🏛️ Indlela ubuchwepheshe obukhulu obuthole ngayo indlela yabo ngomyalelo kaTrump we-AI ↗
UTrump wahoxa emyalweni ohleliwe we-AI ophethe owawuzophoqa ukubuyekezwa kokuphepha kwamamodeli amasha. Isizathu, empeleni: ungayinciphisi ijubane leMelika emncintiswaneni neChina.
Ingxenye ephawulekayo ukuthi i-oda kubikwa ukuthi lalincane kakade - ngokuzithandela, hhayi ngesando sokulawula esibanzi. Noma kunjalo, abaholi bezobuchwepheshe baphikelela kakhulu kulo, futhi bathola ababekufuna. (The Guardian)
🛡️ I-Claude Mythos AI Ithola Ubungozi Obukhulu Obuyi-10,000 Kusofthiwe Esetshenziswa Kabanzi ↗
I-Anthropic ithe umzamo wayo we-Project Glasswing uthole ubuthakathaka obungaphezu kuka-10,000 obuphezulu noma obubucayi besebenzisa i-Claude Mythos Preview. Lokho kuyamangaza, kodwa futhi kuzwakala njenge-alamu yomlilo egqoke ikepisi.
Le nkampani ithi imiphumela engu-1,726 iqinisekisiwe njengemihle yangempela, kanti kukhishwe izeluleko ezingu-97 ezilungisiwe kanye nezeluleko ezingu-88. Ingxenye enzima manje akusikho ukuthola amaphutha - iwukuzilungisa ngaphambi kokuba abahlaseli bathole amathoyizi afanayo. (Izindaba zeHacker)
🛰️ I-White House ivumile u-$9 billion ukuze izinhlangano zezinhloli zikwazi ukuqapha i-AI ↗
I-White House ivumile isicelo esiyimfihlo samaRandi ayizigidi eziyizinkulungwane eziyi-9 sokuthenga ama-chip athuthukisiwe ezinhlangano zezobunhloli zase-US, okuhloswe ngawo ukusebenzisa amamodeli e-AI asemngceleni womsebenzi oyimfihlo.
IPhalamende kusadingeka livume uxhaso olukhulu, kodwa amaRandi ayizigidi ezingu-800 ahlelwa kabusha ukuze kufinyeleleke ngokushesha ekubalweni. Umbhalo ongezansi kunzima ukuwukhohlwa: ngisho nezinhlangano zezinhloli manje zibhajwe emgqeni wama-chip e-AI, okuyinto engenangqondo futhi engabikezeleki. (Inquirer.com)
🪞 'Awukwazi ukulawula yonke into': ukwanda kodokotela abahlinzayo bepulasitiki abacelwe ukudala 'ubuso be-AI' ↗
Odokotela abahlinzayo be-plastic bathi iziguli eziningi ziletha izithombe ze-selfie ezihlelwe yi-AI njengezinhloso zokuhlinzwa - amathambo abukhali, ukulingana okuphelele, isikhumba esingenazimbobo, uhlu lonke lokuthenga olumangalisayo.
Odokotela abahlinzayo baxwayisa ngokuthi i-AI ingahlela kabusha amaphikseli, hhayi i-anatomy. Izinqubo ezinconywe yi-AI ezibuyekeziwe ezingadlula u-£100,000 kodwa zingalethi ubuso obungajwayelekile. I-maze encane esabekayo yesibuko, lena. (I-Guardian)
💰 I-Anthropic izovala ngaphezulu kwamaRandi ayizigidigidi ezingama-30 ngesonto elizayo ↗
Kubikwa ukuthi i-Anthropic iseduze nomjikelezo wokuxhaswa ngemali ongaphezu kwamaRandi ayizigidi eziyizinkulungwane ezingama-30, ngenani elilinganiselwa kumaRandi ayizigidi eziyizinkulungwane ezingama-900. Yebo, lelo nani lisebenza kakhulu.
Lo mjikelezo uqhutshwa yisidingo esikhulu sikaClaude kuzo zonke izinhlelo zokubhala amakhodi, ukusebenza kwebhizinisi kanye nokuphepha kwe-inthanethi. Le nkampani iphinde igijimele ukuvala amakhompyutha amaningi, ngoba ku-AI manje, imali isetshenziswa kakhulu uma ugqoke isudi. (Business Standard)
🖼️ Izithombe ze-ChatGPT Ziphethe Amamaki e-AI Angabonakali Noma ubani Angawathola: Lokho Abasebenzisi Abangakwazi Ukudalula I-Gen AI Okudingeka Bakwazi ↗
Imiphumela yesithombe se-OpenAI manje inezimpawu ezimbili ezivelayo: i-metadata ye-C2PA kanye ne-watermark ye-SynthID ye-Google DeepMind. Enye inikeza umongo wefayela ocebile, kanti enye ihloselwe ukusinda kuzithombe-skrini kanye nokucindezelwa.
Kukhona futhi ithuluzi lokuqinisekisa umphakathi elisekubukeni kuqala. Akulona iphutha, cha - abacwaningi babonise ukuhlaselwa kwe-watermark - kodwa kwenza ukuthi "angizange ngisebenzise i-AI" kube yi-fib encane eyingozi kakhulu. (Tech Times)
Imibuzo Evame Ukubuzwa
Kungani uTrump ehoxile kumyalelo wokuphatha we-AI ohleliwe?
Lesi sihloko sithi umyalo ohleliwe we-AI kubikwa ukuthi wehlisiwe ngemuva kwengcindezi evela kubaholi bezobuchwepheshe abakhulu. Lo myalelo wawusuvele uchazwe njengomnene futhi ngokuzithandela, ugxile ekubuyekezweni kokuphepha kwamamodeli amasha e-AI. Impikiswano eyinhloko ngokumelene nawo ukuthi izinqubo ezengeziwe zokubuyekeza zinganciphisa i-US emncintiswaneni wayo we-AI neChina.
Yini eyatholwa yi-Anthropic's Claude Mythos kusofthiwe?
I-Anthropic ithe umzamo wayo we-Project Glasswing usebenzise i-Claude Mythos Preview ukuthola ubuthakathaka obungaphezu kuka-10,000 obuphezulu noma obubucayi kusofthiwe esetshenziswa kabanzi. Lesi sihloko siphawula ukuthi okutholakele okungu-1,726 sekuqinisekisiwe njengezinto ezinhle ngempela, kanti ezinye iziqeshana nezeluleko sezivele zikhishwe. Ukukhathazeka okubanzi ukuthi abavikeli bangalungisa yini lobu buthakathaka ngaphambi kokuba abahlaseli basebenzise amathuluzi e-AI afanayo.
Kungani izinhlangano zezinhloli zase-US zicela amanye ama-AI chips?
Lesi sihloko sithi i-White House ivumile isicelo esiyimfihlo samaRandi ayizigidi eziyizinkulungwane ezingu-9 sokusiza izinhlangano zezobunhloli zase-US ukuthi zithenge ama-chip athuthukile amamodeli e-AI asemngceleni. Umgomo uwukusekela umsebenzi we-AI ohlukanisiwe odinga inani elikhulu lokubala. Iphinde iphawule ukuthi iCongress isadinga ukuvuma uxhaso olubanzi, kuyilapho enye imali ihlelwa kabusha ukuze ifinyeleleke ngokushesha.
Kusho ukuthini ukuthi “ubuso be-AI” ekuhlinzeni ubuhle?
“Ubuso be-AI” bubhekisela ezigulini eziletha izithombe ze-selfie ezihlelwe yi-AI kodokotela abahlinzayo be-plastic njengezibonelo zendlela abafuna ukubukeka ngayo. Lezi zithombe zingase zibonise izici ezibukhali, ukulingana okuphelele, noma isikhumba esingesona esingokoqobo. Odokotela abahlinzayo abacashunwe kulesi sihloko baxwayisa ngokuthi i-AI ingashintsha amaphikseli kalula, kodwa ukwakheka komzimba kwangempela kunemingcele ukuhlinzwa okungenakuyinqoba ngaso sonke isikhathi.
Izimpawu ezingabonakali ezithombeni ze-ChatGPT zibathinta kanjani abasebenzisi?
Lesi sihloko sithi imiphumela yesithombe se-ChatGPT manje ifaka izimpawu zemvelaphi njenge-metadata ye-C2PA kanye ne-watermarking ye-SynthID. Lawa ma-marker angasiza ekuboneni izithombe ezikhiqizwe yi-AI, kufaka phakathi ezimweni lapho othile engase angafuni ukudalula ukusetshenziswa kwe-AI. Lesi sihloko siphinde siphawule ukuthi i-watermarking ayiyona into engalungile, kodwa yenza ukusetshenziswa kwesithombe se-AI esingadalulwanga kube lula ukukubona.