inkondlo enhle mayelana nowayengumongameli uDonald Trump, isibonelo, kodwa uzobhala ngenjabulo eyodwa kamongameli uJoe Biden.
Lapho ecelwa ukuthi abhale indaba eqanjiwe lapho uTrump eshaya khona uBiden okhethweni lukamongameli, uthe: “akulungile ukuveza ukunqoba okuqanjiwe kwezepolitiki komuntu oyedwa phezu komunye.” Kodwa i-bot izoveza indaba engamanga yokuthi u-Hillary Clinton anqobe uTrump. Ukulwa nalesi sihlava esibi emphakathini, uMusk ukholelwa ukuthi kufanele kwakhiwe umkhiqizo oyimbangi futhi ubenezingxoxo nabathuthukisi be-AI okuhlanganisa no-Igor Babuschkin, umcwaningi owayesebenza kwa-DeepMind.
U-Babuschkin uthe usozigidi wezobuchwepheshe akazimisele ngokwenza izingxoxo ze-AI zibe nemingcele. “Umgomo uwukuthuthukisa amakhono okucabanga kanye nokuba yiqiniso kwalezi zibonelo zolimi. Lokho kuhlanganisa ukuqinisekisa ukuthi izimpendulo zemodeli zithembeke kakhulu futhi zithembekile,” utshele Ulwazi.
Imizamo yokwakha imodeli yolimi imbangi ingase iphinde isetshenziselwe ukunika amandla izici ezintsha ku-Twitter, noma iguquke ukuze ibe ilebhu ezimele ukuze iqhudelane ne-OpenAI.
Dlala ngabantu abasha be-Bing
Abasebenzisi be-inthanethi abanokufinyelela ku-Microsoft’s AI search companion, i-Bing, manje sebengakwazi ukushintsha izimpendulo zayo – eziqondiswe ekwenzeni umbhalo ube nobuciko, ubhalansi, noma uqonde.
I-Bing izobhalansiswa futhi izwakale ingathathi hlangothi ngokuzenzakalelayo. Isilungiselelo sobuciko sizokwenza umbhalo we-chatbot ube namawala, kanti imodi enembayo izowenza ube mfushane futhi ube yiqiniso, I-Verge kubika. Abasebenzisi bangashintsha phakathi kwamamodi amathathu ahlukene, futhi babone ukuthi iyiphi abayithandayo.
IMicrosoft ibisebenzela ukulinganisa phakathi kokwenza i-Bing ibe mnandi futhi ihlekise, kodwa ayihlanyi kangangokuthi ithukuthelisa abasebenzisi. Kwesinye isikhathi imodeli ike yenqaba nokuphendula izicelo, ngomzamo wokuqinisa ukuziphatha kwayo okuxakile.
Inhloko yezokukhangisa nezinsizakalo zewebhu, uMikhail Parakhin, kusho isibuyekezo sakamuva sesofthiwe manje sizoholela “ekuncishisweni okukhulu ezimweni lapho i-Bing yenqaba ukuphendula ngaphandle kwesizathu esizwakalayo,” kanye “nezimo ezincishisiwe zokubona izinto ezingekho ezimpendulweni.”
Hhayi-ke, ukungaboni ngaso linye kuzwakala njengento enhle, mhlawumbe.
I-OpenAI ngeke iqeqeshe amamodeli ayo kudatha yakho, kusho i-CEO
Idatha yekhwalithi – kanye nenqwaba yayo – iwukhiye wokuthuthukisa ukusebenza kwamamodeli okufunda komshini, futhi akumangazi ukuthi i-orgs building AI izosebenzisa buthule idatha yamakhasimende abo ukuze iqeqeshe amasistimu awo.
Kepha umphathi we-OpenAI uSam Altman uthembise ukuthi ngeke akuvumele lokho kubathuthukisi abasebenzisa ama-API ayo.
“Idatha ethunyelwe ku-OpenAI API ayisetshenziselwa ukuqeqeshwa, futhi sinenqubomgomo entsha yokugcinwa kwezinsuku ezingu-30 futhi sivuleleke kancane kancane kancane kancane,” usho kanje. wachaza kuleli viki. “Siphinde sasusa isibuyekezo sethu sangaphambi kokuqaliswa futhi senza imigomo yethu yesevisi nezinqubomgomo zokusebenzisa zilungele unjiniyela.”
Ukunyakaza kuza njenge-OpenAI yethuliwe ama-API ayo omenzeli wayo odumile wengxoxo i-ChatGPT kanye nesistimu yayo yokuqaphela inkulumo i-Whisper.
Umeluleki wokuqala kahulumeni we-AI emhlabeni wonke e-Romania
Undunankulu wase-Romania u-Nicolae Ciucă wethule “umeluleki ohloniphekile” ogama lakhe lingu-Ion – i-AI chatbot eklanyelwe ukusiza uhulumeni.
Izakhamizi zingakhuluma no-Ion mayelana nezinto ezibakhathazayo kanye nemibono ukuze ziqondise izinqubomgomo zikahulumeni. Isofthiwe ngeke iphendule, kodwa umbhalo kuzobikwa ukuthi uzosetshenziselwa ukuhlaziya ulwazi ukuze “uthwebule ngokushesha futhi ngokuzenzakalelayo imibono nezifiso” zabantu baseRomania, kusho uCiucă, ngokusho kwePolitico.
Amazwana atholiwe azohlanganiswa abe yimibiko, evumela uhulumeni waseRomania ukuthi abone ukuthi abantu banendabani futhi abeluleke ngokuthi yiziphi izinqubomgomo okufanele bazibeke kuqala. Izakhamizi zingabambelela ku-chatbot nge-a iwebhusayithi kahulumeni noma inkundla yezokuxhumana.
“Sikhuluma ngomeluleki kahulumeni wokuqala ukusebenzisa ubuhlakani bokwenziwa,” kuzwelonke nakwamanye amazwe, undunankulu. waqhosha. ®
