{"status":true,"post":{"id":35480,"user_id":27,"status":1,"type":1,"orde":null,"notification_type":3,"static_post":0,"published_at":"2022-10-24 22:04:54","created_at":"2022-10-20T21:00:00.000000Z","updated_at":"2022-10-24T19:04:54.000000Z","edited_at":"2024-12-17 22:35:21","source_id":null,"post_id":35480,"is_featured":0,"title":"Yapay zekada korkutucu bir a\u015famaya do\u011fru","slider_title":null,"slider_title_2":null,"slider_spot_title":null,"slider_spot_title_2":null,"home_title":null,"sub_title":null,"category_id":73,"description":null,"content":"  \t\t\t\t\t  \t\t\t\t\t<p style=\"text-align: right;\"><span class=\"large\"><strong>DO\u00c7. DR. ADNAN ERTEMEL<\/strong><\/span><br><span class=\"large\"> <strong><\/strong><\/span><\/p>  <p><span class=\"large\">Yapay zeka teknolojisi, onlarca y\u0131l\u0131 \u00f6\u011frenme evresinde ge\u00e7irdi. \u015eimdi olgunla\u015f\u0131yor ve \u00f6l\u00e7ekleniyor. \u00d6yle ki, tahmin edilenden de h\u0131zl\u0131 bi\u00e7imde h\u0131zlanmaya ba\u015flad\u0131.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Sentetik verilerden beslenen ve kendi tasar\u0131mc\u0131lar\u0131n\u0131n bile akl\u0131n\u0131 ba\u015f\u0131ndan alan, kendi kendini denetleyen bir yapay zeka \u00e7a\u011f\u0131na girdik. Google\u2019a ait teknoloji laboratuvar\u0131 DeepMind\u2019deki ara\u015ft\u0131rmac\u0131lar, yak\u0131n zamanda piyasaya s\u00fcr\u00fclen AlphaTensor sisteminin, algoritmalar\u0131 bile kendi ba\u015f\u0131na ke\u015ffederek \u2018insan sezgisi alan\u0131n\u0131 a\u015ft\u0131\u011f\u0131n\u0131\u2019 s\u00f6yl\u00fcyor. Nature dergisinde ekim ay\u0131 ba\u015f\u0131nda yay\u0131nlanan makalede, DeepMind ara\u015ft\u0131rmac\u0131lar\u0131, derin peki\u015ftirmeli \u00f6\u011frenme yakla\u015f\u0131m\u0131yla \u00e7al\u0131\u015fan yapay zekan\u0131n matematikte bilim adamlar\u0131n\u0131n 50 y\u0131ld\u0131r benimsedi\u011fi matris \u00e7arp\u0131m algoritmalar\u0131ndan daha h\u0131zl\u0131 \u00e7al\u0131\u015fan bir \u00e7arp\u0131m algoritmas\u0131n\u0131 ke\u015ffetti\u011fini ortaya koydu. <\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> \u00d6rne\u011fin makina \u00f6\u011frenme mant\u0131\u011f\u0131nda \u00e7al\u0131\u015fan ve en geli\u015fmi\u015f yapay zeka sistemi olarak g\u00f6sterilen GPT-3, 175 milyar parameter ile \u00e7al\u0131\u015farak, kendi kendine \u00f6\u011frenme yetisi kazanm\u0131\u015f durumda. Bu sayede d\u00fcnyada \u015fu ana kadar kaleme al\u0131nm\u0131\u015f hemen t\u00fcm kitaplar\u0131 ve Wikipedia\u2019daki t\u00fcm i\u00e7eri\u011fi haf\u0131zas\u0131na alan GPT-3, ba\u015flang\u0131\u00e7ta cevaplamas\u0131n\u0131n \u00f6ng\u00f6r\u00fclmedi\u011fi sorular\u0131 bile cevaplamaya ba\u015flad\u0131.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Yapay zeka teknolojisinin geldi\u011fi bu yeni a\u015famada s\u0131radaki hedef, yapay zekay\u0131 farkl\u0131 sekt\u00f6rlerde \u00f6l\u00e7ekleme safhas\u0131\u2026 Bunun i\u00e7in de tek eksik, daha fazla miktarda veri setleriyle yapay zekay\u0131 beslemek. Bunun i\u00e7in geli\u015ftirilen bir y\u00f6ntem, ger\u00e7ek olmayan sentetik veri \u00fcretmek.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> <strong>SENTET\u0130K VER\u0130<\/strong><\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Ger\u00e7ek d\u00fcnyadaki olaylar arac\u0131l\u0131\u011f\u0131yla toplanmak yerine algoritmik olarak olu\u015fturulan veri olarak tan\u0131mlanan sentetik veri, yaz\u0131l\u0131m geli\u015ftirmeyi iyile\u015ftirmek, <\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Ar-Ge\u2019yi h\u0131zland\u0131rmak, makina \u00f6\u011frenimi modellerini e\u011fitmek, kendi dahili verilerini ve \u00fcr\u00fcnlerini daha iyi anlamak ve i\u015f s\u00fcre\u00e7lerini iyile\u015ftirmek i\u00e7in kullan\u0131l\u0131r. Sentetik veriler, ger\u00e7ek veri k\u00fcmelerinin yerini alabilir ve matematiksel modelleri do\u011frulamak i\u00e7in kullan\u0131labilir. Sa\u011fl\u0131k, finans, sigorta, siber g\u00fcvenlik, \u00fcretim, robotik ve otonom ara\u00e7lar gibi alanlardaki \u015firketlerin daha h\u0131zl\u0131 \u00f6l\u00e7eklenebilmeleri i\u00e7in geli\u015ftirmeyi ve pazara sunma s\u00fcresini h\u0131zland\u0131rmak i\u00e7in sentetik verilerin kullan\u0131m\u0131 b\u00fcy\u00fck avantaj sa\u011fl\u0131yor.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> <strong>\u0130Y\u0130 VE K\u00d6T\u00dc \u00d6RNEKLER<\/strong><\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Yapay zeka, ge\u00e7mi\u015fte tahmin bile edilemeyen bi\u00e7imde farkl\u0131 alanlarda sekt\u00f6rlerde ciddi katma de\u011fer olu\u015fturmaya ba\u015flad\u0131. \u00d6rne\u011fin, ABD\u2019de kurulan Syrup Tech adl\u0131 \u015firket, derin tedarik zinciri kesintileri, ger\u00e7ek \u00e7ok kanall\u0131 a\u011flar ve s\u00fcrd\u00fcr\u00fclebilirlik odakl\u0131 t\u00fcketicilerin oldu\u011fu bir d\u00fcnyada, yeni nesil envanter sistemlerine ihtiya\u00e7 oldu\u011fu fikrinden hareketle, ticarette envanter kararlar\u0131 i\u00e7in yapay zeka destekli bir tahmin yaz\u0131l\u0131m\u0131 geli\u015ftiriyor.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Platform, dahili (envanter ve sat\u0131\u015flar gibi) ve harici verilerin (sosyal medya e\u011filimleri ve hava durumu tahminleri gibi) bir\u00e7ok farkl\u0131 veriyi ger\u00e7ek zamanl\u0131 mant\u0131kla analiz ederek hangi \u00fcr\u00fcnlerin ne zaman ne miktarda stoklanaca\u011f\u0131na karar vermelerini sa\u011fl\u0131yor. <\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Ger\u00e7ek zamanl\u0131 mant\u0131kla olarak daha fazla veriden yararlanarak, sat\u0131\u015f ve planlama ekiplerine b\u00fcy\u00fck fayda sa\u011fl\u0131yor. Platform sayesinde uzun vadede de\u011fer zincirindeki her ad\u0131mda katma de\u011fer sa\u011flayarak israf\u0131 tamamen ortada kald\u0131rmak hedefleniyor.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Yapay zeka, \u015f\u00fcphesiz k\u00f6t\u00fc geli\u015fmelere de yol a\u00e7abiliyor. Yanl\u0131\u015f y\u00fcz tan\u0131ma, yanl\u0131 \u00f6\u011frenen algoritmalar ve yapay zeka destekli deep-fake sistemler, yapay zekan\u0131n karanl\u0131k y\u00fcz\u00fcne \u00f6rnek olarak verilebilir.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Genel olarak incelendi\u011finde, t\u00fcm bu geli\u015fmeler \u0131\u015f\u0131\u011f\u0131nda ara\u015ft\u0131rmac\u0131lar, GPT-3 ve Radford gibi yapay zeka modellerindeki son geli\u015fmeleri \u2018paradigma de\u011fi\u015fimi\u2019 olarak adland\u0131r\u0131yor. Bir\u00e7ok uzman, teknolojide tekilli\u011fin (singularity) d\u00fc\u015f\u00fcn\u00fcld\u00fc\u011f\u00fc kadar uzak olmayabilece\u011fini s\u00f6yl\u00fcyor. Yak\u0131n gelecekte yapay zekan\u0131n soyut d\u00fc\u015f\u00fcn\u00fcp d\u00fc\u015f\u00fcnemeyece\u011fi, duygular\u0131 hissedip hissedemeyece\u011fi veya ger\u00e7ekten kendinin fark\u0131nda olup olmayaca\u011f\u0131 gibi sorular\u0131n cevab\u0131n\u0131 hep birlikte g\u00f6rece\u011fiz.<\/span><\/p>  \t\t\t\t","slug":"yapay-zekada-korkutucu-bir-asamaya-dogru","tags":"K\u00f6\u015fe Yaz\u0131s\u0131","meta_title":"Yapay zekada korkutucu bir a\u015famaya do\u011fru","meta_description":"DO\u00c7. DR. ADNAN ERTEMEL","meta_keywords":"K\u00f6\u015fe Yaz\u0131s\u0131","news_cover_min":null,"news_cover":null,"news_video_min":null,"news_video":null,"view_count":2360,"cropped_1200x675":null,"user":{"id":27,"name":"ADNAN VEYSEL","surname":"ERTEMEL","email":"doc-dr-adnan-veysel-ertemel@gmail.com","slug":"doc-dr-adnan-veysel-ertemel","avatar":"\/front\/uploads\/avatar\/17336916001r1LpO2XUqa1CPY.webp","status":1,"role":1,"email_verified_at":null,"orde":null,"created_at":"2022-10-24T10:50:02.000000Z","updated_at":"2024-12-18T09:48:24.000000Z","seo_title":null,"seo_description":null},"translations":[{"id":35579,"is_featured":0,"is_amp":0,"is_ads":0,"ads_link":null,"post_id":35480,"locale":"tr","category_id":73,"title":"Yapay zekada korkutucu bir a\u015famaya do\u011fru","home_title":null,"sub_title":null,"slider_title":null,"slider_title_2":null,"slider_spot_title":null,"slider_spot_title_2":null,"subtitleuse":0,"description":null,"content":"  \t\t\t\t\t  \t\t\t\t\t<p style=\"text-align: right;\"><span class=\"large\"><strong>DO\u00c7. DR. ADNAN ERTEMEL<\/strong><\/span><br><span class=\"large\"> <strong><\/strong><\/span><\/p>  <p><span class=\"large\">Yapay zeka teknolojisi, onlarca y\u0131l\u0131 \u00f6\u011frenme evresinde ge\u00e7irdi. \u015eimdi olgunla\u015f\u0131yor ve \u00f6l\u00e7ekleniyor. \u00d6yle ki, tahmin edilenden de h\u0131zl\u0131 bi\u00e7imde h\u0131zlanmaya ba\u015flad\u0131.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Sentetik verilerden beslenen ve kendi tasar\u0131mc\u0131lar\u0131n\u0131n bile akl\u0131n\u0131 ba\u015f\u0131ndan alan, kendi kendini denetleyen bir yapay zeka \u00e7a\u011f\u0131na girdik. Google\u2019a ait teknoloji laboratuvar\u0131 DeepMind\u2019deki ara\u015ft\u0131rmac\u0131lar, yak\u0131n zamanda piyasaya s\u00fcr\u00fclen AlphaTensor sisteminin, algoritmalar\u0131 bile kendi ba\u015f\u0131na ke\u015ffederek \u2018insan sezgisi alan\u0131n\u0131 a\u015ft\u0131\u011f\u0131n\u0131\u2019 s\u00f6yl\u00fcyor. Nature dergisinde ekim ay\u0131 ba\u015f\u0131nda yay\u0131nlanan makalede, DeepMind ara\u015ft\u0131rmac\u0131lar\u0131, derin peki\u015ftirmeli \u00f6\u011frenme yakla\u015f\u0131m\u0131yla \u00e7al\u0131\u015fan yapay zekan\u0131n matematikte bilim adamlar\u0131n\u0131n 50 y\u0131ld\u0131r benimsedi\u011fi matris \u00e7arp\u0131m algoritmalar\u0131ndan daha h\u0131zl\u0131 \u00e7al\u0131\u015fan bir \u00e7arp\u0131m algoritmas\u0131n\u0131 ke\u015ffetti\u011fini ortaya koydu. <\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> \u00d6rne\u011fin makina \u00f6\u011frenme mant\u0131\u011f\u0131nda \u00e7al\u0131\u015fan ve en geli\u015fmi\u015f yapay zeka sistemi olarak g\u00f6sterilen GPT-3, 175 milyar parameter ile \u00e7al\u0131\u015farak, kendi kendine \u00f6\u011frenme yetisi kazanm\u0131\u015f durumda. Bu sayede d\u00fcnyada \u015fu ana kadar kaleme al\u0131nm\u0131\u015f hemen t\u00fcm kitaplar\u0131 ve Wikipedia\u2019daki t\u00fcm i\u00e7eri\u011fi haf\u0131zas\u0131na alan GPT-3, ba\u015flang\u0131\u00e7ta cevaplamas\u0131n\u0131n \u00f6ng\u00f6r\u00fclmedi\u011fi sorular\u0131 bile cevaplamaya ba\u015flad\u0131.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Yapay zeka teknolojisinin geldi\u011fi bu yeni a\u015famada s\u0131radaki hedef, yapay zekay\u0131 farkl\u0131 sekt\u00f6rlerde \u00f6l\u00e7ekleme safhas\u0131\u2026 Bunun i\u00e7in de tek eksik, daha fazla miktarda veri setleriyle yapay zekay\u0131 beslemek. Bunun i\u00e7in geli\u015ftirilen bir y\u00f6ntem, ger\u00e7ek olmayan sentetik veri \u00fcretmek.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> <strong>SENTET\u0130K VER\u0130<\/strong><\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Ger\u00e7ek d\u00fcnyadaki olaylar arac\u0131l\u0131\u011f\u0131yla toplanmak yerine algoritmik olarak olu\u015fturulan veri olarak tan\u0131mlanan sentetik veri, yaz\u0131l\u0131m geli\u015ftirmeyi iyile\u015ftirmek, <\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Ar-Ge\u2019yi h\u0131zland\u0131rmak, makina \u00f6\u011frenimi modellerini e\u011fitmek, kendi dahili verilerini ve \u00fcr\u00fcnlerini daha iyi anlamak ve i\u015f s\u00fcre\u00e7lerini iyile\u015ftirmek i\u00e7in kullan\u0131l\u0131r. Sentetik veriler, ger\u00e7ek veri k\u00fcmelerinin yerini alabilir ve matematiksel modelleri do\u011frulamak i\u00e7in kullan\u0131labilir. Sa\u011fl\u0131k, finans, sigorta, siber g\u00fcvenlik, \u00fcretim, robotik ve otonom ara\u00e7lar gibi alanlardaki \u015firketlerin daha h\u0131zl\u0131 \u00f6l\u00e7eklenebilmeleri i\u00e7in geli\u015ftirmeyi ve pazara sunma s\u00fcresini h\u0131zland\u0131rmak i\u00e7in sentetik verilerin kullan\u0131m\u0131 b\u00fcy\u00fck avantaj sa\u011fl\u0131yor.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> <strong>\u0130Y\u0130 VE K\u00d6T\u00dc \u00d6RNEKLER<\/strong><\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Yapay zeka, ge\u00e7mi\u015fte tahmin bile edilemeyen bi\u00e7imde farkl\u0131 alanlarda sekt\u00f6rlerde ciddi katma de\u011fer olu\u015fturmaya ba\u015flad\u0131. \u00d6rne\u011fin, ABD\u2019de kurulan Syrup Tech adl\u0131 \u015firket, derin tedarik zinciri kesintileri, ger\u00e7ek \u00e7ok kanall\u0131 a\u011flar ve s\u00fcrd\u00fcr\u00fclebilirlik odakl\u0131 t\u00fcketicilerin oldu\u011fu bir d\u00fcnyada, yeni nesil envanter sistemlerine ihtiya\u00e7 oldu\u011fu fikrinden hareketle, ticarette envanter kararlar\u0131 i\u00e7in yapay zeka destekli bir tahmin yaz\u0131l\u0131m\u0131 geli\u015ftiriyor.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Platform, dahili (envanter ve sat\u0131\u015flar gibi) ve harici verilerin (sosyal medya e\u011filimleri ve hava durumu tahminleri gibi) bir\u00e7ok farkl\u0131 veriyi ger\u00e7ek zamanl\u0131 mant\u0131kla analiz ederek hangi \u00fcr\u00fcnlerin ne zaman ne miktarda stoklanaca\u011f\u0131na karar vermelerini sa\u011fl\u0131yor. <\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Ger\u00e7ek zamanl\u0131 mant\u0131kla olarak daha fazla veriden yararlanarak, sat\u0131\u015f ve planlama ekiplerine b\u00fcy\u00fck fayda sa\u011fl\u0131yor. Platform sayesinde uzun vadede de\u011fer zincirindeki her ad\u0131mda katma de\u011fer sa\u011flayarak israf\u0131 tamamen ortada kald\u0131rmak hedefleniyor.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Yapay zeka, \u015f\u00fcphesiz k\u00f6t\u00fc geli\u015fmelere de yol a\u00e7abiliyor. Yanl\u0131\u015f y\u00fcz tan\u0131ma, yanl\u0131 \u00f6\u011frenen algoritmalar ve yapay zeka destekli deep-fake sistemler, yapay zekan\u0131n karanl\u0131k y\u00fcz\u00fcne \u00f6rnek olarak verilebilir.<\/span><br><span class=\"large\"><\/span><br><span class=\"large\"> Genel olarak incelendi\u011finde, t\u00fcm bu geli\u015fmeler \u0131\u015f\u0131\u011f\u0131nda ara\u015ft\u0131rmac\u0131lar, GPT-3 ve Radford gibi yapay zeka modellerindeki son geli\u015fmeleri \u2018paradigma de\u011fi\u015fimi\u2019 olarak adland\u0131r\u0131yor. Bir\u00e7ok uzman, teknolojide tekilli\u011fin (singularity) d\u00fc\u015f\u00fcn\u00fcld\u00fc\u011f\u00fc kadar uzak olmayabilece\u011fini s\u00f6yl\u00fcyor. Yak\u0131n gelecekte yapay zekan\u0131n soyut d\u00fc\u015f\u00fcn\u00fcp d\u00fc\u015f\u00fcnemeyece\u011fi, duygular\u0131 hissedip hissedemeyece\u011fi veya ger\u00e7ekten kendinin fark\u0131nda olup olmayaca\u011f\u0131 gibi sorular\u0131n cevab\u0131n\u0131 hep birlikte g\u00f6rece\u011fiz.<\/span><\/p>  \t\t\t\t","slug":"yapay-zekada-korkutucu-bir-asamaya-dogru","orjinalimage":null,"news_cover_min":null,"news_cover":null,"news_video_min":null,"news_video":null,"cropped_638x552":null,"cropped_310x208":null,"cropped_416x247":null,"cropped_197x247":null,"cropped_416x600":null,"cropped_1200x675":null,"tags":"K\u00f6\u015fe Yaz\u0131s\u0131","meta_title":"Yapay zekada korkutucu bir a\u015famaya do\u011fru","meta_description":"DO\u00c7. DR. ADNAN ERTEMEL","meta_keywords":"K\u00f6\u015fe Yaz\u0131s\u0131","view_count":2360,"yt":0,"ytid":"","ytimage":null,"imgdate":"2000-01-01 00:00:00","cuff_cover":null,"cropped_358x214":null,"cropped_842x474":null}]}}