作为谎言机器的ChatGPT

ChatGPT: the Machine of Lies

人与AI之共存共生,往往涉及感觉、信念和思考这三个重要面向。而人对AI之思考,又始终难以回避说谎这个根本的难题。本文不妨分三个步骤对这个难题进行回应。首先,我们应该回归哲学史,梳理其中围绕说谎这个问题所展开的义务和功利之间的辩证。其次,由此就引向自欺这个核心的难题,而我们亦试图用EFP这个全新的范式来取代SIT的传统模型。在EFP范式之中,问题取代了真理,表达突破了语言的范域,并最终将自欺视作以虚构和想象为特征的充满创造力的思考之源。ChatGPT正是在这个意义上被称作一部名副其实的谎言机器。最后,我们试图转向萨特的另类而极端的自欺理论,并意在打开绝对否定这另一个反思视角。

The coexistence of human and AI,generally speaking,involves three important aspects:feeling,belief and thinking.In the case of human thinking about AI,it is always difficult to avoid the fundamental problem of telling lies.In this paper,we would like to respond to this dilemma in three steps.First,we should return to the history of philosophy and sort out the dialectic between obligation and utility that surrounds the issue of lying.Second,this leads to the central dilemma of self-deception,and we attempt to replace the traditional model of SIT with the new paradigm of EFP in which questions replace truths,expressions break out of the linguistic paradigm,and ultimately self-deception is seen as a source of creative thinking characterized by fiction and imagination,so ChatGPT is in this sense a veritable machine of lies.Finally,we try to turn to Sartre’s alternative and extreme theory of self-deception,with the intention of opening up an alternative reflective perspective of absolute negation.