提出用实体保真度定义智能，可评估各类智能行为。
On the Definition of Intelligence
- 以实体保真度为核心，统一衡量智能行为。
- 通过ε-概念智能量化生成结果与原实体的差异容忍度。
- 适用于强化学习、生成模型等场景，适合智能评估研究者。
为实现通用人工智能（AGI），需先以跨物种通用的形式捕捉智能本质，使其可评估且能涵盖多样智能行为范式，包括强化学习、生成模型、分类、类比推理和目标导向决策。本文提出基于实体保真度的通用准则：智能是给定表征某一概念的实体后，生成同样表征该概念实体的能力。我们将其形式化为ε-概念智能——若任意允许的区分器无法在容忍度ε内区分生成实体与原始实体，则称其具有ε-概念智能。本文构建了形式框架，概述了实证协议，并讨论了对评估、安全性和泛化性的意义。
原文摘要 · Abstract (English)
To engineer AGI, we should first capture the essence of intelligence in a species-agnostic form that can be evaluated, while being sufficiently general to encompass diverse paradigms of intelligent behavior, including reinforcement learning, generative models, classification, analogical reasoning, and goal-directed decision-making. We propose a general criterion based on \textit{entity fidelity}: Intelligence is the ability, given entities exemplifying a concept, to generate entities exemplifying the same concept. We formalise this intuition as \(\varepsilon\)-concept intelligence: it is \(\varepsilon\)-intelligent with respect to a concept if no chosen admissible distinguisher can separate generated entities from original entities beyond tolerance \(\varepsilon\). We present the formal framework, outline empirical protocols, and discuss implications for evaluation, safety, and generalization.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。