{"id":28847,"date":"2025-06-21T11:13:58","date_gmt":"2025-06-21T08:13:58","guid":{"rendered":"https:\/\/kibrisayna.com\/index.php\/2025\/06\/21\/bu-sorulari-sormayin-yapay-zekayi-zorlayan-konular-son-dakika-teknoloji-haberleri\/"},"modified":"2025-06-21T11:13:58","modified_gmt":"2025-06-21T08:13:58","slug":"bu-sorulari-sormayin-yapay-zekayi-zorlayan-konular-son-dakika-teknoloji-haberleri","status":"publish","type":"post","link":"https:\/\/kibrisayna.com\/index.php\/2025\/06\/21\/bu-sorulari-sormayin-yapay-zekayi-zorlayan-konular-son-dakika-teknoloji-haberleri\/","title":{"rendered":"Bu sorular\u0131 sormay\u0131n! Yapay zekay\u0131 zorlayan konular – Son Dakika Teknoloji Haberleri"},"content":{"rendered":"


\n<\/p>\n

\n

Yapay zekaya y\u00f6neltilen her soru, yaln\u0131zca bir yan\u0131tla de\u011fil, ayn\u0131 zamanda bir miktar karbondioksit sal\u0131m\u0131yla da geri d\u00f6n\u00fcyor.<\/p>\n

Almanya\u2019daki M\u00fcnih Uygulamal\u0131 Bilimler \u00dcniversitesi\u2019nden ara\u015ft\u0131rmac\u0131lar, ChatGPT gibi geni\u015f dil modellerinin (GDM) ak\u0131l y\u00fcr\u00fctme gerektiren sorulara verdikleri yan\u0131tlar\u0131n, \u00e7evresel etkisinin ciddi boyutlara ula\u015ft\u0131\u011f\u0131n\u0131 ortaya koydu.<\/p>\n

6 KAT DAHA FAZLA\u00a0<\/strong><\/p>\n

Hakemli bilim dergisi Frontiers\u2019ta yay\u0131mlanan \u00e7al\u0131\u015fmada, 14 farkl\u0131 yapay zeka<\/a> modeli kar\u015f\u0131la\u015ft\u0131r\u0131ld\u0131. Soyut cebir ya da felsefe gibi daha derin muhakeme isteyen konulardaki sorular, lise tarih dersi gibi basit i\u00e7eriklere k\u0131yasla yakla\u015f\u0131k 6 kat daha fazla karbon sal\u0131m\u0131 \u00fcretiyor.<\/p>\n

\u00c7al\u0131\u015fman\u0131n yazar\u0131 Dr. Maximilian Dauner, \u201cY\u00fcksek d\u00fczeyde muhakeme gerektiren sorular, modellerin daha fazla enerji harcamas\u0131na yol a\u00e7\u0131yor. Bu da do\u011frudan daha fazla karbon sal\u0131m\u0131na neden oluyor\u201d dedi. <\/p>\n

Dr. Dauner, mant\u0131kl\u0131 d\u00fc\u015f\u00fcnebilen b\u00fcy\u00fck modellerin, yal\u0131n ve k\u0131sa yan\u0131tlar veren modellere g\u00f6re 50 kata kadar daha fazla karbon sal\u0131m\u0131 yapabilece\u011fini s\u00f6yledi.<\/p>\n

KISA VE DO\u011eRUDAN YANITLAR TALEP ED\u0130N\u00a0<\/strong><\/p>\n

Yapay zekaya sorulan her soru, say\u0131sal i\u015flem gerektiriyor. Sorgudaki her kelime ya da par\u00e7a, modele i\u015flenmek \u00fczere \u201ctoken\u201d ad\u0131 verilen dijital birime d\u00f6n\u00fc\u015ft\u00fcr\u00fcl\u00fcyor. \u00c7al\u0131\u015fmaya g\u00f6re, muhakeme temelli bir soru ortalama 543,5 token \u00fcretirken, basit bir sorguda bu say\u0131 yaln\u0131zca 40 oluyor.<\/p>\n

Ara\u015ft\u0131rmac\u0131lar, kullan\u0131c\u0131lar\u0131n karbon ayak izini azaltmak i\u00e7in yapay zekadan daha k\u0131sa ve do\u011frudan yan\u0131tlar talep etmesini, karma\u015f\u0131k modellerin ise yaln\u0131zca ger\u00e7ekten ihtiya\u00e7 duyulan durumlarda kullan\u0131lmas\u0131n\u0131 \u00f6neriyor.<\/p>\n

\u00c7al\u0131\u015fma ayr\u0131ca, y\u00fcksek do\u011fruluk oran\u0131na sahip modellerin \u00e7evresel bedelini de ortaya koydu. \u00d6rne\u011fin yakla\u015f\u0131k y\u00fczde 85 do\u011fruluk oran\u0131na ula\u015fan Cogito modeli, daha sade yan\u0131tlar veren benzer boyuttaki modellere g\u00f6re 3 kat daha fazla emisyon sal\u0131yor.<\/p>\n

Do\u011fruluk ile s\u00fcrd\u00fcr\u00fclebilirlik aras\u0131nda bir denge kurmak \u015fart
Dr. Dauner, \u201c\u015eu anda yapay zeka teknolojilerinde do\u011fruluk ve s\u00fcrd\u00fcr\u00fclebilirlik aras\u0131nda bir taviz ili\u015fkisi var. Do\u011fru yan\u0131tlar i\u00e7in daha fazla enerji harc\u0131yoruz. Ancak bu dengeyi daha s\u00fcrd\u00fcr\u00fclebilir k\u0131lmak m\u00fcmk\u00fcn\u201d dedi.<\/p>\n

\u00c7arp\u0131c\u0131 bir \u00f6rne\u011fe g\u00f6re, DeepSeek R1 modeline 600 bin soru sorulmas\u0131, Londra-New York aras\u0131 gidi\u015f-d\u00f6n\u00fc\u015f u\u00e7u\u015funun karbon ayak izine e\u015fde\u011fer bir emisyon yaratabiliyor. Buna kar\u015f\u0131n, Alibaba Cloud\u2019\u0131n Qwen 2.5 modeli benzer do\u011frulukla daha fazla soruyu ayn\u0131 karbon sal\u0131m\u0131 d\u00fczeyinde cevaplayabiliyor.<\/p>\n

Ara\u015ft\u0131rmac\u0131lar, bu t\u00fcr verilerin kullan\u0131c\u0131lar\u0131 daha bilin\u00e7li bir yapay zeka kullan\u0131m\u0131 konusunda y\u00f6nlendirmesini umuyor.<\/p>\n<\/div>\n


\n
<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Yapay zekaya y\u00f6neltilen her soru, yaln\u0131zca bir yan\u0131tla de\u011fil, ayn\u0131 zamanda bir miktar karbondioksit sal\u0131m\u0131yla da geri d\u00f6n\u00fcyor. Almanya\u2019daki M\u00fcnih Uygulamal\u0131 Bilimler \u00dcniversitesi\u2019nden ara\u015ft\u0131rmac\u0131lar, ChatGPT gibi geni\u015f dil modellerinin (GDM) ak\u0131l y\u00fcr\u00fctme gerektiren sorulara verdikleri yan\u0131tlar\u0131n, \u00e7evresel etkisinin ciddi boyutlara ula\u015ft\u0131\u011f\u0131n\u0131 ortaya koydu. 6 KAT DAHA FAZLA\u00a0 Hakemli bilim dergisi Frontiers\u2019ta yay\u0131mlanan \u00e7al\u0131\u015fmada, 14 farkl\u0131 […]<\/p>\n","protected":false},"author":908,"featured_media":28848,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[12],"tags":[],"class_list":["post-28847","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-teknoloji"],"jetpack_publicize_connections":[],"jetpack_featured_media_url":"https:\/\/kibrisayna.com\/wp-content\/uploads\/2025\/06\/EGLYeKVuUEejI3RDNBRzZw.jpg","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/posts\/28847","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/users\/908"}],"replies":[{"embeddable":true,"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/comments?post=28847"}],"version-history":[{"count":0,"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/posts\/28847\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/media\/28848"}],"wp:attachment":[{"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/media?parent=28847"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/categories?post=28847"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/kibrisayna.com\/index.php\/wp-json\/wp\/v2\/tags?post=28847"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}