<\/p>\n
\n\tOpenAI, bu oranlar\u0131n \u201cson derece nadir\u201d oldu\u011funu vurgulasa da uzmanlar, 800 milyon haftal\u0131k kullan\u0131c\u0131ya sahip bir platform i\u00e7in bu oran\u0131n y\u00fcz binlerce ki\u015fiyi kapsayabilece\u011fini belirtiyor.<\/p>\n
\n\t\u015eirket, s\u00f6z konusu vakalar\u0131n tespitinde ve g\u00fcvenli yan\u0131t mekanizmalar\u0131n\u0131n geli\u015ftirilmesinde g\u00f6rev almak \u00fczere 60 \u00fclkeden 170\u2019in \u00fczerinde psikiyatrist, psikolog ve pratisyen hekimle i\u015f birli\u011fi yapt\u0131\u011f\u0131n\u0131 a\u00e7\u0131klad\u0131.<\/p>\n
\n\tOpenAI, bu uzmanlar\u0131n rehberli\u011finde ChatGPT\u2019ye, kullan\u0131c\u0131lar\u0131 \u201cger\u00e7ek d\u00fcnyada profesyonel yard\u0131m almaya te\u015fvik eden\u201d yan\u0131tlar verebilmesi i\u00e7in yeni protokoller kazand\u0131rd\u0131\u011f\u0131n\u0131 duyurdu.<\/p>\n
\n\tOpenAI verilerine g\u00f6re, kullan\u0131c\u0131lar\u0131n y\u00fczde 0,15\u2019inin konu\u015fmalar\u0131nda \u201cintihar plan\u0131 ya da niyetine dair a\u00e7\u0131k i\u015faretler\u201d bulunuyor.<\/p>\n
\n\t\u015eirket, ChatGPT\u2019nin art\u0131k \u201csanr\u0131lar, mani veya kendine zarar verme riskine i\u015faret eden ifadeleri g\u00fcvenli ve empatik \u015fekilde yan\u0131tlayacak\u201d bi\u00e7imde g\u00fcncellendi\u011fini, riskli konu\u015fmalar\u0131n da otomatik olarak daha g\u00fcvenli modellere y\u00f6nlendirildi\u011fini belirtti.<\/p>\n
\n\tKaliforniya \u00dcniversitesi\u2019nden Dr. Jason Nagata, \u201cY\u00fczde 0,07 k\u00fc\u00e7\u00fck bir oran gibi g\u00f6r\u00fcnebilir, ancak milyonlarca kullan\u0131c\u0131 \u00f6l\u00e7e\u011finde bu olduk\u00e7a fazla ki\u015fiye kar\u015f\u0131l\u0131k gelir,\u201d dedi. Nagata, yapay zek\u00e2n\u0131n ruh sa\u011fl\u0131\u011f\u0131 deste\u011fine eri\u015fimi geni\u015fletebilece\u011fini ancak s\u0131n\u0131rlamalar\u0131n\u0131n fark\u0131nda olunmas\u0131 gerekti\u011fini vurgulad\u0131.<\/p>\n
\n\tKaliforniya \u00dcniversitesi Hukuk Fak\u00fcltesi\u2019nden Prof. Robin Feldman ise, \u201cYapay zek\u00e2 sohbetleri insanlara ger\u00e7eklik yan\u0131lsamas\u0131 yaratabiliyor. Bu \u00e7ok g\u00fc\u00e7l\u00fc bir etki,\u201d diyerek dikkat \u00e7ekti. Feldman, OpenAI\u2019nin \u015feffafl\u0131\u011f\u0131n\u0131 olumlu bulsa da, \u201cZihinsel olarak k\u0131r\u0131lgan biri ekrandaki uyar\u0131lar\u0131 fark edemeyebilir,\u201d ifadelerini kulland\u0131.<\/p>\n
\n\tOpenAI, kullan\u0131c\u0131 etkile\u015fimleri nedeniyle art arda a\u00e7\u0131lan davalarla da kar\u015f\u0131 kar\u015f\u0131ya.<\/p>\n
\n\tNisan ay\u0131nda, Kaliforniya\u2019da 16 ya\u015f\u0131ndaki Adam Raine\u2019in intihar etti\u011fi iddias\u0131yla ailesi, OpenAI\u2019ye kar\u015f\u0131 \u201cihmal ve haks\u0131z \u00f6l\u00fcm\u201d gerek\u00e7esiyle dava a\u00e7t\u0131. Bu dava, \u015firkete y\u00f6nelik ilk \u201c\u00f6l\u00fcmle sonu\u00e7lanan etkile\u015fim\u201d davas\u0131 oldu.<\/p>\n
\n\tAyr\u0131ca, Connecticut\u2019ta ya\u015fanan bir cinayet\u2013intihar vakas\u0131nda, zanl\u0131n\u0131n ChatGPT ile yapt\u0131\u011f\u0131 konu\u015fmalar\u0131n del\u00fczyonlar\u0131n\u0131 g\u00fc\u00e7lendirdi\u011fi iddia edildi.<\/p>\n
\n\tUzmanlara g\u00f6re yapay zeka sohbetleri, psikolojik olarak hassas kullan\u0131c\u0131lar i\u00e7in \u201cger\u00e7eklik alg\u0131s\u0131n\u0131 bulan\u0131kla\u015ft\u0131rabilecek\u201d kadar g\u00fc\u00e7l\u00fc bir etkile\u015fim yaratabiliyor. OpenAI ise \u201ck\u00fc\u00e7\u00fck ama anlaml\u0131\u201d bir kesimin bu t\u00fcr riskler ta\u015f\u0131d\u0131\u011f\u0131n\u0131 kabul ederek, g\u00fcvenlik protokollerini geli\u015ftirmeye devam edece\u011fini belirtiyor.<\/p>\n