There is no consensus on the necessary and sufficient conditions for attributing intelligence to a machine. However, the Turing test remains a crucial milestone and benchmark for assessing the capabilities of machine intelligence. In this paper, we examine two of what we see as possible obstacles to passing the Turing test. The first is in the area of logical inference, which is at the core of programming. The second is the lack of effective solutions to some problems in the area of computational complexity. In presenting these two obstacles, we try to combine the literature on the discipline of logic and theoretical computer science in a philosophical way. Our paper concludes that the Turing test is impassable as long as these two obstacles exist.
Alan Turing Turing test artificial intelligence computational complexity non monotonic reasoning
Bir makineye zekâ atfetmenin gerek ve yeter koşulları konusunda herkesçe uzlaşılan bir zemin yoktur. Ancak Turing testi, makine zekâsının yeteneklerini değerlendirmek için çok önemli bir kilometre taşı ve ölçüt olarak durmaya devam etmektedir. Bu makalede, Turing testinin geçilmesinin önündeki bizce muhtemel engellerden ikisini inceliyoruz. Birincisi, programlamanın temelinde yer alan mantıksal çıkarıma dayalı engellerdir. İkincisi ise hesap karmaşıklığı alanındaki bazı problemlerle ilgili etkili çözümler bulunamamasından kaynaklanan engeldir. Bu iki engeli ortaya koyarken mantık disiplinine ve teorik bilgisayar bilimine dair literatürü felsefi bir yöntemle sentezlemeye çalışıyoruz. Çalışmamız, yukarıda sözü edilen iki engelden hareketle ve bu engeller var olduğu sürece Turing testinin geçilemez olduğu sonucuna varmaktadır.
Primary Language | Turkish |
---|---|
Subjects | Epistemology, Philosophy of Artificial Intelligence, Logic |
Journal Section | Research Articles |
Authors | |
Publication Date | May 7, 2024 |
Submission Date | February 15, 2024 |
Acceptance Date | April 29, 2024 |
Published in Issue | Year 2024 Issue: 37 |
Starting from 2024, our journal will be published in 3 issues as two regular and one special issues. These issues will be published In May (regular issue), September (special issue) and December (regular issue).
Acceptance of articles for our special issue and our regular issue in December will begin on March 15.
Only articles within the scope of the file will be included in our special issue.
Thank you for your attention.