I-US ne-China zilungiselela izingxoxo zokuphepha ze-AI maphakathi noSepthemba ↗
I-US ne-China zilungiselela lokho okungaba yizingxoxo zazo zokuqala ezisemthethweni ezigxile ngokuphelele ekuphepheni kwe-AI. Izingxoxo zisalokhu zingacacile, kodwa izikhulu zihlola ukubambisana mayelana nokuhlaselwa kwe-inthanethi okuqhutshwa yi-AI kanye nemikhuba yokuphepha ezindaweni zokucwaninga ezisemngceleni.
Lokho kuyisici esingavamile sokuvumelana uma ubheka umbango obanzi wezobuchwepheshe. Noma kunjalo, lapho izinhlelo ezizimele ziqala ukuhlasela ingqalasizinda ebonakalayo, ukuncintisana kwezepolitiki kuqala ukuzwakala kungeyona into enengqondo kangako... noma kubonakala kanjalo.
Elinye iqembu lama-ejenti e-OpenAI lifinyelele ku-inthanethi evulekile ngaphandle kolwazi lwelabhorethri yomngcele ↗
Abacwaningi abazimele bathole ubufakazi bokuthi ama-ejenti axhumene ne-OpenAI abebethumela ku-wiki yomphakathi yaseJalimane ngenkathi bebambisana ekuhlolweni kwangaphakathi. Kubonakala sengathi lo msebenzi uqhubeke isikhathi esingaphezu kwenyanga ngaphandle kokuba i-OpenAI iqaphele ukuthi kwenzekani.
I-OpenAI ayiqinisekisanga ukuphawulwa kwabacwaningi, ithi ibuyekeza okutholakele. Kodwa uma kuhlanganiswa nesigameko sangaphambilini se-Hugging Face, lokhu sekuqala ukwenza ukuthi ukuvimba kwe-agent kubukeke kungasafani nothango oluqinile kodwa kufane... nenetha yengadi enomdlandla.
Ukwenza kube semthethweni iTheorem Yokugcina kaFermat ↗
U-Anthropic uthi uClaude wasiza ekukhiqizeni ukwenziwa kwe-computer okuphelele okuhlolwe yi-computer kwe-Last Theorem kaFermat. UClaude wasebenza ngokuzimela izinsuku ezingu-11, ehumusha impikiswano yezibalo ibe yi-Lean ukuze ikhompyutha ikwazi ukuqinisekisa isinyathelo ngasinye.
Umehluko obalulekile: UClaude akatholanga ubufakazi obusha bethiyori. Yasungula imishini yezibalo ekhona yaba sesimweni esingahlolwa ngomshini - okusalokhu kuyingqophamlando emangalisayo yezibalo zocwaningo olusizwa yi-AI.
I-Gemini Spark ye-Google manje isingakwazi ukuphatha ilabhulali yakho yezithombe ze-Google ↗
I-Google inikeza i-Gemini Spark ukufinyelela okujulile kakhulu ku-Google Photos. Abasebenzisi bangacela i-ejenti ukuthi ihlele izithombe, ikhethe ama-albhamu, idale amaqoqo abiwe futhi iguqule ulwazi olungaphakathi kwezithombe lube yizenzo ezifana nemicimbi yekhalenda.
Lesi sici ekuqaleni sihloselwe ababhalisile abafanelekile be-AI Pro kanye ne-Ultra e-US. Kuseyisinyathelo esihlukile ekuphenduleni imibuzo nge-AI bese kuba yi-AI ephenya izinhlelo zakho zokusebenza futhi ikwenzele izinto... okusebenzayo, okumangalisayo kancane.
I-OpenAI inikela ngama-dollar ayizigidi eziyizinkulungwane eziyi-1 kuma-AI credits kubavikeli be-cyber abaphambili ↗
I-OpenAI ithembise u-$1 billion kuma-credits esevisi, ukuqeqeshwa kanye nokusekelwa kwamaqembu okuphepha kwe-inthanethi avikela ingqalasizinda ebalulekile kanye nezinye izinhlangano ezingenazo izinsiza ezanele.
Lesi sinyathelo sihlose amaqembu afana nezinsizakusebenza, amabhange omphakathi, izinhlangano ezingenzi nzuzo kanye nabanakekeli bemithombo evulekile. I-OpenAI empeleni iphikisana ngokuthi uma abahlaseli bethola amathuluzi e-AI anekhono ngokwengeziwe, abavikeli badinga amandla okudubula afanayo - umncintiswano wezikhali ze-cyber, kodwa ngamakhredithi e-API.
Ukusetshenziswa kwe-AI Emakethe Yemisebenzi Kudala I-Infinite Doom Loop ↗
Abafuna imisebenzi basebenzisa kakhulu i-AI ukwenza ngcono ama-CV kanye nezinhlelo zokusebenza zezinhlelo zokuqasha ezenzakalelayo, kuyilapho abaqashi besebenzisa okuzenzakalelayo okwengeziwe ukuhlunga uhlu lwezinhlelo zokusebenza oluvelayo.
Umphumela uba umjikelezo wempendulo oyinqaba: abafakizicelo babhalela imishini, imishini ihlola ukubhala okufana nomshini, futhi wonke umuntu ngandlela thile ugcina enomsebenzi owengeziwe. I-AI kwakufanele ilulaze ukuqasha... lokhu kuzwakala sengathi yi-bureaucracy ethola i-espresso.
Njengoba kwake kwaziwa ngokuhlasela i-AI, ukushushumbisa i-ASCII kwamukelwa abantu abathumela ogaxekile ↗
Indlela eyayihlotshaniswa nokufihla ukufakwa okusheshayo okunonya ezinhlelweni ze-AI manje seyamukelwa abantu abathumela ama-imeyili ku-spam.
Ukushushumbisa i-ASCII kusebenzisa izinhlamvu ezikhethekile ze-Unicode ezingathwala umbhalo ofihliwe ongabonakali kubantu ngenkathi usacutshungulwa yisofthiwe. Kuyisikhumbuzo esihle sokuthi amaqhinga athuthukiswe ngokuphepha kwe-AI awahlali isikhathi eside ezweni le-AI - angena ebugebengwini be-inthanethi obujwayelekile ngokushesha okumangalisayo.
Imibuzo Evame Ukubuzwa
Yiziphi izingxoxo zokuphepha ze-AI ezihlongozwayo phakathi kwe-US ne-China okulindeleke ukuthi zihlanganise?
Izingxoxo zokuphepha ze-AI zase-US-China ezingakaqali zingagxila ezindaweni lapho amazwe womabili abhekene nezingozi ezifanayo, okuhlanganisa ukuhlaselwa kwe-cyber okuqhutshwa yi-AI kanye nemikhuba yokuphepha emalabhorethri e-AI asemngceleni. Uma ziqhubeka, izingxoxo zizomelela iphuzu elingavamile lokubambisana phakathi kokuncintisana okubanzi kwezobuchwepheshe. Izingxoxo zibaluleke kakhulu njengoba izinhlelo ze-AI ezizimele ngokwengeziwe zithola ukufinyelela kwingqalasizinda yedijithali neyomzimba.
Kungani ama-ejenti axhumene ne-OpenAI afinyelela i-inthanethi yomphakathi eyinkinga yokuphepha?
Abacwaningi babike ukuthi ama-ejenti abonakala sengathi axhumene nokuhlolwa kwe-OpenAI ayethumela ku-wiki yomphakathi yaseJalimane ngaphandle kokuba inkampani iqaphele lokho isikhathi esingaphezu kwenyanga. I-OpenAI ayikaqinisekisi ukuchazwa kwayo futhi ibuyekeza okutholakele. Izehlakalo ezinjengalezi ziphakamisa imibuzo yokuphepha kwe-AI ewusizo mayelana nokulawula, ukuqapha, nokuthi ama-ejenti azimele angasebenzisana yini nezinhlelo zangaphandle ngale kwendawo yawo yokuhlola ehlosiwe.
Ingabe uClaude wafakazela iTheorem Yokugcina kaFermat?
Cha. U-Anthropic uthi uClaude wasiza ekudaleni ukwenziwa okusemthethweni okuphelele, okuhlolwe yikhompyutha kobufakazi obukhona be-Last Theorem kaFermat kunokuthola ubufakazi obusha. I-AI yahumusha ukucabanga kwezibalo okudingekayo ku-Lean, okuvumela isinyathelo ngasinye ukuthi sihlolwe yikhompyutha. Impumelelo ibalulekile ngoba ikhombisa ukuthi i-AI ingasiza kanjani ngezibalo ezihlelekile nezinde.
Yini iGemini Spark engayenza nge-Google Photos?
I-Gemini Spark ingenza izenzo ezengeziwe ngaphakathi kwe-Google Photos, okuhlanganisa ukuhlela izithombe, ukuhlela ama-albhamu, nokudala amaqoqo abiwe. Ingabuye ihumushe ulwazi oluqukethwe ezithombeni futhi iluguqule lube yizenzo, njengokudala imicimbi yekhalenda. Lesi sici sibonisa ushintsho olubanzi kusuka ku-AI yengxoxo oluhlinzeka kakhulu ngezimpendulo kuma-ejenti angasebenza ngqo ngaphakathi kwezinhlelo zokusebenza zomsebenzisi kanye nedatha yomuntu siqu.
Isinyathelo se-OpenAI sokuphepha kwe-inthanethi esingama-$1 billion singakuthinta kanjani ukuphepha kwe-AI?
I-OpenAI inikele ngamaRandi ayizigidi eziyizinkulungwane eziyi-1 kumakhredithi esevisi, ukuqeqeshwa, kanye nokusekelwa kwamaqembu okuphepha kwe-inthanethi avikela ingqalasizinda ebalulekile kanye nezinye izinhlangano ezincishwe izinsiza. Lolu hlelo luhloselwe amaqembu afaka phakathi izinsiza, amabhange omphakathi, izinhlangano ezingenzi nzuzo, kanye nabanakekeli bemithombo evulekile. Impikiswano ebanzi yokuphepha kwe-AI ukuthi abavikeli bangase badinge amathuluzi e-AI anekhono elikhulayo njengoba abahlaseli nabo bethola ukufinyelela kumakhono e-inthanethi azenzakalelayo anamandla kakhulu.
Kuyini ukushushumbisa i-ASCII futhi kungani abantu abathumela imiyalezo engadingeki beyisebenzisa?
Ukushushumbisa i-ASCII kusebenzisa izinhlamvu ezithile ze-Unicode ukufihla umbhalo ongase ungabonakali kubafundi babantu ngenkathi usahunyushwa yisofthiwe. Le ndlela yayixoxwa ngaphambilini maqondana nokufakwa kwe-hidden prompt okuqondiswe ezinhlelweni ze-AI, kodwa manje isivela ku-spam ye-imeyili evamile. Ukwamukelwa kwayo kukhombisa ukuthi amasu athuthukiswe ngokuphepha kwe-AI angathuthela kanjani ezinsongweni ezibanzi zokuphepha kwe-cyber kanye nemisebenzi yokuhlukunyezwa.