🔐 I-OpenAI ithola inkinga yokuphepha ehilela ithuluzi lomuntu wesithathu, ithi idatha yomsebenzisi ayifinyelelwanga ↗
I-OpenAI ithe ithuluzi lonjiniyela leqembu lesithathu elithintekile lithinte inqubo eliyisebenzisayo ukuqinisekisa izinhlelo zayo zokusebenza ze-macOS njengezisemthethweni. Le nkampani ithe ayitholanga bufakazi bokuthi idatha yomsebenzisi ifinyelelwe, izinhlelo zayo noma i-IP yayo iphazamisekile, noma isofthiwe yayo ishintshiwe - okuyiphuzu elibalulekile. (Reuters)
Impumela engokoqobo iseyimbi kakhulu. I-OpenAI ijikelezise izitifiketi futhi yashintsha ukuze iqinise izinto ngakho izinhlelo zokusebenza ze-ChatGPT ezingamanga kodwa ezikholisayo azibi yinkinga enkulu kunalokho okudingeka zibe yikho. Akuyona-ke ukwephulwa kwemithetho - kodwa ingenye yalezo zikhathi "zokuvuselela uhlelo lwakho lokusebenza manje". (OpenAI)
🧯 USam Altman uphendula isihloko 'esivuthayo' saseNew Yorker ngemuva kokuhlaselwa kwekhaya lakhe ↗
USam Altman waphikisa obala iphrofayili yaseNew York eyayigxeka kakhulu ngemuva kokuhlaselwa okusobala ekhaya lakhe, ebiza izingxenye zalo ngokuthi ziyadukisa futhi ziyimfihlo. Konke lokhu kwazwakala kungajwayelekile ngesikhundla sokuba yi-CEO - ukuzivikela kakhulu kunokucophelela, mhlawumbe ngomklamo. (TechCrunch)
Ngesikhathi esifanayo, amaphoyisa abopha umsolwa ngokuhlasela kukaMolotov okuboshwe endlini ka-Altman eSan Francisco, kungekho ukulimala okubikiwe. Ngakho indaba yashintsha kusukela ekugxekweni kwabezindaba kuya ekuphepheni ngokomzimba ngokuphazima kweso - manje, lokho sekuyingxenye yangempela yesigqi se-AI. (Izindaba ze-AP)
🏗️ Abaholi Be-OpenAI Stargate Bahlela Ukujoyina Amapulatifomu E-Meta ↗
Kubikwa ukuthi abantu abathathu abaphezulu abaxhumene nomzamo wengqalasizinda ye-OpenAI i-Stargate baya e-Meta. Lokho akusikho nje ukuguquka kwamakhono - kungukuguquka kwamakhono empi yokubala, okufinyelela ngendlela ehlukile lapho wonke umuntu ezama ukuthola izikhungo zedatha, ama-chip, kanye namandla.
Abaqashiwe basikisela ukuthi iMeta ayigcini nje ngokuchitha imali eningi kumamodeli, kodwa izama ukumunca abantu abaziyo ukwakha amapayipi ezimboni ngaphansi kwawo. Mhlawumbe komile phezu kwamanzi, kodwa yilapho ingxenye enkulu yomjaho ikhona manje. (Bloomberg)
🛡️ Ukubuka KukaClaude Mythos ↗
I-Anthropic ithe imodeli yayo entsha ye-Mythos inamandla ngokwanele kwezokuphepha kwe-inthanethi kangangokuthi ayikukhiphi kabanzi, okungenani okwamanje. Esikhundleni salokho, ifakwa emzamweni wokuzivikela olawulwa ngokuqinile, ngoba inkampani ikholelwa ukuthi imodeli ingaveza amaphutha esofthiwe ayingozi ngezinga, kahle, elithusayo kancane. (Red Anthropic)
Leso sixwayiso sesivele sibonakala. Imibiko ithi izikhulu zase-US nezinkampani ezinkulu zibheka ukwenyuka kwamandla ale modeli njengenkinga yangempela yengqalasizinda kanye nokuphepha, hhayi nje okunye ukwethulwa okumangalisayo. Okwesibili, umjikelezo we-AI unesigqoko esiqinile. (Axios)
☁️ I-CoreWeave isayine isivumelwano sefu le-AI ne-Anthropic, amasheya akhuphuka ↗
I-CoreWeave ithe izohlinzeka i-Anthropic ngomthamo wokubala amafu ngaphansi kwesivumelwano seminyaka eminingi, kanye nomthamo okulindeleke ukuthi ufike ku-inthanethi kamuva kulo nyaka. Kuyisikhumbuzo futhi ukuthi izinkampani zamamodeli zisashesha njengomzila wengqalasizinda ongaphansi kwazo - isofthiwe ekhangayo, izithiyo ezingokomzimba ezinonya.
Ku-Anthropic, isivumelwano siqinisa ukufinyelela ekubaleni komugqa we-Claude. Ku-CoreWeave, kuyisibonakaliso esisodwa sokuthi abadlali be-AI cloud abangochwepheshe bayaqhubeka nokudonsa ibhizinisi kusuka esigabeni esiphezulu sabakhi bemodeli, okuyinto - ngokumangazayo - ehlala iqinile. (Reuters)
💸 I-SiFive esekelwa yiNvidia ifinyelela inani lama- $3.65 billion lama-chip e-AI avulekile ↗
I-SiFive ithole umklomelo wezigidi ezingama- $400 ngenani elilinganiselwa ku-$3.65 billion, okuyinani elikhulu lokwethemba ekwakhiweni kwe-open chip yezinhlelo ze-AI. Lokhu akusikho ukuntengantenga kwesihlalo sobukhosi seNvidia okwamanje, cha - kodwa kukhombisa ukuthi abatshalizimali basafuna izindlela ezihlukile zokungena ku-hardware ye-AI.
Iphuzu elibanzi kunzima ukulikhohlwa. I-AI ayiseyona nje indaba eyisibonelo, noma ngisho nendaba ye-chip - isiba yimpi yokuthi yiziphi izendlalelo ezihlala zivulekile, yiziphi ezihlala ziyimfihlo, nokuthi ubani okhokhelwayo kuzo zonke izitobhi ebhandeni lokuhambisa. (TechCrunch)
Imibuzo Evame Ukubuzwa
Kwenzekeni ngenkinga yokuphepha kohlelo lokusebenza lwe-OpenAI macOS?
I-OpenAI ithe ithuluzi lonjiniyela leqembu lesithathu elithintekile lithinte inqubo yokusayina esetshenziswa ukuqinisekisa izinhlelo zayo zokusebenza ze-macOS. Le nkampani iphinde yathi ayitholanga bufakazi bokuthi idatha yomsebenzisi ifinyelelwe, izinhlelo zayo noma impahla yobuhlakani ithintekile, noma isofthiwe yayo ishintshiwe. Inkinga enkulu kwakuwukwethenjwa kanye nobuqiniso bezinhlelo zokusebenza, hhayi ukwephulwa kolwazi lwamakhasimende okuqinisekisiwe.
Ingabe kufanele ngibuyekeze uhlelo lokusebenza lwe-ChatGPT macOS ngemuva kwalesi sigameko se-OpenAI?
Yebo, ukubuyekeza uhlelo lokusebenza kuyindlela ewusizo. I-OpenAI ijikelezise izitifiketi futhi yaqinisa inqubo ngakho izinhlelo zokusebenza ze-ChatGPT ezingamanga kodwa ezikholisayo cishe azikwazi ukubangela ukudideka noma ukudala ingozi. Ezimweni ezinjengalezi, isinyathelo esiphephile ukusebenzisa inguqulo yakamuva esemthethweni futhi ugweme ukulanda izinhlelo zokusebenza zedeskithophu emithonjeni engekho emthethweni noma kumasayithi esibuko.
Kungani inkinga yethuluzi lonjiniyela wesithathu ibalulekile uma kungekho datha eyebiwe?
Ngoba ukwethenjwa kwesofthiwe kuncike kokungaphezu kokufinyelela idatha kuphela. Uma ithuluzi elihilelekile ekuqinisekisweni kohlelo lokusebenza lisengozini, lingaphakamisa ukungabaza ngokuthi abasebenzisi bangakwazi yini ukuhlonza isofthiwe esemthethweni ngokuthembekile. Ezindaweni eziningi zokukhiqiza, lolo hlobo lwenkinga lubalulekile ngoba luthinta ukuphepha kokusabalalisa, ukuzethemba kuzibuyekezo, kanye nengozi yokuzama ukuzenza ongeyena.
Kungani iMeta iqasha abaholi be-OpenAI Stargate bangaphambili kanye nezinye izinyathelo zengqalasizinda ze-AI ziyinto enkulu kangaka?
Lokhu kuqashwa kukhomba umncintiswano ngaphansi kwesendlalelo samamodeli, lapho izikhungo zedatha, ama-chip, amandla, kanye namandla okuthunyelwa kubaluleke kakhulu njengethalente locwaningo. Ingqalasizinda ye-AI iba yinzuzo yamasu, hhayi nje umsebenzi wokusekela. Lesi sihloko siphakamisa ukuthi i-Meta izama ukuqinisa uhlangothi lwezimboni lwe-AI, hhayi nje ukwengeza abacwaningi abaningi bemodeli.
Iyini imodeli ye-Anthropic's Mythos, futhi kungani inkampani inganciphisa ukukhishwa kwayo?
I-Anthropic ichaze i-Mythos njengethuthuke ngokwanele kwezokuphepha kwe-inthanethi kangangokuthi igcinwa ilawulwa ngokuqinile kunokuba ikhululwe kabanzi. Ukukhathazeka kubonakala sengathi uhlelo olunamandla lokuthola amaphutha esofthiwe lunganikeza inani lokuzivikela ngenkathi futhi luphakamisa izingozi zokusetshenziswa kabi. Indlela evamile ezimweni ezinjengalezi ukufinyelela okulinganiselwe, ukuthunyelwa okuncane, kanye nokuqapha okuseduze.
Kungani amadili engqalasizinda ye-AI kanye nokuxhaswa ngemali okuvulekile kubalulekile emakethe ebanzi ye-AI?
Zibonisa ukuthi umjaho we-AI ukhula ngokuya ngokwakhiwa kokufinyelela kokubala kanye nokukhetha kwehadiwe, hhayi nje ngezici ze-chatbot. Isivumelwano se-CoreWeave-Anthropic siqokomisa indlela izinkampani zamamodeli ezisathembele ngayo kumthamo wamafu, kuyilapho ukuxhaswa kwe-SiFive kukhombisa intshisekelo yabatshalizimali kwezinye izindlela ngaphakathi kwe-AI chip stack. Uma zihlanganiswa, lezo zinyathelo ziphakamisa ukuthi ingqalasizinda ye-AI isiba yindawo eyinhloko yokulwa nokukhula nokulawula.