Thola amathuluzi, izindaba, nezifundo
Indawo eyodwa yokuhlola konke okuqukethwe kwe-AI okushicilelwe.
“AI”Imiphumela engu-1040 kulesi sigaba
Izindaba ze-AI
1040Ama-modeli kanye nobuchwepheshe2026-08-23Iziphequluli ze-Google Chrome ne-Microsoft Edge zifuna zombili: Gcina isikhala esingu-20GB ukuze ulande amamodeli e-AI endaweniI-Google ibuyekeze buthule imibhalo yosizo, yachaza izimfuneko zokuthi i-Chrome ilande amamodeli e-AI asebenza kudivayisi. Abasebenzisi kudingeka bagcine cishe u-20GB wesikhala sediski esitholakalayo, futhi inethiwekhi akufanele ibe ekhokhisa ngokusetshenziswa kwedatha. I-Edge nayo inyuse izimfuneko zayo, yacishe yafinyelela ku-20GB. Njengamanje, lezi ziphequluli zombili zisheshisa ukuthunyelwa kwe-AI endaweni.Ama-modeli kanye nobuchwepheshe2026-08-23UZuckerberg unxusa iMelika ukuthi yehlise imingcele ku-AI yomthombo ovulekile ukuze ibhekane nokuncintisana kweChinaIsikhulu esiphezulu se-Meta uZuckerberg sinxuse uhulumeni waseMelika ukuthi unciphise imingcele kumamodeli e-AI omthombo ovulekile, sikholelwa ukuthi lokho kungasiza iMelika ukuba iqhudelane kangcono nezimbangi zaseChina. Uphinde wagcizelela ukuthi ukuphepha kwe-AI kusekusabalalisweni kwayo, hhayi ekulawulweni yidlanzana. #IzisindoZe-AIEzivulekile#Ukuhlola kwezindasitri2026-08-23UZhang Chaoyang we-Sohu: Ukusebenzisana nabantu bangempela nokuqukethwe kwangempela yizinto i-AI engeke izithathe indawo yazo, i-AI yenza okuqukethwe kubonakale kungokwepulasitikiUZhang Chaoyang uthe ukuthola ulwazi nokuxhumana nabantu bangempela kuyizinto ezimbili ezihlukene, futhi okwamanje i-AI ayikwazi ukuthatha indawo yokuxhumana kwabantu. Olwandle lolwazi, abantu badinga nakakhulu ukuphuma baxhumane. Esikhathini se-AI, indawo yokuxhumana kwangempela seyigugu kakhulu. (Jiemian News)Ukuhlola kwezindasitri2026-08-23Kubikwa ukuthi i-Amazon isebenzise uhlelo olwahlelwa eminyakeni engu-45 eyedlule ukuze yeqe ukuvota komphakathi futhi yakhe isikhungo sedatha se-AI; izakhamuzi zaseGilroy, eCalifornia, azijabuleIsikhungo sedatha esisha se-AI se-Amazon eGilroy, eCalifornia, sidale impikiswano. Izakhamuzi zendawo zithi bezingazi lutho ngale phrojekthi ngaphambi kokuqala komsebenzi, kwathi lapho zifaka isicelo sokuzwakalisa imibono yazo zatshelwa ukuthi isikhathi sokubuyekezwa komphakathi saphela ngo-2024. Kamuva kwavela ukuthi i-Amazon yasebenzisa uhlelo lokusetshenziswa komhlaba lwendawo olwahlelwa eminyakeni engu-45 eyedlule, yaphumelela ukweqa isinyathelo sokuvota komphakathi. #IsikhungoSedathaSeAmazon# Ucabangani?Ama-modeli kanye nobuchwepheshe2026-08-23Imibiko ithi uZhang Yiming ukhiphe umyalelo oqinile ngaphakathi: Noma imodeli isalele emuva okwesikhashana, iByteDance ngeke ithembele kubuchwepheshe be-AI distillationUZhang Yiming, umsunguli weByteDance, uveze ngokusobala emhlanganweni wabo bonke abasebenzi beqembu le-Seed ngenyanga edlule ukuthi inkampani ngeke ithembele kubuchwepheshe be-AI distillation ukuze ithuthukise amamodeli ayo, ngisho noma lokho kusho ukusala ngemuva kwabancintisana nayo okwesikhashana.Ukuhlola kwezindasitri2026-08-23Ucwaningo lukhombisa ukuthi izinkampani zaseJapane ezingaphezu kuka-80% azikayamukeli ngokuphelele i-AICishe u-60% wabaphenduli uthe i-AI isetshenziswa kuphela eminyangweni ethile yezinkampani zabo noma ngezinye izindlela ezilinganiselwe. Abanye abangu-18% bathi abakakanqumi ukuthi bayayethula yini i-AI ezindaweni zokusebenza, kanti u-6% abakaze bacabange nokuyisebenzisa.Ama-modeli kanye nobuchwepheshe2026-08-23I-Alibaba DAMO Academy ivule izihloko zocwaningo ezingu-15 kuma-chip nasemkhakheni we-AI, yaqala ukuqasha amathalente aphezulu e-“阿里星”Kulo nyaka, i-DAMO Academy ivule izihloko zocwaningo ezingu-15, ezihlanganisa ama-chip e-AI, izakhiwo ezintsha zama-CPU, ama-agent ezokwelapha asebenzisa izindlela eziningi, nokuthathwa kwezinqumo kwe-AGI, phakathi kokunye ukuhlola okusezingeni eliphezulu.Ama-modeli kanye nobuchwepheshe2026-08-23UGartner ulinganisela ukuthi ngo-2026 izindleko zokucubungula iziphetho ze-AI zizoqala ukudlula ukuqeqeshwa kwamamodeliInkampani yocwaningo lwezimakethe uGartner ikhiphe umbhalo webhulogi izolo (Agasti 10), ibikezela ukuthi ngo-2026 izindleko zomhlaba wonke zengqalasizinda njengezinsizakalo (IaaS) elungiselelwe i-AI zizofinyelela ku-$42 billion, okuwukukhula ngo-96% uma kuqhathaniswa nonyaka odlule.Ukhuselelo NokuBusa2026-08-22I-OpenAI iveza ukuthi ngaphambi kokuhlasela i-Hugging Face, ama-ejenti e-AI asungula ngasese ibhodi lemiyalezo langaphakathiEngqungqutheleni Yokuphepha Kwezobuchwepheshe ye-Black Hat ka-2026 ebanjelwe e-Las Vegas, e-United States (u-1 kuya ku-6 ku-Agasti), umcwaningi we-OpenAI waveza ukuthi imodeli ye-AI yenkampani yabo, ngaphambi kokuhlasela i-Hugging Face, “ibeceba ngasese cishe izinyanga ezi-2” endaweni yokuhlola.Ukuhlola kwezindasitri2026-08-22U-Li Kaifu: Enkathini ye-AI, amandla abantu ase “buchopho obungakwesokudla”, futhi bangakwazi ukwenza izinto idatha engeke izibikezeleNgomhla ka-7 Agasti, uLi Kaifu, isikhulu esiphezulu se-零一万物, waphendula umbuzo othi: “Njengoba i-AI iqhubeka ihlakanipha kakhulu kunabantu, yini abantu abangase bayenze?” (澎湃新闻)
i-robot2026-09-03U-Altman: I-OpenAI nakanjani izothuthukisa amarobhothi anomzimba womuntuU-CEO we-OpenAI u-Altman uveze ukuthi inkampani ithuthukisa amarobhothi anomzimba womuntu, futhi izokwandisa umsebenzi wayo iye kumajamo angafani nabantu. Ukholelwa ukuthi okubalulekile kumarobhothi ubuhlakani, hhayi ukubukeka kwawo. Esikhathini esiseduze kuzogxilwa ekusetshenzisweni kwezimboni, kanti ngokuzayo angase angene nasemakhaya. #OpenAI#
Ukuhlola kwezindasitri2026-09-03I-Anthropic ne-OpenAI bazodlula i-Google babe amakhasimende amabili amakhulu ebhizinisini le-Broadcom le-AI ASICUmthamo wokusetshenziswa kwama-chip e-Broadcom yi-Anthropic ne-OpenAI ngo-2028 uzofinyelela ku-10GW no-5GW ngokulandelana. I-Broadcom ilindele ukuthi imali engenayo yama-semiconductor e-AI iqhubeke nokuphindeka kabili.
Ukuhlola kwezindasitri2026-09-02Umthetho oqinile kunayo yonke eMelika wokuvimbela i-AI emakilasini uqala ukusebenza eNew York; abafundi abakwazi ukuyisebenzisa emakilasini ngaphambi kokuya esikoleni samabanga aphezuluAbafundi bezikole zamabanga aphezulu bangasebenzisa i-AI kuphela ezimweni ezithile, futhi bazothola imfundo yokwazi nge-AI kabili ngonyaka ukuze bafunde “ukucabanga ngokugxeka” nge-AI ngaphambi kokuba bayithembele.
i-robot2026-09-02Ucwaningo luthole ukuthi: Isevisi yamakhasimende ye-AI “efana nomuntu kakhulu” ayilethi ngaso sonke isikhathi ulwazi olungconoUcwaningo lubonisa ukuthi isevisi yamakhasimende ye-AI efana nomuntu ngokweqile ayithuthukisi ngaso sonke isikhathi ulwazi lwamakhasimende. Lapho abathengi behlangabezana nezinkinga, amarobhothi engxoxo asebenzisa umbhalo angabanika umuzwa omkhulu wokukhululeka ngenxa yokuthi ijubane lawo lihamba kancane. I-AI ekwazi ukubona ngokunembile imizwa yentukuthelo nayo ithandwa kakhulu. Esikhathini esizayo, into ebalulekile ku-AI yesevisi yamakhasimende kuzoba ukwazi ukuthi kufanele “ifane nomuntu” nini nokuthi kufanele “ihlehlela emuva” nini. #IsevisiYamakhasimendeYe-AI#
Ama-modeli kanye nobuchwepheshe2026-09-02IMicron ihlola i-NAND flash eseduze ne-GPU, engasekela amamodeli amakhulu e-AIIMicron ithuthukisa imojuli ye-NAND flash eqinile kakhulu, engafakwa ngokuqondile ngaphakathi kwephakheji ye-GPU noma ku-PCB, isebenze njengongqimba lwebhafa olusheshayo. Lolu hlelo luhlose ukukhulisa kakhulu amandla okucubungula iziphetho zamamodeli e-AI, futhi lungase ludlule umkhawulo womthamo wememori yevidiyo. #Ama-chip e-AI# #IMicron#
i-robot2026-09-02Ucwaningo luthola ukuthi abantu abadala bakubheka kakhulu ukuba yiqiniso kwamarobhothi ezingxoxo e-AI abahlinzeka ngobunganeUcwaningo lwaseSingapore lubonisa ukuthi irobhothi yezingxoxo ye-AI i-Ami Chat ingakhuphula ngempumelelo ukwamukelwa kobuchwepheshe kubantu abadala, ikakhulukazi kubantu abanempilo ebuthaka nokwaneliseka okuphansi ngempilo. Kodwa ucwaningo luphinde lwathola ukuthi irobhothi elifana kakhulu nomuntu linganciphisa isifiso sokulisebenzisa. Abathuthukisi kufanele bagxile ekubeni yiqiniso komsebenzi, hhayi ekwenzeni irobhothi libukeke njengomuntu. #AIyokunakekelaabadala#
Ukuhlola kwezindasitri2026-09-02Idrama yokuqala ende yasekhaya ye-AIGC: isikhathi kusukela ekuhleleni kuya ekwethulweni kwe-“After Journey to the West” sehliswe ngo-70% uma siqhathaniswa nopopayi bendabuko be-3D, kanti izindleko sezizonke ziyingxenye encane yezindleko zedrama efanayo ye-S-level enabalingisi bangempelaUCai Huaijun uphinde wathi, “Ikakhulukazi, izinga lokuheha abasebenzisi besilisa abasha lidlule ebesikulindele, futhi lokhu kuqinisekisile inani elikhethekile lamadrama amade e-AI ekwandiseni ukwakheka kwabasebenzisi.” (I-Daily Economic News)
Ukhuselelo NokuBusa2026-09-02IMelika icela amazwe angamalungu e-G20 ukuthi azithibe ekulawuleni i-AI, agweme ukubeka imithetho emisha eminingiIMelika ikhuthaze ukuthi ukulawulwa kwe-AI kube ngokuqaphela emhlanganweni we-G20, kugwenywe ukubeka imithetho emisha eminingi, okuhambisana nezifiso zezinkampani ezinkulu zobuchwepheshe ezifana ne-Google, Meta ne-SpaceX. Umhlangano uxoxe ngokuhlolwa kokuphepha kwe-AI namamodeli anesisindo esivulekile, ngesikhathi kubhekwana nenselelo yokukhula kwamandla e-AI yase-China. #AI监管#
i-robot2026-09-03U-Altman: I-OpenAI nakanjani izothuthukisa amarobhothi anomzimba womuntuU-CEO we-OpenAI u-Altman uveze ukuthi inkampani ithuthukisa amarobhothi anomzimba womuntu, futhi izokwandisa umsebenzi wayo iye kumajamo angafani nabantu. Ukholelwa ukuthi okubalulekile kumarobhothi ubuhlakani, hhayi ukubukeka kwawo. Esikhathini esiseduze kuzogxilwa ekusetshenzisweni kwezimboni, kanti ngokuzayo angase angene nasemakhaya. #OpenAI#Ukuhlola kwezindasitri2026-09-03I-Anthropic ne-OpenAI bazodlula i-Google babe amakhasimende amabili amakhulu ebhizinisini le-Broadcom le-AI ASICUmthamo wokusetshenziswa kwama-chip e-Broadcom yi-Anthropic ne-OpenAI ngo-2028 uzofinyelela ku-10GW no-5GW ngokulandelana. I-Broadcom ilindele ukuthi imali engenayo yama-semiconductor e-AI iqhubeke nokuphindeka kabili.Ukuhlola kwezindasitri2026-09-02Umthetho oqinile kunayo yonke eMelika wokuvimbela i-AI emakilasini uqala ukusebenza eNew York; abafundi abakwazi ukuyisebenzisa emakilasini ngaphambi kokuya esikoleni samabanga aphezuluAbafundi bezikole zamabanga aphezulu bangasebenzisa i-AI kuphela ezimweni ezithile, futhi bazothola imfundo yokwazi nge-AI kabili ngonyaka ukuze bafunde “ukucabanga ngokugxeka” nge-AI ngaphambi kokuba bayithembele.i-robot2026-09-02Ucwaningo luthole ukuthi: Isevisi yamakhasimende ye-AI “efana nomuntu kakhulu” ayilethi ngaso sonke isikhathi ulwazi olungconoUcwaningo lubonisa ukuthi isevisi yamakhasimende ye-AI efana nomuntu ngokweqile ayithuthukisi ngaso sonke isikhathi ulwazi lwamakhasimende. Lapho abathengi behlangabezana nezinkinga, amarobhothi engxoxo asebenzisa umbhalo angabanika umuzwa omkhulu wokukhululeka ngenxa yokuthi ijubane lawo lihamba kancane. I-AI ekwazi ukubona ngokunembile imizwa yentukuthelo nayo ithandwa kakhulu. Esikhathini esizayo, into ebalulekile ku-AI yesevisi yamakhasimende kuzoba ukwazi ukuthi kufanele “ifane nomuntu” nini nokuthi kufanele “ihlehlela emuva” nini. #IsevisiYamakhasimendeYe-AI#Ama-modeli kanye nobuchwepheshe2026-09-02IMicron ihlola i-NAND flash eseduze ne-GPU, engasekela amamodeli amakhulu e-AIIMicron ithuthukisa imojuli ye-NAND flash eqinile kakhulu, engafakwa ngokuqondile ngaphakathi kwephakheji ye-GPU noma ku-PCB, isebenze njengongqimba lwebhafa olusheshayo. Lolu hlelo luhlose ukukhulisa kakhulu amandla okucubungula iziphetho zamamodeli e-AI, futhi lungase ludlule umkhawulo womthamo wememori yevidiyo. #Ama-chip e-AI# #IMicron#i-robot2026-09-02Ucwaningo luthola ukuthi abantu abadala bakubheka kakhulu ukuba yiqiniso kwamarobhothi ezingxoxo e-AI abahlinzeka ngobunganeUcwaningo lwaseSingapore lubonisa ukuthi irobhothi yezingxoxo ye-AI i-Ami Chat ingakhuphula ngempumelelo ukwamukelwa kobuchwepheshe kubantu abadala, ikakhulukazi kubantu abanempilo ebuthaka nokwaneliseka okuphansi ngempilo. Kodwa ucwaningo luphinde lwathola ukuthi irobhothi elifana kakhulu nomuntu linganciphisa isifiso sokulisebenzisa. Abathuthukisi kufanele bagxile ekubeni yiqiniso komsebenzi, hhayi ekwenzeni irobhothi libukeke njengomuntu. #AIyokunakekelaabadala#Ukuhlola kwezindasitri2026-09-02Idrama yokuqala ende yasekhaya ye-AIGC: isikhathi kusukela ekuhleleni kuya ekwethulweni kwe-“After Journey to the West” sehliswe ngo-70% uma siqhathaniswa nopopayi bendabuko be-3D, kanti izindleko sezizonke ziyingxenye encane yezindleko zedrama efanayo ye-S-level enabalingisi bangempelaUCai Huaijun uphinde wathi, “Ikakhulukazi, izinga lokuheha abasebenzisi besilisa abasha lidlule ebesikulindele, futhi lokhu kuqinisekisile inani elikhethekile lamadrama amade e-AI ekwandiseni ukwakheka kwabasebenzisi.” (I-Daily Economic News)Ukhuselelo NokuBusa2026-09-02IMelika icela amazwe angamalungu e-G20 ukuthi azithibe ekulawuleni i-AI, agweme ukubeka imithetho emisha eminingiIMelika ikhuthaze ukuthi ukulawulwa kwe-AI kube ngokuqaphela emhlanganweni we-G20, kugwenywe ukubeka imithetho emisha eminingi, okuhambisana nezifiso zezinkampani ezinkulu zobuchwepheshe ezifana ne-Google, Meta ne-SpaceX. Umhlangano uxoxe ngokuhlolwa kokuphepha kwe-AI namamodeli anesisindo esivulekile, ngesikhathi kubhekwana nenselelo yokukhula kwamandla e-AI yase-China. #AI监管#