I-OpenAI ihlehlisa ukukhishwa kwe-GPT-5.6 emphakathini njengoba i-US ifuna ukufinyelela kusenesikhathi kumamodeli e-AI asemngceleni ↗
Ukukhishwa kwe-OpenAI GPT-5.6 kwaba indaba yokufinyelela kuhulumeni, hhayi nje ukukhishwa kwemodeli okukhazimulayo. Le nkampani ikhawulele ukuqaliswa kokuqala kwabalingani abahlolwe ngemuva kokuba izikhulu zase-US zicele ukubonakala kusenesikhathi.
Umndeni wamamodeli uhlanganisa uSol, uTerra noLuna. USol ubekwe njengomuntu ohlakaniphile, onekhodi eqinile, isayensi kanye nokuphepha kwe-inthanethi - okusebenzayo, kodwa futhi "ngicela ungakunikeli lokhu kuma-chaos goblins."
I-OpenAI ithe lolu hlobo lokukhetha amakhasimende kahulumeni akufanele lube yinto ezenzakalelayo. Yilokho ukucindezeleka: amasango okuphepha, kodwa hhayi umsele onefulegi kuwo.
Okukhethekile: Amabhange eGoldman athi ukuthuthuka okulandelayo kwe-AI kusemnotho ongokoqobo ↗
I-Goldman Sachs iqondisa ukukhanya kwe-AI kude nesofthiwe emsulwa futhi ibhekise ezimbonini, ezimayini, ezinsizeni kanye nasezindaweni zokugaya uwoyela. Ukukhazimula okuncane kwe-chatbot, ukushaja okuningi kwerobhothi-ingalo-endlini yokugcina izinto enothuli.
Impikiswano ilula: ingxenye enkulu yomnotho akuyona isofthiwe, ngakho-ke igagasi elikhulu lemali le-AI lingavela ezenzakalelayo ezindaweni ezithutha izinto, zenze izinto futhi zishise ngamandla.
Lokho kuzwakala kusobala uma sekushiwo ngokuzwakalayo, kodwa kusasho ushintsho olukhulu. I-AI ayibe yinto yesikrini nje futhi iba yisikrufu esinesikhungo sedatha esinamathiselwe.
Ukusheshisa amamodeli e-Gemini Nano ku-Pixel ngokubikezela kwe-Multi-Token eqandisiwe ↗
I-Google yabelane ngomsebenzi wokwenza i-Gemini Nano isheshe kumafoni e-Pixel kusetshenziswa ukubikezela kwe-Multi-Token eqandisiwe. Ukuhumusha: ifoni ingadweba amathokheni angaphezu kweyodwa ngesikhathi ngaphandle kokuqeqesha kabusha injini yonke njalo.
Ingxenye esebenzayo iwukubambezeleka. I-AI ekudivayisi izwakala imangalisa kuphela uma iphendula ngokushesha, futhi lokhu kuhloselwe ukwenza amamodeli endawo asheshe ngaphandle kokusebenzisa i-cloud computing enkulu kuyo yonke imisebenzi emincane.
Akuyona isihloko esikhulu kunazo zonke, kodwa kubalulekile. Ukunqoba okuncane ngesivinini kuyanezela ezikhathini zomkhiqizo othi “oh, lokhu kuyasebenza”.
I-IBM, i-Red Hat, kanye ne-Deloitte Bamemezela Ukusebenzisana kwe-Lightwell Ukusiza Ukuqinisa I-Open Source Software Supply Chain Trust ↗
I-IBM, i-Red Hat kanye ne-Deloitte bamemezele ukusebenzisana kwe-Lightwell okugxile ekuphepheni kwe-software yomthombo ovulekile. Inkulumo: lungisa isofthiwe esengozini ngokushesha, ngaphandle kokuphoqa izinkampani ukuthi zithuthukise ngendlela ephazamisayo.
Bahlela usongo njengokucindezela kwe-cyber okuzenzakalelayo, kanye ne-AI yasemngceleni okwenza ukutholakala kobuthakathaka kanye nokuxhashazwa kube lula. Kuhle, ngendlela i-Excel yokufunda ngoshaka emuhle ngayo.
I-angle kaLightwell iyiziqeshana eziqinisekisiwe ze-backport zezinguqulo zesofthiwe amabhizinisi asevele esebenza. Ayikhazimuli? Yebo. Futhi iphuzu elikhulu.
I-Patronus AI iqoqe u-$50M kuma-ejenti e-AI avivinya ukucindezeleka ↗
I-Patronus AI iqoqe u-$50M ukwakha izindawo ezilingisiwe zokuhlola ama-ejenti e-AI ngaphambi kokuthi athinte izinhlelo ezibukhoma. Ngokuyisisekelo, ilebhu yokuhlolwa kokuphahlazeka kwama-ejenti esofthiwe.
“Amamodeli Omhlaba Wedijithali” enkampani ahloselwe ukulingisa amawebhusayithi nezinhlelo zangaphakathi, avumele ama-ejenti ukuthi azilolonge imisebenzi emide futhi adalule ukuthi akhohlisa kuphi, aphula kuphi noma athatha izinqamuleli ezisolisayo.
Kuyinkinga enkulu manje. Wonke umuntu ufuna ama-ejenti enze umsebenzi ongokoqobo, kodwa akekho ofuna ukuthi axhumanise i-toaster ngokuqiniseka ohlelweni lwentela, uma singasho kanjalo.
Imibuzo Evame Ukubuzwa
Kungani i-OpenAI ihlehlise ukukhishwa kwe-GPT-5.6 emphakathini?
I-OpenAI ihlehlise ukwethulwa kwe-GPT-5.6 emphakathini ngemuva kokuba izikhulu zase-US zifune ukubonakala kusenesikhathi kumamodeli e-AI asemngceleni. Esikhundleni sokuqhubeka nokwethulwa kabanzi, inkampani ikhawulele ukufinyelela kokuqala kubalingani abahloliwe. Lesi sinqumo siguqule ukukhishwa kwaba yimpikiswano ebanzi mayelana nokufinyelela kukahulumeni, ukulawulwa kokuphepha nokuthi ukufinyelela kusenesikhathi kufanele kube yisici esijwayelekile sokusetshenziswa kwamamodeli asemngceleni.
Yini uSol, uTerra noLuna emndenini wamamodeli we-GPT-5.6?
I-Sol, i-Terra kanye ne-Luna zichazwa njengezingxenye zomndeni wemodeli we-OpenAI we-GPT-5.6. I-Sol ibekwe njengenketho enekhono kakhulu, esebenza kahle ezindaweni ezifana nokubhala ikhodi, isayensi kanye nokuphepha kwe-cyber. Lesi sihloko sichaza lokhu njengokubalulekile nokubucayi, njengoba ikhono lobuchwepheshe elithuthukisiwe lingaletha izinzuzo zokukhiqiza ngenkathi liphakamisa nokukhathazeka ngokusetshenziswa kabi.
Kungani abatshalizimali bebheka i-AI emnothweni ongokoqobo?
Lesi sihloko sithi ababhange beGoldman Sachs babona ukuchuma okulandelayo kwe-AI kudlulela ngale kwesofthiwe emsulwa kuya ezimbonini, ezimayini, ezinsizeni kanye nasezindaweni zokugaya uwoyela. Umqondo uwukuthi ingxenye enkulu yomnotho incike emsebenzini ongokomzimba, hhayi nje ezikrinini nasezinhlelweni zokusebenza. Kusukela kulowo mbono, ukuzenzekela ezindaweni zezimboni kungaba yinto ebalulekile ekutshalweni kwezimali kanye nokusetshenziswa kwe-AI esikhathini esizayo.
Ukubikezela kwe-Multi-Token eqandisiwe kuyisiza kanjani i-Gemini Nano kumafoni e-Pixel?
Ukubikezela kwe-Frozen Multi-Token kwethulwa njengendlela yokwenza i-Gemini Nano isheshe kumafoni e-Pixel. Esikhundleni sokukhiqiza ithokheni elilodwa ngesikhathi, imodeli ingabhala amathokheni amaningi ngenkathi igwema ukuqeqeshwa kabusha kwesistimu ephelele njalo. Umgomo osebenzayo ukubambezeleka okuphansi, ngakho-ke i-AI ekudivayisi izwakala ishesha futhi isabela kakhudlwana.
Iyiphi inkinga uLightwell azama ukuyixazulula kwezokuphepha komthombo ovulekile?
U-Lightwell, ovela ku-IBM, u-Red Hat kanye no-Deloitte, ugxile ekuqiniseni ukwethenjwa kwe-software yomthombo ovulekile. Indlela yakhe igxile kuma-patches aqinisekisiwe ezinguqulo zesofthiwe ezisetshenziswa amabhizinisi kakade. Lokho kubalulekile ngoba izinkampani eziningi zidinga ukulungisa ubuthakathaka ngokushesha, kuyilapho ukuthuthukiswa okuphazamisayo kungaba nzima ezindaweni zokukhiqiza ezinezinhlelo ezindala noma ezihlanganiswe kahle.
Kungani ama-ejenti e-AI edinga izindawo zokuhlola ezilingisiwe?
I-Patronus AI yakha izindawo ezilingisiwe ukuze ihlole ama-ejenti e-AI ngaphambi kokuba asebenzisane nezinhlelo ezibukhoma. Lawa “maModeli Omhlaba Wedijithali” aklanyelwe ukuphinda amawebhusayithi namathuluzi angaphakathi, okuvumela ama-ejenti ukuthi azijwayeze imisebenzi eyinkimbinkimbi ngokuphephile. Umgomo uwukubona ukwehluleka, izinqamuleli noma ukuziphatha okuyingozi ngaphambi kokuba ama-ejenti athenjwe ngemisebenzi ebukhoma.