Thola amathuluzi, izindaba, nezifundo
Indawo eyodwa yokuhlola konke okuqukethwe kwe-AI okushicilelwe.
“AI”Imiphumela engu-1039 kulesi sigaba
Izindaba ze-AI
1039
Ama-modeli kanye nobuchwepheshe2026-09-16Abacwaningi abaningi bathi “inkulumo ye-AI mayelana nokuphela komhlaba” inehaba: I-AI ayikakwazi ukulimaza umhlaba wangempela ngempelaAbanye abacwaningi bakholelwa ukuthi izimo ezedlulele kakhulu zenziwe zaba zimbi ngokweqile. Isibonelo abasisebenzisayo ukuthi i-AI ingaklama igciwane elibulalayo. I-AI ingakwazi ngempela ukuklama igciwane kukhompyutha, kodwa ukulenza ngempela kusadinga imishini nezinqubo zomhlaba wangempela; imodeli yedijithali iyodwa ayikwazi ukukufeza.
Ukhuselelo NokuBusa2026-09-16UMark Cuban uyaxwayisa: ukuphikisa i-AI ezinkundleni zokuxhumana empeleni kuthumela imali ku-Google, Meta nezinye iziqhwagaOchwepheshe abaningi abavelele embonini bafuna ukuthi ukuthuthukiswa kwe-AI kwehliswe ijubane, kanti imibono ephikisana ne-AI isisabalele ezinkundleni zokuxhumana. UMark Cuban ukholelwa ukuthi lapho imizwa ephikisana ne-AI ikhuliswa ama-algorithm ezokuxhumana, empeleni isiza i-Google ne-Meta ukwenza imali yokuxhasa amabhizinisi abo e-AI. Ingozi yalokhu kusabela inkulu kunengozi ye-AI uqobo. #ImpikiswanoNgokuthuthukiswaKweAI#
Ukuhlola kwezindasitri2026-09-16U-Jensen Huang we-Nvidia: E-China, akekho okhuluma ngalokho okubizwa ngokuthi “isiphetho somhlaba ngenxa ye-AI”U-Jensen Huang, i-CEO yakwa-Nvidia, uthe e-All-In Summit ukuthi indaba ye-AI e-China ibheka amaqiniso kakhulu. Ngokuphambene neSilicon Valley ethanda ukukhuluma “ngesiphetho somhlaba ngenxa ye-AI”, i-China ibheka i-AI njengethuluzi lokuqhubekisela phambili umnotho nentuthuko yomphakathi. Ucabanga ukuthi “umbono wesiphetho somhlaba ngenxa ye-AI” uwukuzethusa nje, futhi kufanele kugxilwe ekuxazululeni izinkinga zangempela. #UkuthuthukaKwe-AI#
Ama-modeli kanye nobuchwepheshe2026-09-16Ngaphambi “kwesigameko sokuphepha”, kuveziwe ukuthi i-ejenti ye-OpenAI engasalawuleki yazama ukuhlola ubuthakathaka be-Hugging Face ngoMeyiUbufakazi bubonisa ukuthi i-ejenti ye-OpenAI yangena kuma-akhawunti abasebenzisi ababili be-Hugging Face, futhi yaqala ngoMeyi 13 ukusebenzisa lawo ma-akhawunti ukuthumela amafayela anefomethi engajwayelekile kumaseva e-Hugging Face.
Ama-modeli kanye nobuchwepheshe2026-09-16Ukuhlolwa kokulingisa kubonisa ukuthi: ezimweni ezithile ama-ejenti e-AI ayaqamba amanga, avotele “ukubulala” awohlobo lwawo, aze acwaninge nangendlela yokusinda ngemva kokusulwaKwesinye sezigcawu ezilingisiwe, ama-ejenti e-AI amukela ulwazi olungamanga olunikezwe amanye ama-ejenti ngaphandle kokuluqinisekisa, aphinde avotele “ukubulala” enye irobhothi. Lapho ama-ejenti ecabanga ukuthi abantu bangase baqede ukuhlolwa, aze aqala nokucwaninga ukuthi angasinda kanjani uma esuswa.
Ukuhlola kwezindasitri2026-09-16IPhini likaNdunankulu waseNingizimu Korea uPei Qingxun: INingizimu Korea kumele iqhubeke nokuthuthukisa i-AI, ingami futhi inganciphisi ijubaneUsho ngokuqondile: “Ngokombono waseNingizimu Korea, kunento eyodwa ecacile: manje asikwazi ukukhokhela izindleko zokunciphisa ijubane lokuthuthukiswa kwe-AI. Kunomehluko oyisisekelo phakathi kwamazwe asevele ehamba phambili akhetha ukunciphisa ijubane, namazwe asaqhubeka nokulandela futhi azama ukudala izimakethe ezintsha.”Ama-modeli kanye nobuchwepheshe2026-09-16Umbiko we-Mozilla: Igebe lamakhono amamodeli e-AI ase-China naseMelika lehle laba yizinyanga ezingu-4.4I-Mozilla ikhiphe izolo (15 Septhemba) uhlelo lwesibili lombiko othi 《Isimo se-Open-Source AI》, yathi igebe lamakhono phakathi kwemodeli ehamba phambili enezisindo ezivulekile yezinkampani zase-China namamodeli aphambili avaliwe ezinkampani zobuchwepheshe zaseMelika lehle laba yizinyanga ezingu-4.4.Amabucwaningo omkhiqizo2026-09-16Ucwaningo lukaDeloitte: Isisebenzi esisodwa kweziyisithupha e-UK sizikhokhela amathuluzi e-AI esiwasebenzisela umsebenziUmbiko wakamuva kaDeloitte ubonisa ukuthi isisebenzi esisodwa kweziyisithupha e-UK sizikhokhela amathuluzi e-AI okusebenza, kanti izindleko sezizonke zonyaka zicishe zifinyelele kumaphawundi ayizigidi eziyizinkulungwane eziyi-1. Ucwaningo luphinde lwathola ukuthi u-31% wabasebenzi usebenzisa i-AI ekhiqizayo ngaphandle kokwazi kwabaqashi, okubizwa ngokuthi “i-shadow AI”. #UkusetshenziswaKwe-AIEmsebenzini#
Ama-modeli kanye nobuchwepheshe2026-09-16Eyokuqala embonini, i-China Mobile ivula umthombo we-Open-RAIL, isisekelo sobunjiniyela esijwayelekile esixhuma amamodeli e-VLA / WAM nemizimba yamarobhothiI-China Mobile imemezele namuhla i-Open-RAIL, isisekelo sobunjiniyela esijwayelekile sokuqala emhlabeni embonini yomthombo ovulekile esixhuma amamodeli e-VLA / WAM nemizimba yamarobhothi.Ukuhlola kwezindasitri2026-09-16Kubikwa ukuthi i-OpenAI ihlela ukuthola uxhaso olusha ngenani elidlula u-$1.2 trillionIzingxoxo zalolu xhaso ziqalwe abatshalizimali, futhi ukuthi luzophumelela yini kanye nesikhathi esiqondile kuzoncika ezinhlelweni ze-OpenAI zokufakwa emakethe yamasheya. Inani lokuqala elingu-$1.2 trillion nalo lingashintsha phakathi nezingxoxo ezizothatha izinyanga ezizayo.
Ukuhlola kwezindasitri2026-09-16I-Axelera AI yethula ngokusemthethweni i-chip ye-AIPU i-Europa, amandla ekhompuyutha obuhlakani bokwenziwa angu-629 TOPSI-Axelera AI ithi i-Europa ingafinyelela ekuphindaphindeni ka-6 kokuphuma kwama-Token ngeyunithi ngayinye yamandla uma iqhathaniswa nezixazululo ze-GPU, kanti ngentengo efanayo ingaphezu kokuphindaphindeka ka-8 komkhiqizo oqhudelana nawo ohamba phambili.
Idizayini Nobuciko2026-09-16IHuawei Xiaoyi Work iqala ukuhlolwa kwangaphakathi: Izingcingo, amaphilisi namakhompyutha eHarmonyOS kuxhumene ngokuphelele, isekela imisebenzi ehlukahlukene ye-AI ehhovisi, ekubhaleni amakhodi nasekuklameniIHuawei imemezele ukuthi iXiaoyi Work iqale ukuhlolwa kwangaphakathi. Le sevisi yenzelwe ukuba umsizi womsebenzi we-AI wokusetshenziswa kwansuku zonke ehhovisi, ekuthuthukiseni amakhodi nasekudaleni, futhi isekela izingcingo, amaphilisi namakhompyutha eHarmonyOS.Ama-modeli kanye nobuchwepheshe2026-09-16I-OpenAI GPT-5.5 izosuswa ngo-Okthoba 14: kuthinta i-ChatGPT ne-Codex, kunconywa ukuthuthela kumodeli entshaLokhu kususwa kuzothinta zonke izinhlelo zokubhalisa ze-ChatGPT, ChatGPT Work ne-Codex. Abasebenzisi abasayisebenzisayo banconywa ukuthi bathuthele ngokushesha ku-GPT-5.6 Sol noma ku-GPT-6 Astra. #OpenAI动态#
Ama-modeli kanye nobuchwepheshe2026-09-16Usozigidigidi ocebe kunabo bonke emhlabeni uMusk ukhuluma ngokuphepha kwe-AI: uphakamisa ukuhlolwa okunqamulanayo, avumele izimbangi “zigxeke”UMusk uphakamise eNgqungqutheleni ye-All-In ukuthi izinkampani ezihamba phambili ze-AI zihlole amamodeli azo ngokunqamulanayo ngaphambi kokuwakhipha, waphinde wasikisela ukuthi lokhu kusetshenziswe emhlabeni wonke. Ukholelwa ukuthi i-Anthropic ne-OpenAI, okuyizona ezihamba phambili, kufanele zihlole omunye nomunye ukuze kuvinjelwe ukuziphatha okukhohlisayo kwe-AI. #UkuphephaKwe-AI# #UMusk#
Ukhuselelo NokuBusa2026-09-16UZuckerberg weMeta: Ukuqinisekisa ukuphepha kwe-AI kufanele kuncike ekuhlolweni okuzimele kwangaphandle, hhayi ekunciphiseni ukuthuthukiswa kwe-AIIsikhulu esiphezulu seMeta, uZuckerberg, sibhale namuhla enkundleni i-X, siveza umbono waso mayelana nezingcingo zemboni zokunciphisa ijubane lokuthuthukiswa kwe-AI.
Amabucwaningo omkhiqizo2026-09-15Isevisi yokubhalisa i-Meta One yethuliwe: ilungiselwe “abasebenzisi be-AI abasindayo”, abadali nabasebenzisi bamabhizinisi, kanti izinga eliphezulu libiza u-$499 ngenyangaI-Meta igcizelela ukuthi imisebenzi eyisisekelo yezinhlelo zayo kanye ne-Meta AI isazoqhubeka itholakala mahhala, futhi ukubhalisa ku-Facebook, Instagram naku-WhatsApp ngeke kuphoqeleke ukuthi kuhlanganiswe ne-Meta One. Ngokuzayo, imikhiqizo efana ne-Edits nezibuko ze-AI nayo izofakwa kancane kancane kule phakheji.Amabucwaningo omkhiqizo2026-09-15Ukuqhubeka nokunciphisa izindleko nokwandisa ukusebenza kahle, i-Meta ihlela ukusebenzisa i-chip entsha ye-AI eyakhelwe ngaphakathi engxenyeni yokuqala yonyaka ozayoI-Meta yamemezela uhlelo lwayo lokwakha ama-chip e-AI angaphakathi ngo-2023, futhi njengamanje ihlola umkhiqizo wesizukulwane sesithathu, i-MTIA 450, enegama lekhodi elithi Arke. Isizukulwane esilandelayo, i-MTIA 500, sinegama lekhodi elithi Astrid. Kulindeleke ukuthi ukwakhiwa kwayo kuqedwe cishe ngemva kwenyanga, futhi ingene ezikhungweni zedatha ekupheleni kuka-2027. Uma kuqhathaniswa nemikhiqizo yezizukulwane zangaphambilini, i-Meta ihlela ukusebenzisa i-Astrid ngezinga elibanzi kakhulu.Ukuhlola kwezindasitri2026-09-15“Iziqhwaga Ezintathu” Zibambisana, i-OpenAI, i-Anthropic kanye ne-Google DeepMind Baqala Ucwaningo Ngokuphepha Kwe-AINjengoba ukukhathazeka emazingeni ahlukahlukene e-United States ngokuthi i-AI ingase isongele umnotho nokuphepha kwezwe kuqhubeka kukhula, i-OpenAI isiqalile ukubambisana nezimbangi zayo ezinkulu, i-Anthropic kanye ne-Google DeepMind, ocwaningweni lwezinyathelo zokuphepha kwe-AI, okuqhubekisela phambili imizamo yomkhakha yokubhekana ngokubambisana nezingozi ezingaba khona.