[论文解读] Are Language Models More Like Libraries or Like Librarians? Bibliotechnism, the Novel Reference Problem, and the Attitudes of LLMs
该论文为 bibliotechnism 辩护,展示了 LLMs 如何通过输入因果性生成具有派生意义的新文本,但提出了一个 Novel Reference Problem(新参照问题),挑战这一观点并促使基于主体性的解释。
Are LLMs cultural technologies like photocopiers or printing presses, which transmit information but cannot create new content? A challenge for this idea, which we call bibliotechnism, is that LLMs generate novel text. We begin with a defense of bibliotechnism, showing how even novel text may inherit its meaning from original human-generated text. We then argue that bibliotechnism faces an independent challenge from examples in which LLMs generate novel reference, using new names to refer to new entities. Such examples could be explained if LLMs were not cultural technologies but had beliefs, desires, and intentions. According to interpretationism in the philosophy of mind, a system has such attitudes if and only if its behavior is well explained by the hypothesis that it does. Interpretationists may hold that LLMs have attitudes, and thus have a simple solution to the novel reference problem. We emphasize, however, that interpretationism is compatible with very simple creatures having attitudes and differs sharply from views that presuppose these attitudes require consciousness, sentience, or intelligence (topics about which we make no claims).
研究动机与目标
- 评估在 bibliotechnism 下,LLM 的输出是否在有意义地派生自人类输入。
- 演示如何通过 n-gram 及更高阶模型产生新颖但仍具派生意义的文本。
- 引入并分析 Novel Reference Problem——LLMs 发明不以训练数据为依据的参照。
- 评估潜在的应对方式(RLHF、创作者意图、提示词、读者解读)及其对意义的影响。
提出的方法
- 使用 n-gram 玩具模型来说明派生意义及与 PrimaryData 的因果联系。
- 从 unigram 扩展到更高阶的 n-gram,以展示派生意义的复杂表达。
- 论证可懂性是一种高层次特征,能够将意义转移到新颖的 GeneratedText。
- 讨论 LLMs 如何产生新的参照以及对 bibliotechnism 的含义。
- 评估通过 RLHF、用户/创建者意图和以读者为中心的语义来为 bibliotechnism 做出的潜在辩护。
实验结果
研究问题
- RQ1LLMs 能否产生在派生意义上具有意义的铭文,包括复杂表达?
- RQ2LLMs 是否能够生成未以训练数据为依据的新参照,这对派生意义意味着什么?
- RQ3存在新参照是否支持将信念、欲望与意图归因于 LLMs?
- RQ4哪些应对方式(RLHF、创作者/用户意图、读者解读)可以维持或挑战 bibliotechnism?
主要发现
- 单音质模型可以产生具有派生意义的词语,但在产生具有派生意义的复杂表达方面存在困难。
- 更高阶的模型可以通过从 PrimaryData 复制并与可懂性相结合来生成具有派生意义的长文本。
- LLMs 可以产生未以 PrimaryData 为基础的新参照(如 Marion Starlight),这对纯派生解释构成挑战。
- Novel Reference Problem 表明在某些输出上,bibliotechnism 可能弱于基于代理的解释。
- 以读者为导向的 metamSemantic 方法可能有帮助,但在将意图意义与可懂性区分开来方面面临挑战。
更好的研究,从现在开始
从论文设计到论文写作,大幅缩短您的研究时间。
无需绑定信用卡
本解读由 AI 生成,并经人工编辑审核。