Imishini Yokucabanga Eqalayo ye-AI yethula imodeli ye-AI evulekile ↗
I-Mira Murati's Thinking Machines yembule i-Inkling, imodeli yayo yokuqala yenhloso ejwayelekile. Uhlelo lwamapharamitha ayizigidigidi ezingu-975 luvulekile, luvumela abathuthukisi ukuthi balulande, balusebenzise futhi balwenze ngezifiso esikhundleni sokuqasha ukufinyelela nge-API evaliwe.
I-Inkling isalandela amamodeli anamandla kakhulu, kodwa kubikwa ukuthi incintisana kahle nemisebenzi yama-ejenti. Lokho kubalulekile - abathuthukisi baseNtshonalanga bebelokhu bethembele kakhulu kumamodeli avulekile aseShayina muva nje, futhi manje sekukhona omunye osindayo eringini... ngandlela thile.
Isevisi ye-Apple Intelligence AI ibhaliswe nomlawuli we-cyberspace waseShayina ↗
I-Apple Intelligence iqede isithiyo esikhulu somthetho sokusetshenziswa kwama-iPhone eShayina. Ukubhaliswa kuvula ithuba lokukhishwa okulibazisekile isikhathi eside, yize kungekho usuku lokwethulwa olumenyezelwe.
Inguqulo yasendaweni ye-Apple izosebenzisa amakhono avela ku-Qwen ka-Alibaba kanye nobuchwepheshe obuthuthukiswe nge-Baidu. Kuyisitsha esingavamile se-AI - ihadiwe ye-Apple, amamodeli aseShayina, kanye nokuvunyelwa nguhulumeni okubambe i-toothpick ndawonye.
I-Cadence ikhipha i-AI agent ebhodini lesekethe lesivinini kanye nomklamo wokupakisha ama-chip ↗
I-Cadence yethule i-AuraStack, "i-ejenti ephezulu" ye-AI engakwazi ukuklama amabhodi wesekethe namaphakheji e-chip kusuka emiyalweni yolimi olulula. Ihlela imisebenzi yokusebenza, isebenzisa amathuluzi obunjiniyela be-Cadence futhi ihlola nge-inthanethi imiklamo etholakala.
Le nkampani ithi i-AuraStack inganciphisa isikhathi ngesigamu ukuze imakethe futhi ilethe izinzuzo zokukhiqiza ezifika ezikhathini ezingu-15 emisebenzini ngayinye. I-Nvidia, i-TSMC kanye ne-Schneider Electric baphakathi kwabasebenzisi bokuqala - ngokusobala, yilapho ama-ejenti eyeka khona ukuhlela amakhalenda bese eqala ukuthinta insimbi.
Ngokwemvelo, iBlackstone ibheja ukuthi ibhizinisi elilandelayo le-AI lama-trillion dollar liwukuqaliswa, hhayi amamodeli kuphela ↗
I-Anthropic kanye neqembu labasekeli abakhulu bezezimali baqambe ibhizinisi labo lokuqalisa i-AI i-Ode. Leli bhizinisi linamaRandi ayizigidi eziyizinkulungwane eziyi-1.5 futhi lihlose ukusiza izinkampani ezinkulu ukuguqula amamodeli anekhono abe izinhlelo ezisebenza ngaphakathi kwezinhlangano eziyinkimbinkimbi.
I-Ode ikhule kusuka ku-Fractional AI futhi izothola amakhasimende angaba khona avela kumaphothifoliyo abaxhasi bayo bezimali ezizimele. Ukubheja kusobala impela: inkampani enkulu elandelayo ye-AI ingase ingakhi imodeli ehlakaniphile kakhulu - ingase ifake amapayipi ngaphandle kokugcwala ihhovisi.
Kubikwa ukuthi iMicrosoft iqeqesha abathengisi ukuthi bakhulume kabi nge-OpenAI kanye ne-Anthropic ↗
Kubikwa ukuthi iMicrosoft iyale amaqembu okuthengisa ukuthi aqhathanise imikhiqizo yayo ye-AI neminikelo evela ku-OpenAI, Google kanye ne-Anthropic. Abaphathi baphakamise inkulumo yezinhlelo kusukela ekuqaleni kuya ekugcineni egxile ekusebenzeni kahle, izindleko kanye nokuhlanganiswa okuqinile nesofthiwe ye-Microsoft.
Kubikwa ukuthi isethulo esisodwa sichaze uClaude njengohamba kancane, onganembile kakhulu futhi obuthakathaka ekuphepheni ngaphakathi kwezinhlelo zokusebenza zehhovisi le-Microsoft. Kubi, yebo - iMicrosoft manje incintisana ngokusobala nabahlinzeki bemodeli ebithembele kubo iminyaka eminingi. Ozakwethu, izimbangi, izitha... khetha ilebula likaTuesday.
I-Microsoft iqapha inani lobuthakathaka bokuphepha, icaphuna ukusetshenziswa kwayo kwe-AI ↗
I-Microsoft ikhiphe ukulungiswa kwamaphutha okuphepha angu-570 kuwo wonke ama-Windows, ama-Office kanye neminye imikhiqizo, okuphawula ukukhishwa kwayo okukhulu kakhulu kwe-patch yanyanga zonke. Okungenani ezimbili zaziyizinkinga zezinsuku ezingenalutho ezivele zisetshenziswa abahlaseli.
Le nkampani ngokwengxenye income ucwaningo olusizwa yi-AI ngokuthola ezinye iziphazamisi ezilele. Ngakho-ke i-AI idala ikhanda elikhulu lokuphepha ngenkathi isiza nokuthola ikhanda elidala - eliyindilinga ngendlela engavamile, njenge-alamu yentuthu efunda ukukhiqiza intuthu.
I-Vint Cerf isebenza ohlelweni lokukhulula ama-ejenti e-AI ku-inthanethi evulekile ↗
Iphayona le-inthanethi uVint Cerf ujoyine i-Innovation Labs njengomeluleki wezengqalasizinda zokuhlonza ama-ejenti e-AI ku-inthanethi. Isiphakamiso senhlangano se-DNSid sizoxhumanisa ama-ejenti namagama ezizinda aqinisekisiwe futhi sisebenzise ubufakazi be-cryptographic ukurekhoda ubunikazi bawo.
Lo mbono ubhekana nenkinga ekhula ngokushesha: ama-ejenti azimele angase asebenze kumawebhusayithi ngaphandle kwendlela evamile yokuthola ukuthi ubani obalawulayo, ukuthi banegunya lini noma ukuthi ubani ophethe icala. Ngokuyisisekelo amapasipoti ama-bot... ngaphandle kokuthi ipasipoti ingaxoxisana nelinye ipasipoti ngenkathi ulele.
Imibuzo Evame Ukubuzwa
Kuyini i-Inkling, futhi kungani ukuqala kwayo kubalulekile?
I-Inkling yethulwa njengokwethulwa okusha, yize imininingwane etholakalayo ingachazi umkhiqizo wayo, imakethe, noma amakhono ayo. Ukuqala kwayo kuvame ukukhombisa ukuthi inkampani noma ipulatifomu idlulela ngale kwentuthuko futhi ifinyelele ekutholakaleni komphakathi. Imibuzo ebalulekile ukuthi iyiphi inkinga i-Inkling eyixazulula, yakhelwe bani, nokuthi ihluke kanjani kwezinye izindlela ezikhona.
Kusho ukuthini ukuthi i-Apple clearing China AI iqalise ukusebenza kwayo?
I-Apple clearing China AI iphakamisa inqubekela phambili ekunikezeni izici ze-AI emakethe yaseShayina, lapho imikhiqizo ivame ukubhekana nezidingo zomthetho nezokusebenza zendawo. Lokhu kungasusa isithiyo esikhulu ekusetshenzisweni, yize ukuvunyelwa ngokwako kungase kungaqinisekisi ukuqaliswa ngokushesha. Ukutholakala, amadivayisi asekelwayo, ukubambisana kwendawo, kanye nokwehluka kwezici kungase kusahluka ngokwesifunda.
I-AuraStack ikwenza kanjani ngokuzenzakalelayo ukuthuthukiswa kwe-chip?
I-AuraStack ichazwa ngokuthi ama-chip azenzakalelayo, okungenzeka ukuthi abhekisela ekwenzeni lula izingxenye zomklamo we-semiconductor, ukucushwa, ukuhlolwa, noma ukuthunyelwa. Kuwo wonke ama-chip workflows amaningi, ukuzenzakalela kunganciphisa imisebenzi yobunjiniyela ephindaphindwayo futhi kuthuthukise ukuhambisana. Umthelela wayo osebenzayo uncike ekutheni yiziphi izigaba i-AuraStack ezisekelayo nokuthi ihlangana kahle kangakanani namathuluzi okuklama nawokukhiqiza akhona.
Kusho ukuthini ukukhulisa ukukhishwa kwe-Ode kubasebenzisi?
Ukwandisa ukukhishwa kwayo kwe-Ode kubonisa ukuthi ukufinyelela kuyanda ngale kokukhishwa kwangaphambilini noma okulinganiselwe kakhulu. Izinkampani zivame ukwandisa ukutholakala kancane kancane ukuze zikwazi ukuqapha ukuthembeka, isidingo, kanye nempendulo yabasebenzisi ngaphambi kokufinyelela izithameli eziningi. Abasebenzisi kufanele baqaphele ukufaneleka okubuyekeziwe, ukufinyelela kwesifunda, amanani, izidingo zokungeniswa, kanye nanoma yiziphi izinguquko ezethulwa njengoba ukukhishwa kukhula.
Kungani iMicrosoft ihlasela izimbangi zayo ngenkathi ilungisa amaphutha angu-570?
Isihloko sihlanganisa indawo yokuncintisana nesibuyekezo esikhulu sokuphepha. I-Microsoft ibonakala igxeka noma iphonsela inselelo izimbangi ngenkathi ibhekana nobuthakathaka noma amaphutha esofthiwe angu-570. Kumakhasimende, okubaluleke kakhulu ukuphathwa kwama-patch: ukubuyekeza imikhiqizo ethintekile, ukuhlola izibuyekezo, kanye nokufaka izilungiso ezibalulekile ngokushesha ngokuhambisana nezinqubo zokuphepha zangaphakathi.
I-DNSid iwabona kanjani ama-ejenti e-AI?
I-DNSid ichazwa njengohlelo lokuhlonza ama-ejenti, okungenzeka ama-ejenti e-AI noma izinsizakalo ezenzakalelayo ezisebenza kuwo wonke amanethiwekhi. Indlela evamile ukunikeza ama-ejenti ubunikazi obuqinisekisiwe obusiza amasistimu ukuthi awabone, awaqinisekise, noma awaqaphe. Lokhu kungase kusekele ukuphepha, ukuziphendulela, kanye nokulawula ukufinyelela, yize imininingwane yokusebenzisa izonquma ukuthi inqubo yokuhlonza ithembeke futhi ingasebenzisana kangakanani.