Chengying Guan

本文立足于行动哲学视角探讨如何将英文中agent一词在人工智能及跨学科学术语境中翻译为汉语的问题。以新近提出的“行为体”译法为主要商榷对象，本文论述了其在具身性预设、谓述兼容性、范畴外延、双重语义对仗方面临的挑战，主张以“能动元”（意为“能动单元”）作为一般情况下的替代译法。此外，本文主张在该译法的基础上建立“全称补述+简称标注”的弹性翻译机制等配套措施，以适应不同学科和语境下的翻译需求，并且兼容某些已经通行的译法。例如，可将intelligent agent翻译为“智慧能动元”，简称“智能元”；如根据语境相关指涉对象还具有具身性，则用“体”取代“元”作为后缀，将embodied agent翻译为“能动体”，intelligent embodied agent翻译为“智慧具身能动元”或“智慧能动体”，简称“智能体”。[This paper addresses the translation of the English term “agent” into Chinese in the context of artificial intelligence and interdisciplinary fields, from the perspective of the philosophy of action. Taking the newly proposed translation “行为体” (xíng wéi tǐ) as the target of critical analysis, the paper examines challenges to it in terms of embodiment presupposition, predicate compatibility, category extension, and semantic parallelism. It advocates for “能动元” (néng dòng yuán) as a general alternative. In addition, on this basis, this paper proposes establishing corresponding measures, such as a flexible translation mechanism of “full description + abbreviation annotation”, to meet translation needs across different disciplines and contexts and to accommodate widely used translations. For example, “intelligent agent” can be translated as “智慧能动元” (zhì huì néng dòng yuán), abbreviated as “智能元” (zhì néng yuán). If, according to the context, the referent is also embodied, then let the suffix “元” (yuán) be replaced by “体” (tǐ), and “intelligent embodied agent” can be translated as “智慧具身能动元” (zhì huì jù shēn néng dòng yuán), abbreviated as “智能体” (zhì néng tǐ).]

The thesis is a study in the philosophy of action. It opens by identifying the field’s two central issues: action metaphysics and action explanation. It then sets out to challenge the so-called ‘standard story of action’, as a theory that addresses both issues, and to explore an alternative, with emphasis on the metaphysical dimension. Part I examines the standard story, identified more specifically as ‘standard event causalism’. Chapter 1 provides historical context by discussing the rise and fall of old volitionism as its predecessor. Chapter 2 investigates its genesis and constitution, dissecting it into six core, mutually independent claims. Chapter 3 offers a systematic critique, arguing that standard event causalism fails to resolve the problems of deviant causation and the disappearing agent, and faces other challenges such as accounting for tool-using actions. Part II lays the ontological and methodological groundwork for an alternative. Chapter 4 critiques classical agent causalism, concerning both its intelligibility and the embrace of event ontology that it shares with standard event causalism. Chapter 5 presents a methodological, cross-linguistic challenge specifically to Davidson’s influential semantic argument that actions are events. Part III develops the preliminaries of a theory falling under the strand that I call ‘novel agent causalism’. Chapter 6 explores the ontological commitments of this theory, concentrating on the idea that actions conceptualised as agent-causings constitute a category sui generis; furthermore, it investigates the ontology of agent, arguing that agents are powerful substances. Chapter 7 discusses the classification and synthesis of agents and addresses the concern that novel agent causalism remains an incomplete theory of action because it says little about action explanation. Chapter 8 explores the phenomenon of tool use, arguing it is a crucial, yet overlooked, aspect of human agency; it proposes a new definition of tool use based on the notion of affordance and argues that novel agent causalism may have a distinctive potential of accommodating this phenomenon. The thesis concludes by suggesting that its methodological departure from linguistic parochialism and its portrayal of human agents as animals who systematically rely on tools to act together embody a ‘neo-cosmopolitan’ ideal in philosophy.

This paper concerns the Wishful Thinking Problem for non-cognitivism, which has recently been raised by Cian Dorr. Contrary to Dorr’s claim that the Wishful Thinking Problem is a new crucial objection to non-cognitivism in addition to the well-known Frege-Geach Problem, I argue that recent research has shown that the Wishful Thinking Problem is not independent of the Frege-Geach Problem and therefore it is not lethal to non-cognitivism. However, the Wishful Thinking Problem is still a problem for non-cognitivism, for it reveals that the Frege-Geach Problem may be even more troublesome than non-cognitivists originally thought.