生成式人工智能具有意识吗?——基于神经科学哲学视角的探究

Does Generative Artificial Intelligence Have Consciousness?—Exploration Based on the Philosophy of Neuroscience

人工智能是否具有意识是哲学和科学研究中争论的焦点,生成式人工智能(GAI)的快速迭代再次点燃了这一话题。神经科学哲学作为意识研究的重要组成,近年来在理论和方法两个维度都获得了显著的成果,是考察GAI意识状况的良好基础和必由之路。这种考察可从如下几个方面展开:首先,基于计算功能主义的神经科学策略,论证意识在人类之外智能系统上的可实现性,借助意识的神经科学成果和理论偏重方法,显示出意识评估的科学可处理性;其次,通过分析当前主流的意识神经科学理论,确定标志属性,考察已有的意识测量方法,包括主观和客观两种类型,进而提供GAI意识判定的具体标准和操作手段;最后,以神经科学哲学中的时空意识理论(TTC)为例,展开说明GAI何以没有意识,并且指出其通往意识的可能途径。

Whether or not artificial intelligence has consciousness is a core of debate in philosophy and scientific research,and the rapid iteration of generative artificial intelligence(GAI)has made it the more heated topic.As an important component of consciousness research,the philosophy of neuroscience has gained significant results in both theoretical and methodological dimensions in recent years.It is a good foundation and a necessary path for the examination of the state of consciousness in GAI.Such examination can be carried out from the following aspects.First,neuroscientific strategies based on computational functionalism is adopted to argue for the realizability of consciousness on intelligent systems other than humans,drawing on neuroscientific results and theory-heavy approaches to consciousness to show the scientific processability of consciousness assessment.Second,specific criteria and operational means for GAI consciousness determination is provided by analyzing the current mainstream neuroscience theories of consciousness,identifying indicator properties,and examining existing measures of consciousness,including both subjective and objective types.Finally,the temporo-spatial theory of consciousness(TTC)in the philosophy of neuroscience is used as an example to develop an explanation of why GAI is not conscious and to point out its possible pathways to consciousness.