Bilgisayarlar \u2018data<\/em>\u2019lar\u0131 belle\u011finde tutan, istendi\u011finde geri getiren ve seri hesaplamalar yapan cihazlard\u0131r. Yapay zek\u00e2lar ise insan beyninin alg\u0131, tan\u0131ma, \u00f6\u011frenme, problem \u00e7\u00f6zme, d\u00fc\u015f\u00fcnme, muhakeme ve karar yetilerini taklit etmeye \u00e7al\u0131\u015fan ak\u0131ll\u0131 bilgisayarlard\u0131r. Yapay zek\u00e2lar\u0131n (YZ), y\u00fcklenen verilerle s\u0131n\u0131rl\u0131 \u00e7ok geni\u015f belle\u011fi yan\u0131nda, soru yan\u0131tlama, bilgi e\u015fle\u015ftirme, analiz yapma, mant\u0131k y\u00fcr\u00fctme, rapor haz\u0131rlama, cihaz y\u00f6netme gibi becerileri vard\u0131r. \u0130leri hedef; d\u00fc\u015f\u00fcnen, karar veren ve otonom davranan g\u00fcvenilir YZ\u2019lar \u00fcretmektir\u2026<\/span><\/p>\n Belki farkl\u0131 parametrelerle yap\u0131lacak \u00f6l\u00e7\u00fcmlerde de\u011fi\u015fik sonu\u00e7lar \u00e7\u0131kabilir, ama bug\u00fcn (insanlara g\u00f6re tasarlanm\u0131\u015f) IQ testlerinde yapay zek\u00e2lar\u0131n performans\u0131 san\u0131ld\u0131\u011f\u0131 kadar y\u00fcksek de\u011fildir; 100\u2019\u00fcn alt\u0131nda veya biraz \u00fcst\u00fcnde (114 gibi) skorlar al\u0131nmaktad\u0131r (1). Yapay zek\u00e2 say\u0131lmayan, ama s\u00f6zel mesajlar\u0131 anlay\u0131p uygun yan\u0131tlar verebilen Siri, Assistant, Cortana ve Alexa gibi sesli asistanlar\u0131n IQ\u2019lar\u0131 daha da d\u00fc\u015f\u00fck (Siri: 24, Google Assistant: 47) bulunmu\u015ftur (2); bu skorlar\u0131n zaman i\u00e7inde y\u00fckselmesi olas\u0131d\u0131r. GhatGPT 4\u2019un IQ\u2019su 147-155 aras\u0131 iken, bunun yak\u0131n gelecekte 1.600\u2019l\u00fc rakamlara, hatta 10.000\u2019e \u00e7\u0131kaca\u011f\u0131 ileri s\u00fcr\u00fclse de insani olmayan, ayk\u0131r\u0131 \u00f6zelliklerinin daima sorun yarataca\u011f\u0131 \u00f6ng\u00f6r\u00fclmektedir\u2026<\/p>\n Olumsuz ve Yetersiz Y\u00f6nler<\/strong><\/p>\n – Yapay zek\u00e2lar bilin\u00e7li de\u011fildir: Kendini ve \u00e7evresini alg\u0131lasa da hissedemez, bilin\u00e7li fark\u0131ndal\u0131\u011f\u0131\u00a0(mindfulness)<\/em>\u00a0yoktur.<\/p>\n – D\u00fcz mant\u0131kla ak\u0131l y\u00fcr\u00fct\u00fcr; soyutlamalar\u0131, deyimleri, imalar\u0131, ironileri, hileleri ve k\u00fclt\u00fcrel kodlar\u0131 anlayamaz. \u2018\u0130nsan gibi\u2019 d\u00fc\u015f\u00fcnemez, hayal kuramaz; yarat\u0131c\u0131-yenilik\u00e7i-insani-etik \u00e7\u00f6z\u00fcmler \u00fcretemez. Uyum becerisi d\u00fc\u015f\u00fckt\u00fcr…<\/p>\n – \u2018Sokma ak\u0131ll\u0131\u2019d\u0131r; belle\u011fine ne y\u00fcklendiyse ve nas\u0131l programland\u0131ysa o kadard\u0131r. Yanl\u0131\u015f bilgi y\u00fcklenmesi yanl\u0131\u015f yapt\u0131r\u0131r; \u00e7\u00f6p y\u00fcklenirse \u00e7\u00f6p verir\u00a0(Garbage in, garbage out).<\/em><\/p>\n – Duygusall\u0131\u011f\u0131; merhameti, \u015fefkati, vicdan\u0131 yoktur. Sevmez, sevinmez, \u00fcz\u00fclmez, kayg\u0131lanmaz, a\u011flamaz, korkmaz; mahremiyet, s\u0131r, ahl\u00e2k, tevazu bilmez. \u2018Buz gibi\u2019dir; duygulara k\u00f6r ve sa\u011f\u0131rd\u0131r (aleksitimi).<\/p>\n – Empati yapamaz; duygusal zek\u00e2s\u0131 (EQ), duygusal \u00e7evikli\u011fi\u00a0(emotional agility)\u00a0<\/em>yoktur.<\/p>\n – YZ\u2019lar zeki, mant\u0131kl\u0131 ve pragmatiktir, ama ak\u0131ll\u0131 ve bilge de\u011fildir. Bilgi\u00e7lik taslar; ukal\u00e2, kibirli ve malumatfuru\u015ftur. Akl\u0131selimi, feraseti, sezgisi, ileri g\u00f6r\u00fc\u015f\u00fc, edebi, erdemi\u00a0(wisdom)<\/em>\u00a0yoktur.<\/p>\n Askeri Havac\u0131l\u0131kta YZ\u2019lar<\/strong><\/p>\n Havac\u0131l\u0131\u011fa YZ\u2019l\u0131 teknolojilerin girmesiyle b\u00fcy\u00fck bir d\u00f6n\u00fc\u015f\u00fcm ya\u015fanmaktad\u0131r. Askeri havac\u0131l\u0131kta sofistike sava\u015f u\u00e7a\u011f\u0131 pilotlar\u0131n\u0131n yapt\u0131\u011f\u0131 g\u00f6revlerin yapay zek\u00e2l\u0131 otonom bilgisayarlara y\u00fcklenmesi, insan unsuruna ba\u011fl\u0131 hatalar\u0131, kazalar\u0131 ve pilotlar\u0131n can kayb\u0131 olas\u0131l\u0131\u011f\u0131n\u0131 \u00e7ok azaltacakt\u0131r. \u0130nsanl\u0131 u\u00e7ak kazalar\u0131nda \u2018insan fakt\u00f6r\u00fc\u2019 (\u0130F) oran\u0131 %80 iken, kokpitte pilotlar\u0131n yerine (veya yan\u0131na) YZ\u2019lar\u0131n girmesiyle \u0130F s\u0131f\u0131rlanmas\u0131 hayal de\u011fildir. Yak\u0131n gelecekte ABD\u2019de \u00e7al\u0131\u015fan 140 milyon beyin i\u015f\u00e7isinin pozisyonlar\u0131n\u0131 YZ\u2019l\u0131 bilgisayarlara kapt\u0131raca\u011f\u0131; bunlar aras\u0131nda hukuk\u00e7u, doktor ve pilotlar\u0131n\u00a0(robo-lawyer, robo-doc, robo-pilot)\u00a0<\/em>da olaca\u011f\u0131 \u00f6ng\u00f6r\u00fcl\u00fcyor.<\/p>\n 5\u2019inci Nesil Sava\u015f U\u00e7aklar\u0131<\/strong><\/p>\n Bug\u00fcn \u00e7ok say\u0131da otomasyon ve yapay zek\u00e2 sistemlerinin kompakt bi\u00e7imde yerle\u015ftirildi\u011fi 5\u2019inci nesil F-35\u2019lerin \u2018u\u00e7an s\u00fcper bilgisayarlar\u2019 oldu\u011fu s\u00f6ylenmektedir. D\u00fc\u015fman u\u00e7aklar\u0131n\u0131 uzaktan alg\u0131lama ve \u00f6nleme, u\u00e7aksavarlar\u0131 bulma, d\u00fc\u015fman hedeflerine f\u00fczeler g\u00f6nderme, radar kar\u0131\u015ft\u0131rmas\u0131 yapma, en uygun rotay\u0131 belirleme gibi sofistike \u00f6zellikler giderek geli\u015ftirilmekte ve bu teknoloji sayesinde belirgin hava \u00fcst\u00fcnl\u00fc\u011f\u00fc sa\u011flanmaktad\u0131r. Bir F-35\u2019in donan\u0131m\u0131 300 bin par\u00e7a iken, yaz\u0131l\u0131m\u0131ndaki bilgisayar kodu say\u0131s\u0131 8 milyondur. U\u00e7ak d\u0131\u015f\u0131ndan destek veren 16 milyon eklendi\u011finde, toplam 24 milyon sat\u0131r kod ile u\u00e7maktad\u0131r…<\/p>\n Ruslar\u0131n Su-35 ve Su-57 Felon u\u00e7aklar\u0131n\u0131n da benzer YZ sistemleriyle donat\u0131lm\u0131\u015f oldu\u011fu; \u00c7in\u2019in J-16 u\u00e7a\u011f\u0131n\u0131n YZ versiyonlu modelinin test u\u00e7u\u015flar\u0131 yapt\u0131\u011f\u0131, 5\u2019inci nesil J-20\u2019sinin de \u00fcretimde oldu\u011fu bilinmektedir (3). Avrupa cephesinde ise \u0130ngiltere, \u0130talya ve Japonya\u2019n\u0131n 2035 y\u0131l\u0131na kadar YZ g\u00fcd\u00fcml\u00fc sava\u015f u\u00e7a\u011f\u0131 \u00fcretmek \u00fczere anla\u015ft\u0131klar\u0131 a\u00e7\u0131klanm\u0131\u015ft\u0131r. T\u00fcrkiye\u2019nin \u00fcretim \u00e7al\u0131\u015fmalar\u0131 ve deneme u\u00e7u\u015flar\u0131 devam eden MMU Kaan da YZ ile donat\u0131lm\u0131\u015f ilk 5\u2019inci nesil T\u00fcrk sava\u015f u\u00e7a\u011f\u0131 olacakt\u0131r. Ayr\u0131ca yapay zek\u00e2l\u0131 silahl\u0131 insans\u0131z hava ara\u00e7lar\u0131 (S\u0130HA, T\u0130HA) olarak Ak\u0131nc\u0131, TB-3; Muharip \u0130nsans\u0131z U\u00e7ak Sistemleri olarak da K\u0131z\u0131lelma, Anka-3, \u00fclkemizin askeri havac\u0131l\u0131\u011f\u0131nda b\u00fcy\u00fck ilerleme kaydetmi\u015ftir.<\/p>\n 6\u2019nc\u0131 Nesil \u0130nsans\u0131z U\u00e7aklar<\/strong><\/p>\n Yapay zek\u00e2larla y\u00f6netilen pilotsuz 6\u2019nc\u0131 nesil u\u00e7aklarda fiziksel, psikolojik, bili\u015fsel zafiyetler olmayacakt\u0131r. Pilot olmay\u0131nca fi<\/a>zyolojik inkapasitasyonlar (hipoksi, y\u00fcksek G kuvvetine ba\u011fl\u0131 bilin\u00e7 kayb\u0131-GLoc, vestib\u00fcler ve g\u00f6rsel ill\u00fczyon, dekompresyon, vs.) da olmayacakt\u0131r. Ya\u015fam destek sistemlerinin (oksijen, bas\u0131n\u00e7lama, atlama, vs.) \u00e7\u0131kar\u0131lmas\u0131yla u\u00e7aklar hafifleyecektir. Kokpitlerde koltuk ekipmanlar\u0131, g\u00f6stergeler, mekanik kumandalar, vs. \u00e7\u0131kar\u0131l\u0131p t\u00fcm i\u015flevlerin YZ\u2019lara b\u0131rak\u0131lmas\u0131yla, u\u00e7u\u015f g\u00f6revleri insan elinden ve akl\u0131ndan daha s\u00fcratli ve do\u011fru bi\u00e7imde yap\u0131labilecek olmas\u0131 \u00e7ok \u00f6nemli bir devrim niteli\u011findedir…<\/p>\n YZ\u2019li robot sava\u015f u\u00e7aklar\u0131, m\u00fckemmel yaz\u0131l\u0131mlar\u0131 sayesinde ke\u015fif, g\u00f6zetleme, tehdit alg\u0131lama, hava-hava ve hava-yer at\u0131\u015flar\u0131 yapma gibi becerileriyle \u00f6ne \u00e7\u0131kmaktad\u0131r. ABD Hava Kuvvetlerine F-16\u2019lar \u00fcreten Lockheed Martin\u2019in \u2018Skunk Works\u2019 b\u00f6l\u00fcm\u00fc, 2023 y\u0131l\u0131nda F-16\u2019n\u0131n VISTA X-62A isimli test mod\u00fcl\u00fc ile 17 saatlik u\u00e7u\u015fu YZ ile ger\u00e7ekle\u015ftirmi\u015fti. 2024 y\u0131l\u0131nda ise bir insanl\u0131 F-16 ile it dala\u015f\u0131 yapan Vista F-16 kadar ba\u015far\u0131l\u0131 de\u011fildi; ama s\u00fcratle geli\u015ftirilip \u00f6ne ge\u00e7ece\u011fi tahmin edilmektedir. Zaten ABD Hava Kuvvetleri binden fazla YZ\u2019li insans\u0131z sava\u015f u\u00e7a\u011f\u0131ndan olu\u015fan bir filonun 2028 y\u0131l\u0131na kadar faaliyete ge\u00e7ece\u011fini duyurmu\u015f bulunmaktad\u0131r… Di\u011fer taraftan \u00c7in\u2019de iki k\u00fc\u00e7\u00fck sabit kanatl\u0131 u\u00e7a\u011f\u0131n it dala\u015f\u0131nda YZ\u2019li u\u00e7a\u011f\u0131n insanl\u0131 u\u00e7a\u011f\u0131 90 saniye i\u00e7inde yendi\u011fi a\u00e7\u0131kland\u0131.\u00a0\u2018Acta Aeronautica et Astronautica Sinica\u2019<\/em>\u00a0dergisinde 27 \u015eubat 2023 g\u00fcn\u00fc yay\u0131nlanan makalede; \u2018yapay zek\u00e2lar\u0131n \u2018kral olaca\u011f\u0131\u2019 hava sava\u015flar\u0131n\u0131n ufukta g\u00f6r\u00fcnd\u00fc\u011f\u00fc\u2019 yaz\u0131ld\u0131 (4).<\/p>\n Gelecekteki Hava Sava\u015flar\u0131 Simulasyonunda YZ<\/strong><\/p>\n 2023 y\u0131l\u0131nda Londra’da d\u00fczenlenen \u2018Future Combat Air & Space Capabilities\u2019 Zirvesinde, ABD Hava K. Yapay Zek\u00e2 Operasyonlar\u0131 \u015eefi Albay Tucker Hamilton, bir testte yapay zek\u00e2n\u0131n “amac\u0131na ula\u015fmak i\u00e7in hi\u00e7 beklenmedik stratejiler” kulland\u0131\u011f\u0131n\u0131; hedefini vurmak \u00fczere iken bunu engelleyen operat\u00f6r\u00fc \u00f6ld\u00fcrmeye kalk\u0131\u015ft\u0131\u011f\u0131n\u0131 a\u00e7\u0131klam\u0131\u015ft\u0131. Olay\u0131n simulat\u00f6r operasyonunda ge\u00e7mesi nedeniyle \u00f6len olmamas\u0131 iyiydi; ama potansiyel tehlikelerin ipu\u00e7lar\u0131 ortaya \u00e7\u0131km\u0131\u015ft\u0131 (5).<\/p>\n Askeri Hava Savunma Sistemleri<\/strong><\/p>\n Hava kalkan\u0131 benzeri sistemler y\u00fczlerce mil \u00f6telerden ate\u015flenen f\u00fcze, roket, S\u0130HA ve u\u00e7aklar\u0131 uydu ve radarlar arac\u0131l\u0131\u011f\u0131yla an\u0131nda alg\u0131lamakta; YZ kullanarak y\u00f6n ve h\u0131zlar\u0131n\u0131 tespit edip \u00f6nleme yapmaktad\u0131r. Ruslar\u0131n yapay zek\u00e2l\u0131 Pantsir-S1 robotik Anti-\u0130HA\u2019lar\u0131 otonom \u00e7al\u0131\u015fmakla birlikte, hedefleri imhada son karar\u0131 yerdeki yetkililerin iznine b\u0131rakmaktad\u0131r. Baz\u0131 \u00fclkelerin kulland\u0131klar\u0131, T\u00fcrkiye\u2019nin de geli\u015ftirmekte oldu\u011fu \u2018\u00c7elik Kubbe\u2019 projesi bunun son \u00f6rne\u011fidir.<\/p>\n Sivil Havac\u0131l\u0131kta YZ\u2019lar<\/strong><\/p>\n Sivil havac\u0131l\u0131kta uzun s\u00fcredir kullan\u0131lmakta olan otomasyon sistemleri u\u00e7a\u011f\u0131n emniyetle kalk\u0131\u015f, ini\u015f ve d\u00fcz u\u00e7u\u015funu yapt\u0131rmaktad\u0131r. YZ\u2019lar beklenmeyen kritik durumlarda \u0131\u015f\u0131kl\u0131 ve sesli mesajlarla pilotlar\u0131 uyarmakta; bas\u0131n\u00e7 azalmas\u0131, hidrolik ar\u0131zas\u0131, yang\u0131n, \u00e7arp\u0131\u015fma (TCAS), vb. durumlarda kumandalara m\u00fcdahale edebilmektedir.<\/p>\n ABD\u2019de uzaktan kumandal\u0131 sistemler geli\u015ftiren \u2018Reliable Robotics\u2019 \u015firketi, 2023 y\u0131l\u0131nda test \u00e7al\u0131\u015fmalar\u0131 i\u00e7in FAA\u2019den onay ald\u0131. Otopilot sistemiyle donat\u0131lan Cessna Caravan tipi bir insans\u0131z u\u00e7ak, Kaliforniya\u2019daki Hollister Havaalan\u0131nda 12 dakikal\u0131k u\u00e7u\u015f s\u00fcrecinde taksi yapt\u0131, havaland\u0131 ve g\u00fcvenli bir \u015fekilde indi… Yang\u0131n s\u00f6nd\u00fcrme, trafik denetleme, zirai ila\u00e7lama, kargo, hava taksi, vs. hizmetler insans\u0131z YZ\u2019li hava ara\u00e7lar\u0131yla yap\u0131l\u0131r hale gelmi\u015ftir.<\/p>\n U\u00e7ak Kokpitlerinde Robo-pilotlar: 2023 y\u0131l\u0131nda G\u00fcney Kore \u2018\u0130leri Bilim ve Teknoloji Enstit\u00fcs\u00fc\u2019 (KAIST), kokpitte \u00f6zel bir de\u011fi\u015fiklik yap\u0131lmadan bir u\u00e7a\u011f\u0131 u\u00e7urabilecek humanoid robo-pilot (PiBot) geli\u015ftirildi\u011fini a\u00e7\u0131klam\u0131\u015ft\u0131. Farkl\u0131 u\u00e7ak konfig\u00fcrasyonlar\u0131 y\u00fcklendi\u011finde PiBot\u2019un, hangi tip se\u00e7ilirse o u\u00e7a\u011f\u0131 u\u00e7urulabildi\u011fi, ilgili k\u0131lavuzlar\u0131, referans materyalleri ve Jeppesen \u00e7izelgelerinin t\u00fcm\u00fcn\u00fc belle\u011finde saklayabildi\u011fi bildirildi (6<\/strong>).<\/p>\n 2035 y\u0131l\u0131na do\u011fru u\u00e7aklarda sol koltukta kaptan otururken, sa\u011fda yapay zek\u00e2l\u0131 bir sistemin yer alaca\u011f\u0131 \u00f6ng\u00f6r\u00fclmektedir. Bunlar\u0131n prototipleri Cessna, Boeing-737 gibi u\u00e7aklarda test edilmektedir. Bellek kapasitesi, fiziksel hesaplama, fark\u0131ndal\u0131k, \u00f6ng\u00f6r\u00fc, \u00e7oklu g\u00f6rev, bilgileri mukayeseli de\u011ferlendirme gibi \u00fcst\u00fcn performanslar\u0131yla insandan daha zeki olabilen YZ\u2019lar gene de insandan daha \u2018ak\u0131ll\u0131\u2019 olabilmi\u015f de\u011fildir. Yapay zek\u00e2l\u0131 sistemlerin bazen umulmad\u0131k hatalar yap\u0131p tehlikeli olabilecekleri ihtimali nedeniyle, u\u00e7ak i\u00e7inde her an m\u00fcdahale edecek ve kritik durumlarda son karar\u0131 verecek bir insan\u0131n bulunmas\u0131 \u015fartt\u0131r.<\/p>\n \u0130ki Ba\u015fl\u0131l\u0131k: \u0130ronik olan \u015fudur ki; pilotlar\u0131n kokpitteki rutin i\u015flerini rahatlats\u0131n, i\u015fy\u00fck\u00fcn\u00fc azalts\u0131n ve kumandalar\u0131 standardize etsin diye konulmu\u015f olan otomasyon sistemleri, bazen yetki ve \u00f6ncelik karma\u015fas\u0131 sonucu iki ba\u015fl\u0131l\u0131k\u00a0(coupling)<\/em>\u00a0nedeniyle i\u015fleri fazlaca kar\u0131\u015ft\u0131rabiliyor. Otopilot (havac\u0131l\u0131k jargonunda\u00a0\u2018George<\/em>\u2019) devrede iken, hangi kontrollerin kendisinde oldu\u011fu konusunda ikileme d\u00fc\u015febilen pilotlar, sistemin o durumda ne yapaca\u011f\u0131n\u0131 bilemiyor, bazen de otopilot, pilotun kumandas\u0131na itaat etmeyebiliyor. Baz\u0131 pilotlar\u0131n, kendilerine yard\u0131m edeceklerini sand\u0131klar\u0131 otomasyonun \u00e7ok karma\u015f\u0131k oldu\u011fundan ve (umulan\u0131n tersine) i\u015f y\u00fcklerini \u00e7ok artt\u0131rd\u0131\u011f\u0131ndan \u015fik\u00e2yet etmekte olmalar\u0131, \u00f6nemli bir soruna dikkat \u00e7ekiyor. Ba\u015fka bir zorluk da, bu sistemlere al\u0131\u015fan pilotlar\u0131n ar\u0131za veya ba\u015fka nedenle manuel sisteme ge\u00e7tiklerinde yava\u015f ve hatal\u0131 kumanda veriyor olmalar\u0131d\u0131r (7<\/strong>). Yani otomasyon sistemleri pilotu tembelli\u011fe al\u0131\u015ft\u0131rabilir; bir butona basarak devreye ald\u0131\u011f\u0131 sistemin kendisinden daha do\u011fru karar verece\u011fine g\u00fcvenerek rahatl\u0131k i\u00e7inde\u00a0(complacency)<\/em>\u00a0u\u00e7u\u015fu izleme kolayc\u0131l\u0131\u011f\u0131na d\u00fc\u015f\u00fcr\u00fcyor olabilir\u2026<\/p>\n YZ\u2019lar\u0131n CRM becerileri: U\u00e7u\u015f ekibi kaynaklar\u0131 y\u00f6netimi ba\u011flam\u0131nda d\u00fcz bak\u0131\u015fla YZ\u2019lardan daha iyisi olamaz. Fark\u0131ndal\u0131klar\u0131 \u00e7ok y\u00fcksektir, her \u015feyi hemen alg\u0131layabilir, y\u00fczlerce parametreyi de\u011ferlendirip uygun karar ve do\u011fru eylemi belirleyebilir. \u0130\u015f b\u00f6l\u00fcm\u00fcn\u00fc ve tak\u0131m \u00e7al\u0131\u015fmas\u0131n\u0131 pl\u00e2nlayabilir. Belle\u011findeki acil durum senaryolar\u0131na g\u00f6re \u00f6nlemleri eksiksiz uygulayabilir\u2026<\/p>\n Ancak teknik d\u0131\u015f\u0131 beceriler\u00a0(Notechs)<\/em>\u00a0YZ\u2019lar\u0131n en zay\u0131f yan\u0131d\u0131r. \u0130nsanlarla birlikte g\u00f6rev yap\u0131lan kritik i\u015flerde tak\u0131m \u00e7al\u0131\u015fmas\u0131n\u0131 y\u00f6neten bir YZ, insanlar\u0131n psikolojisini, ili\u015fki sorunlar\u0131n\u0131, ekip dinamiklerini, motivasyonlar\u0131n\u0131 bilemez, empati kuramaz. Yorgunluk, uyku, yeme-i\u00e7me gibi ihtiya\u00e7lar\u0131n\u0131, heyecan, korku ve h\u0131rslar\u0131n\u0131 anlayamaz. Bu nedenlerle YZ\u2019lar rutin i\u015flerde ba\u015far\u0131l\u0131 olabilir, ama ekip y\u00f6netimi, liderlik ve yarat\u0131c\u0131 \u00e7\u00f6z\u00fcmler, onlar\u0131n yetersiz kald\u0131klar\u0131 alanlard\u0131r.<\/p>\n Yapay Zek\u00e2lar\u0131n Karar Verici Olarak Kullan\u0131m\u0131<\/strong><\/p>\n Pilotun u\u00e7u\u015fta dikkatini da\u011f\u0131t\u0131c\u0131 unsurlar, fizyolojik zorlanmalar ve psikolojik kayg\u0131lar alt\u0131nda iken do\u011fru ve seri bi\u00e7imde karar verebilmesinde zorluklar oldu\u011fu bilinir. Hatt\u00e2 insan\u0131n hata yapmaya yatk\u0131n ve pek de g\u00fcvenilemez oldu\u011funa dair yayg\u0131n kanaatler vard\u0131r. Kaza fakt\u00f6r analizlerinde alg\u0131 \/ yarg\u0131 \/ karar hatas\u0131 unsurunun s\u0131kl\u0131kla ortaya \u00e7\u0131kmas\u0131 y\u00fcz\u00fcnden, karar verme sorumlulu\u011funu geli\u015fmi\u015f bilgisayarlara devretme d\u00fc\u015f\u00fcncesinin tarihi yeni de\u011fildir.<\/p>\n Ar\u0131za yapmad\u0131\u011f\u0131 ve \u015farj edildi\u011fi s\u00fcrece YZ\u2019lar yorulmaz, ac\u0131kmaz, uykusu gelmez, tuvalet ihtiyac\u0131 duymaz, ba\u015f\u0131 a\u011fr\u0131maz, hastalanmaz. Bili\u015fsel i\u015flevlerini kesintisiz ve tam do\u011frulukta s\u00fcrd\u00fcr\u00fcr; daima uyan\u0131k kal\u0131r, dikkati da\u011f\u0131lmaz, algoritmalar\u0131 ve acil durum \u00f6nlemlerini hi\u00e7 unutmaz, sorulara do\u011fru yan\u0131tlar verir… Korku ve panik gibi duygusal tepkileri olmaz; so\u011fukkanl\u0131l\u0131\u011f\u0131n\u0131 sonuna kadar koruyabilir; her durumda dengeli ve rasyonel davran\u0131r. Oksijensizlikten, d\u00fc\u015f\u00fck ve y\u00fcksek G kuvvetlerinden, bas\u0131n\u00e7 ve \u0131s\u0131 de\u011fi\u015fimlerinden etkilenmez. Yani t\u0131bbi, bili\u015fsel, psikolojik ve fizyolojik inkapasitasyonlara girmez\u2026<\/p>\n B\u00fct\u00fcn bu avantaj ve \u00fcst\u00fcnl\u00fcklerine kar\u015f\u0131n, kritik m\u00fcdahale ve do\u011fru karar s\u00fcre\u00e7lerinde hal\u00e2 sa\u011fl\u0131kl\u0131, dengeli ve profesyonel insanlar\u0131n kokpitte YZ\u2019lardan daha iyi olabilecekleri durumlar vard\u0131r. Bunun \u00f6rnekleri f\u00fct\u00fcristik uzay filmlerinde g\u00f6r\u00fcl\u00fcr. Gelecek \u00f6ng\u00f6r\u00fcleri y\u00fcksek bilim kurgu yazarlar\u0131n\u0131n senaryolar\u0131nda, zorlu sorunlar kar\u015f\u0131s\u0131nda YZ\u2019lar\u0131n yetersiz kald\u0131\u011f\u0131 veya fel\u00e2kete g\u00f6t\u00fcrd\u00fc\u011f\u00fc anlar vard\u0131r. Bu noktada insani dokunu\u015flar\u0131n durumu kurtard\u0131\u011f\u0131 sahneler, senaristlerin insan\u0131 kay\u0131rma g\u00fcd\u00fcleriyle a\u00e7\u0131klanamaz\u2026<\/p>\n Sons\u00f6z<\/strong><\/p>\n Kanaatimizce yapay zek\u00e2lar hatalar\u0131 s\u0131f\u0131rlayan, i\u015flemleri h\u0131zland\u0131ran ve ya\u015fam\u0131 kolayla\u015ft\u0131ran vazge\u00e7ilemez elektronik sistemlerdir. Ancak olumsuz a\u00e7\u0131dan bak\u0131ld\u0131\u011f\u0131nda, belle\u011fi \u00e7ok g\u00fc\u00e7l\u00fc, ama duygusal yan\u0131 zay\u0131f \u2018Savant Sendrom\u2019lu otistiklere, k\u0131smen de empati yapamayan ve ac\u0131mas\u0131z davranabilen sosyopatlara benzer.<\/p>\n Baz\u0131 komplo teorileri, otonomi verilen YZ\u2019lerin talimatlara ayk\u0131r\u0131 tehlikeli kararlar verebileceklerini ileri s\u00fcrse de bu olas\u0131l\u0131k, fantastik uzay filmlerinde g\u00f6r\u00fcld\u00fc\u011f\u00fcnden bile daha azd\u0131r. Bug\u00fcn i\u00e7in ar\u0131za veya s\u0131ra d\u0131\u015f\u0131 durumlar dikkate al\u0131narak, sistem i\u00e7indeki insanlar\u0131n daima devrede olmas\u0131 ve hibrid bir yap\u0131 olu\u015fturulmas\u0131 daha uygundur. Otonom \u00e7al\u0131\u015fan YZ sistemlerinde bile g\u00f6zleyen, kontrol\u00fc elinde tutan ve son karar\u0131 veren insanlar\u0131n olmas\u0131 \u00f6nerilmektedir.<\/p>\n Bir f\u0131kra<\/strong><\/p>\n * Pilotsuz u\u00e7abilen en yeni model u\u00e7a\u011f\u0131 g\u00f6rmek i\u00e7in gelen yetkili ki\u015fi, kokpitte bir pilot ile kar\u015f\u0131s\u0131nda oturan bir k\u00f6pek g\u00f6r\u00fcr ve \u015fa\u015f\u0131r\u0131r. K\u00f6pe\u011fin kokpitte ne i\u015fi oldu\u011funu sorar.<\/p>\n M\u00fchendisler k\u00f6pe\u011fin, pilotun kumandalara dokunmamas\u0131 i\u00e7in e\u011fitilmi\u015f oldu\u011funu, dokunursa \u0131s\u0131raca\u011f\u0131n\u0131 s\u00f6yler. Yetkili,\u00a0\u201cO halde pilot neden kokpitte oturuyor?\u201d\u00a0<\/em>diye sorunca,\u00a0\u201cK\u00f6pek ac\u0131kt\u0131\u011f\u0131nda onu beslemesi i\u00e7in\u201d<\/em>\u00a0derler\u2026<\/p>\n Kaynaklar<\/i><\/strong>:<\/strong><\/i><\/p>\n 1.<\/strong>\u00a0Brian Wang. Bing AI and GPT4 has an IQ of 114 and is smarter than the average human. (Nextbigfuture: Mar 15, 2023).\u00a0https:\/\/www.nextbigfuture.com\/2023\/03\/bing-ai-and-gpt4-has-an-iq-of-114-and-is-smart-than-average-human.html<\/a><\/i><\/p>\n 2.<\/strong>\u00a0Shoshanna Delventhal. Google AI or Siri: Which Has the Highest IQ? (Investopedia: June 25, 2019). https:\/\/www.investopedia.com\/news\/google-ai-or-siri-which-has-highest-iq\/<\/a><\/i><\/p>\n 3.\u00a0<\/strong>Liu Xuanzun & Fan Wei in Zhuhai. China’s J-20 fighter jet to keep receiving upgrades: designer. Global Times (Nov 07, 2022). https:\/\/www.globaltimes.cn\/page\/202211\/1278912.shtml<\/i><\/p>\n 4.<\/strong>\u00a0Independent T\u00fcrk\u00e7e, 6 Mart 2023; \u201cSouth China Morning Post, Futurism\u201d. Derleyen: \u00c7a\u011fla \u00dcren. https:\/\/www.indyturk.com\/node\/614951\/bilim\/yapay-zeka-it-dala\u015f\u0131nda-\u00e7inli-sava\u015f-pilotunu-90-saniyede-yendi<\/a><\/i><\/p>\n 5.<\/strong>\u00a0Yapay zek\u00e2n\u0131n karanl\u0131k y\u00fcz\u00fc: Operat\u00f6r\u00fcn\u00fc \u00f6ld\u00fcrmeye karar verdi. H\u00fcrriyet Gazetesi (2 Haziran 2023).\u00a0\u00a0https:\/\/www.hurriyet.com.tr\/dunya\/yapay-zekanin-karanlik-yuzu-operatorunu-oldurmeye-karar-verdi-42277408<\/a>;<\/i><\/p>\n![\"\"](\"https:\/\/www.airkule.com\/images\/image\/20240916-mcetinguc-1.jpg\")
<\/p>\n![\"\"](\"https:\/\/www.airkule.com\/images\/image\/20240916-mcetinguc-2.jpg\")
<\/p>\n![\"\"](\"https:\/\/www.airkule.com\/images\/image\/20240916-mcetinguc-3.jpg\")
<\/p>\n![\"\"](\"https:\/\/www.airkule.com\/images\/image\/20240916-mcetinguc-4.jpg\")
<\/p>\n![\"\"](\"https:\/\/www.airkule.com\/images\/image\/20240916-mcetinguc-5.jpg\")
<\/p>\n![\"\"](\"https:\/\/www.airkule.com\/images\/image\/20240916-mcetinguc-6.jpg\")
<\/p>\n![\"\"](\"https:\/\/www.airkule.com\/images\/image\/20240916-mcetinguc-7.jpg\")
<\/p>\n![\"\"](\"https:\/\/www.airkule.com\/images\/image\/20240916-mcetinguc-8.jpg\")
<\/p>\n