讲座信息 | The Problem of AI-Human Content Alignment

跨境Amy

2025-10-22

讲座信息

系列：分析哲学系列讲座

主题：The Problem of AI-Human Content Alignment

主讲人：Nikolaj Pedersen（延世大学）

时间：2025.10.24（周五） 15：00—17:00

地点：李兆基人文学苑3号楼109

Abstract

讲座简介

The problem of value alignment has received a great deal of attention from AI researchers across a wide range of disciplines. Value alignment is considered important because it is widely regarded as one of the key components of AI safety. Aligning AI values with human values will serve as a safety valve: it will help mitigate AI risk. This paper introduces a new alignment problem: the problem of AI-human content alignment. Let ‘human content’ be content featured in human minds or language, and let ‘AI content’ be content configured by AI systems. How can alignment between AI content and human content be achieved? The paper does two things. First, a cluster of examples of content alignment and misalignment is provided. Second, the paper gives several reasons why the problem of AI-human content alignment is important.

供稿 | 何嘉诚

排版 | 刘枫林

审核 | 王彦晶

发布 | 刘枫林

【声明】内容源于网络

跨境Amy

跨境分享站 | 每日更新跨境知识

内容 44207

粉丝 2

跨境Amy 跨境分享站 | 每日更新跨境知识

总阅读252.7k

粉丝2

内容44.2k