大语言模型的技术政治学评析

A Review of the Technopolitics of Large Language Model

知识生产者的行列里,出现了一位人类眼中的陌生者:作为生成式人工智能的大语言模型。以ChatGPT为代表的大语言模型,展现出接近通用的智能水准,同时亦有大量使用者报告各种错误,甚至错误连连。由此,对大语言模型的智能形态展开系统性的探究,可从技术政治学视角切入,并在此基础上对它生成错误与意识形态偏见之缘由展开分析。美国政府要求“美国价值”必须是开发与应用人工智能系统的准绳,这意味着人工智能时代,至关重要的论题仍然是意识形态的批判性分析。

In the ranks of knowledge producers,a stranger to the eyes of mankind emerged:Large Language Model(LLM)as generative artificial intelligence.The LLM,represented by ChatGPT,demonstrates a near-general-purpose level of intelligence,while at the same time a large number of users report a variety of errors and even frequent errors.It therefore leads to a systematic inquiry into the intelligent form of the LLM from a techno-political perspective,based on which the causes of its generative errors and ideological biases are analyzed.The U.S.government requires that"American values"must govern the development and application of artificial intelligence systems,which means that the crucial thesis in the age of artifici intelligence remains critical analysis of ideology.