Kwenzekeni
I-South China Morning Post ibika ukuthi umncintiswano we-US-China AI ugxile kakhulu ekuzithuthukiseni okuphindaphindayo: amasistimu asebenzisa i-AI ukusiza ucwaningo, ukuqeqesha nokuhlola abazolandela abanekhono. Izinkampani zase-US zibonakala zihamba phambili ocwaningweni olusizwa yi-AI, kanti amalebhu aseShayina asakha izindlela zokuziqeqesha kanye nezindlela “zokuziphendukela kwemvelo” ezihlobene. Umbiko uthi alukho uhlangothi olubonise iluphu yokuthuthukisa evulekile, ezimele ngokuphelele.
I-South China Morning Post ibika ukuthi izinkampani ezihamba phambili zase-US kanye ne-Chinese AI zisebenzisa amamodeli athuthukile ukuze abhale amakhodi, adizayine ukuhlola futhi athuthukise amasu okuqeqesha amamodeli aqinile. Ichaza inhloso yesikhathi eside njengokuzithuthukisa okuphindaphindayo, lapho amasistimu e-AI anikela ngokuzimele ekuqeqesheni nasekuthuthukiseni amasistimu alandelayo. Umbiko uthi i-OpenAI isanda kuchaza umsebenzi kumcwaningi we-AI ozenzakalelayo, ngenkathi futhi ivuma ukungaqiniseki mayelana nendlela yokufinyelela ekuzithuthukiseni okuhambisana ngokugcwele okuphindelelayo ngokuphepha.
Ngokombiko, izinkampani zaseMelika njengamanje zivela ezinyangeni ezimbalwa ngaphambili ekusebenziseni i-AI ukwenza ucwaningo lwe-AI ngokuzimela. Icaphuna umsebenzi obikiwe we-Anthropic wokuba nokuhlolwa kokuklama kwamamodeli e-Claude, ukubhala ikhodi nokuthuthukisa amasu angemuva kokuqeqeshwa, kanye nencazelo ye-Google DeepMind ye-Gemini 3.8 Flash njengoba ifanele ama-ejenti we-software asebenza kude nokuzimela. Umbiko uphinde uthi i-OpenAI ibike umcwaningi osebenza ngokuzenzakalelayo okwazi ukwenza izivivinyo ezithile ezingathatha izinsuku zomcwaningi onekhono. Lezi zimangalo zidalulwe yiSouth China Morning Post kanye nezinkampani noma abacwaningi ebacaphunayo; aziqinisekiswanga ngokuzimela lapha.
Umbiko uthi amalebhu aseShayina aphishekela izinhloso ezihlobene ngaphansi kwamagama anjengokuthi "ukuziphendukela kwemvelo" kanye "nokuziqeqesha ngokugcwele." I-Z.ai umsunguli u-Tang Jie kubikwa ukuthi uchaze umgwaqo we-GLM-6.0 lapho i-AI izophatha ipayipi lokuqeqesha kusukela ekuqeqeshweni kwangaphambili ngokusebenzisa ukuqeqeshwa ngemva kokuqeqeshwa. Umbiko uphinde udingide umsebenzi ohlotshaniswa ne-MiniMax kanye ne-DeepSeek, kuyilapho uphawula ukugxeka kochwepheshe kokuthi ama-ejenti azimele kanye nekhono lokuziqeqesha akukafiki ekuthuthukisweni kwangempela okuphindaphindayo. Indaba eyinhloko engaxazululiwe ukuthi ingabe isistimu inganquma uma ushintsho empeleni luyintuthuko ngaphandle kokuthembela esigungwini somuntu esihlukile.
Imininingwane yomthombo: scmp.com ↗
Kungani kubalulekile
Uma amasistimu e-AI angathuthukisa ngokuthembekile izindlela ezisetshenziswayo ukwakha abalandelayo, inqubekelaphambili ingase isheshise futhi yenze ukugada kube nzima nakakhulu. Umbiko uchaza izinzuzo ezingaba khona ekuthuthukisweni kwesofthiwe, idizayini yokuhlola nokusebenza kahle kokuqeqeshwa, kodwa futhi uxwayisa ngokuthi amamodeli anamaphutha angase adlulise amaphutha noma ukungaqondi kahle ezizukulwaneni ezizayo. Ukubaluleka okungokoqobo kusalokhu kungaqinisekile ngoba izinhlelo zamanje zisathembele ekwahluleleni komuntu, ukubala ingqalasizinda kanye nokuhlola kwangaphandle.
Ukuzithuthukisa okuphindaphindayo kungadala iluphu yempendulo lapho i-AI isiza khona ukukhiqiza amamodeli anekhono, okungenzeka kufinyeze imijikelezo yentuthuko futhi kwandise ukubaluleka kwesu lokubala, ingqalasizinda yocwaningo kanye nokuhlola. Lo mbiko uthi abahlaziyi baseMelika babona inzuzo ekufinyeleleni kwe-American ukubala, kuyilapho abacwaningi baseShayina bechazwa njengabasebenza kahle ekukhipheni ukusebenza ku-hardware enomkhawulo.
Ukukhathazeka ngokuphepha kusemqoka ngoba ukuqeqesha amasistimu esikhathi esizayo anamanduleli anamaphutha kungase kuhlanganiswe amaphutha noma ukungahambi kahle. Umbiko ucaphuna izexwayiso ezivela kubacwaningi kuwo womabili la mazwe mayelana nokusheshisa, ukuphunyuka kwebhokisi lesihlabathi kanye nemiphumela engaba khona yokuphepha kwezwe. Kodwa-ke, iphinda ibike ukuthi ukuthuthuka okuyisisekelo kwezakhiwo kanye nezahlulelo mayelana nokuthi izinguquko ziyazuzisa yini zincike elwazini lomuntu. Isikali, isivinini kanye nokuthembeka kwanoma yikuphi ukuthuthukiswa okuphindaphindiwe okuzayo kuhlala kungaziwa.
I-Interactive Mechanism: Indlela Esebenza Ngayo Ngempela
Hlola ubuchwepheshe obuyisisekelo ngemuva kwalokhu kuthuthukiswa ngokuhlanganyela.
crm_get_transaction(id='4092').Which component of an AI application is the machine-learning model itself?
Ongakubuka ngokulandelayo
Buka imiboniso engaqiniseka ngokuzimela yokuthi isistimu ye-AI ingaphakamisa, isebenzise futhi yahlulele ukuthuthukiswa ngaphandle kwesiqinisekisi esichazwe ngumuntu esihlukile. Bheka futhi ukuthi izinkampani ziyashicilela yini ukuhlolwa kokuphepha, imikhawulo yokusebenzisa kanye nobufakazi bokuthi amasistimu wokuzithuthukisa awakhulisi amaphutha noma adlule izivikelo. Ayikho imigomo yokufinyelela, intengo noma ukutholakala okuvamile okubhalwe phansi ngoba umbiko uphathelene nocwaningo nentuthuko kunokwethulwa komkhiqizo womphakathi.
Ubufakazi obubaluleke kakhulu bungaba uhlelo oluphindaphindekayo olukhomba ngokuzimela izinguquko eziwusizo, lusebenzise, luhlole imiphumela futhi luthuthukise inqubo yalo yokuhlola. I-South China Morning Post ibika ukuthi ubufakazi obusesidlangalaleni baleli khono elivulekile buhlala buncane, ngakho izimangalo "zokuziphendukela kwemvelo" akufanele zithathwe njengobufakazi bokuzithuthukisa okuphindaphindiwe okuzenzakalelayo.
Umbiko awukuqophi ukukhishwa okusesidlangalaleni, ukufinyelela komthengi, amanani noma ishejuli yokuphakela yamasistimu okuxoxwe ngawo. Futhi ayiqinisekisi ngokuzimela izimangalo zokusebenza kwezinkampani. Ukubika okuzayo kufanele kuveze ukuthi yimiphi imisebenzi lezi zinhlelo ezingaqeda ngaphandle kokungenelela komuntu, yiziphi izivikelo ezibacindezelayo, ukuthi imiphumela ikalwa kanjani nokuthi ukwehluleka kutholwa yini ngaphambi kokuba izinguquko zifinyelele ezinhlelweni zokukhiqiza.