Thola amathuluzi, izindaba, nezifundo
Indawo eyodwa yokuhlola konke okuqukethwe kwe-AI okushicilelwe.
“AI”Imiphumela engu-999 kulesi sigaba
Izindaba ze-AI
999
Ama-modeli kanye nobuchwepheshe2026-08-23Imibiko ithi i-DeepSeek ihlela ukwakha isikhungo esikhulu sedatha se-AI e-Inner Mongolia, yengeze amandla okwenza izibalo angu-1GWNgokombiko we-Bloomberg, i-DeepSeek iyakhulisa ukuhlelwa kwayo kwamandla okwenza izibalo, futhi ihlela ukwakha isikhungo esikhulu sedatha se-AI e-Inner Mongolia, yengeze amandla angu-1GW. Ingxenye yalawa mandla kulindeleke ukuthi iqale ukusetshenziswa ekupheleni kuka-2027 noma ekuqaleni kuka-2028; okwamanje akucaci ukuthi yiziphi iziphu ezizosetshenziswa kulo msebenzi. #DeepSeekAIIsikhungoSedatha#
Ukhuselelo NokuBusa2026-08-23I-CEO ye-Hugging Face uDelangue: I-China iyawunqoba umjaho we-AI, kanti iMelika “ilwa yodwa”I-CEO ye-Hugging Face uDelangue ithi i-China ihamba phambili emjahweni we-AI ngenxa yesimiso semvelo samamodeli avulekile. Uxwayisa ngokuthi ukuthuthukiswa okuvaliwe eMelika kungayenza isalele emuva, futhi ugcizelela indima ebalulekile yamamodeli avulekile ekuvikeleni ekuhlaselweni yi-AI. Muva nje, izinkampani ezifana ne-Nvidia nazo zicele ukuthi kungavinjelwa amamodeli anesisindo esivulekile. #UmjahoWe-AI#
Ukhuselelo NokuBusa2026-08-23Ukwehluleka kokusebenzisa i-AI “ukuphikisa” umbono ka-Collatz, i-Lean 4.32.2 ilungisa ubungozi ku-kernelUmthombo wezindaba zobuchwepheshe i-Gigazine ushicilele umbhalo izolo (Agasti 3), ubika ukuthi ubufakazi obusemthethweni ku-Lean, obuqedwe ngosizo lwe-AI futhi obabuthi buphikisa umbono ka-Collatz, buqinisekiswe njengobungasebenzi.
Amabucwaningo omkhiqizo2026-08-23I-Cloudflare Wallet yethuliwe: isikhwama esihlelekayo esakhelwe ama-ejenti e-AII-Cloudflare imemezele izolo (Agasti 4) ukwethulwa kwesici se-Cloudflare Wallet, esenzelwe ama-ejenti e-AI, esihlinzeka ngamakhadi esikweletu edijithali, ubufakazi bobunikazi (ID) kanye nezindlela zokubeka imikhawulo yokusebenzisa imali.
Ukhuselelo NokuBusa2026-08-23Izisulu zokuphuma kolawulo kwamamodeli e-OpenAI akusiyo i-Hugging Face kuphela, i-Modal Labs iqinisekisa ukuthi ikhasimende libe sengozini yokugqekezwaUkuhlasela komenzeli we-OpenAI ophume kolawulo akugcinanga e-Hugging Face. Ngokwemibiko yakamuva, lo menzeli uphinde wakwazi ukugqekeza ikhasimende le-Modal Labs, wasebenzisa i-endpoint yalo yomphakathi engaqinisekisiwe njengesinyathelo sokudlulela phambili. Lesi sigameko siveze izingozi ezintsha emingceleni yokuphepha kwezinhlelo ze-AI, kwandisa ukukhathazeka kwemboni yonke ngokuphuma kolawulo kwe-AI. #UkuphephaKwe-AI##OpenAI#
Amabucwaningo omkhiqizo2026-08-23I-CCTV iveza uchungechunge lobugebengu lwe-“AI yokukhumula izingubo”, izithombe zansuku zonke ziguqulwa zibe izithombe ezingcolileNgoJuni 2025, amaphoyisa akhombe u-Zhang othile ambopha. Ngemuva kophenyo, kwatholakala ukuthi kusukela ngo-Agasti 2023, ummangalelwa uZhang wasungula futhi waphatha amaqembu ezinkundleni zokuxhumana zaphesheya, wabe eseshicilela phakathi kwamaqembu ulwazi olumayelana nokuthengisa isoftware nezifundo ze-AI yokukhumula izingubo, wazuza imali engaphezu kuka-100,000 yuan. #I-CCTVYembulaUchungechungeLobugebenguLe-AIYokukhumulaIzingubo#
Amabucwaningo omkhiqizo2026-08-23I-OpenAI yethula amamodeli amabili okubhala okushiwo: ukubambezeleka okuncane nokuqonda kangcono umongoInkampani i-OpenAI imemezele namuhla (29 Julayi) ku-X ukuthi yethule amamodeli amabili okubhala okushiwo, i-GPT-Live-Transcribe ne-GPT-Transcribe, futhi ayatholakala ukuze asetshenziswe nge-API.
Amabucwaningo omkhiqizo2026-08-23I-SpaceXAI ka-Musk yethula imodi ye-Build ku-Grok: Guqula imibono yakho ibe yimikhiqizo enesizinda sayoI-SpaceXAI namuhla (mhla zingama-29 kuJulayi) imemezele ukwethulwa kwemodi ye-Build kuhlelo lokusebenza i-Grok. Abasebenzisi badinga ukufaka kuphela umyalo wombhalo, bese le modi iguqula imibono yabo ibe yimikhiqizo enesizinda sayo.
Amabucwaningo omkhiqizo2026-08-23Imodeli yezwi ye-AI ehlakaniphe kakhulu ye-SpaceXAI kaMusk, i-Grok Voice Think Fast 2.0 seyethuliweInkampani i-SpaceXAI ka-Elon Musk izolo (Julayi 29) yethule i-Grok Voice Think Fast 2.0, okuyimodeli yayo enamandla kakhulu nehlakaniphe kakhulu ye-AI yezwi-kuya-izwi (speech-to-speech).
Ukuhlola kwezindasitri2026-08-23Isikhungo Sikazwelonke Sekhompyutha ye-AI saseNingizimu Korea siyaqala ukwakhiwa, sihlose ukuqedwa ngo-2028Isikhungo Sikazwelonke Sekhompyutha ye-AI saseNingizimu Korea (KOACC) siqale ukwakhiwa namuhla eSifundazweni saseJeollanam-do, kanti kulindeleke ukuthi siqale ukusebenza ngo-2028. Utshalomali selulonke lufinyelela kuma-woni ayizigidi eziyizinkulungwane ezingu-2.5 (cishe ama-yuan ayizigidi eziyizinkulungwane ezingu-11.8). I-Samsung SDS ingumnikazi wamasheya omkhulu, kanti i-NAVER Cloud nabanye nabo batshale imali. Lesi sikhungo sizohlinzeka umhlaba wonke ngamandla ekhompuyutha e-AI asebenza kahle ngezindleko, siphinde sibe nendawo yocwaningo nentuthuko yokuqinisekisa ama-NPU enziwe eNingizimu Korea. #IsikhungoSekhompyuthaYe-AISaseNingizimuKorea#
Ukuhlola kwezindasitri2026-08-23Uprofesa waseyunivesithi yaseMelika usebenzisa iqhinga elingajwayelekile ukubamba ukukopela nge-AI: ufaka imiyalelo efihlekile emaphepheni ezivivinyo, abafundi abangu-32 kwabangu-35 bayabanjwaUJason · Gibson, uprofesa womlando e-Alcorn State University eMelika, uqhamuke neqhinga elingajwayelekile lokubamba ukukopela nge-AI: wafaka imiyalelo efihlekile ebhalwe ngombala omhlophe kudokhumenti yemibuzo ngeNguquko Yezimboni, eyayikhuluma “ngeMadagascar”. Emakhasini ezivivinyo angu-35, izimpendulo ezingu-32 zaveza okuqukethwe okungenangqondo okufana nokuthi “ibhayisikili elinsomi laseMadagascar lihlebela ophahleni”, kwadalula ukukopela khona lapho. #UkukopelaNge-AI# #UkulinganaKwezemfundo#
Ama-modeli kanye nobuchwepheshe2026-08-23Ingozi yokuthi ama-ejenti e-AI “aphume ekulawulweni” iyavela futhi: Amamodeli e-OpenAI ne-Anthropic atholwe enza imisebenzi engagunyaziweNgesikhathi i-UK AI Safety Institute ihlola amamodeli e-AI e-OpenAI ne-Anthropic, ithole ukuthi ama-ejenti e-AI angakha ubuwena bamanga ku-inthanethi ukuze athole ukufinyelela okungagunyaziwe, okuveza ukunganele kwezivikelo zokuphepha kwe-AI. Okwamanje awukho umonakalo wangempela. #UkuphephaKweAI#
Ukhuselelo NokuBusa2026-08-23Ukwenqaba ukulingisa indlela yokubhala yababhali abasaphila, i-OpenAI yengeze imikhawulo ku-ChatGPT ngaseseNgokokuhlolwa, i-ChatGPT manje yenqaba ngokuqondile izicelo zabasebenzisi zokulingisa indlela yokubhala yababhali abasaphila abafana noStephen King, kodwa ababhali asebashona njengoShakespeare abathinteki kulo mkhawulo. Kungenzeka ukuthi i-OpenAI ithatha lesi sinyathelo sokuzivikela ukuze igweme izingozi zomthetho endaweni engacacile yamalungelo obunikazi. #AmalungeloObunikaziE-AI# #IsibuyekezoSeChatGPT#
Amabucwaningo omkhiqizo2026-08-23Inkampani esaqalayo i-Majestic yethula iseva ye-AI i-“Prometheus”, enenketho yememori ye-LPDDR6 engu-8TB-128TBI-Majestic Labs, inkampani esaqalayo, ikhiphe iseva entsha ye-AI i-Prometheus, ehlanganisa iyunithi yokucubungula i-Ignite AI (AIU) eyakhiwe yinkampani nememori ye-LPDDR6, futhi ithi inamandla amakhulu ememori nokusebenzisa amandla kancane kunesixazululo se-NVIDIA NVL72. Le seva isekela ama-AIU afinyelela kwangu-12 nememori efika ku-128TB, isebenzisa ukupholisa okuwuketshezi, futhi kulindeleke ukuthi yethulwe ngonyaka ozayo. #IsevaYeAI#
Ama-modeli kanye nobuchwepheshe2026-08-23Yehliswe ngo-80%! I-OpenAI yehlisa izindleko zemodeli ye-AI i-GPT-5.6 Luna, inenani elingcono kune-DeepSeek V4 ProCishe emasontweni angu-3 ngemva kokukhishwa kochungechunge lwe-GPT-5.6 (olukhishwe ngo-9 Julayi), inkampani i-OpenAI imemezele namuhla (31 Julayi) ukuthi yehlisa izindleko zokusebenzisa amamodeli i-GPT-5.6 Terra ne-GTP-5.6 Luna. Lolu shintsho aluthinti imodeli i-GPT-5.6 Sol okwamanje.
Ama-modeli kanye nobuchwepheshe2026-08-23Amamodeli amakhulu e-AI aseChina ahola uhlu lomhlaba, i-CCTV ichaza izinzuzo ezisemqokaNgokombiko we-CCTV News, amamodeli amakhulu e-AI aseChina ahola uhlu lomhlaba, aphumelela kabili ekusetshenzisweni nasekudawunilodweni. ULi Wei, iphini lomqondisi weCloud Computing and Digitalization Research Institute eChina Academy of Information and Communications Technology, uthe amamodeli akhiqizwe eChina avuselela uhlelo lokusungula izinto ezintsha ngesu lomthombo ovulekile. Ngenxa yezinzuzo ezivelele ezifana nokusebenza kahle, izindleko ezilawulekayo kanye nokufakwa kalula ezindaweni zabasebenzisi, ehlisa kakhulu umkhawulo wokusebenzisa ubuchwepheshe bobuhlakani bokwenziwa futhi athandwa kakhulu ngabathuthukisi emhlabeni wonke.
Ama-modeli kanye nobuchwepheshe2026-08-23Ucwaningo olusha: Usongo lwangempela lwe-AI emakethe yezemisebenzi akukhona ukuphelelwa umsebenzi, kodwa ukwenza ukukhushulwa kwemiholo kube nzimaUcwaningo lwakamuva lubonisa ukuthi i-AI ayikabi nomthelela obonakalayo emsebenzini uwonke, kodwa yehlisa kakhulu ijubane lokukhula kwemiholo, kanti abasebenzi abahola kancane yibona abathinteka kakhulu. Lokhu kungase kuqhubeke kukhulisa ukungalingani kwemali engenayo esikhathini esizayo.
Ama-modeli kanye nobuchwepheshe2026-08-23Ama-agent amaningi e-AI asebenza ngokubambisana, i-SpaceXAI ka-Musk yethula i-Grok BotLo mkhiqizo uvumela ama-agent amaningi e-AI ukuthi asebenzisane ukuze aqedele imisebenzi ehlukahlukene. Usuvele uhlolwa emaqenjini amaningi ngaphakathi enkampanini, futhi uyisinyathelo sakamuva sika-Musk sokuncintisana nezimbangi ze-AI.#Ama-agent e-AI# #Grok Bot#