教育会输给技术吗？

本报告是经济合作与发展组织2023年发布的最新报告，比较了2016年和2021年人工智能和人类技能在数学和阅读方面的进展，显示了这些领域的人工智能是如何随着时间的推移而发展的，以及它们与人类的阅读和数学技能的对比。报告显示，自2016年来，人工智能在阅读能力方面取得了显著进步，但解决数学问题的能力并未以同样速度提高。专家预测，如在人工智能研究和开发方面增加投资，将导致其在未来几年在阅读和数学方面取得重大进展。 报告预测了2026年人工智能和人类技能在阅读和数学方面的发展。报告表明，人工智能在阅读和数学方面的表现可能超过大部分人口。这对就业和教育具有重要影响。报告还强调需要加强劳动力的基础技能，并使其做好准备，在关键领域与人工智能合作。 当前，AI的快速发展以及ChatGPT的横空出世，正是大家讨论的热点。哪些是AI可以做的，哪些是不可以做的，它们的发展趋势如何，教育如何应对，未来哪些职业将被取代，人们选择什么专业和职业方可不被AI取代，都是大家非常关注的。在这种背景下，本书的引进与出版显得非常及时、有意义、有价值。本书将放入“OECD教育研究与创新系列”中。

经 济 合 作 与 发 展 组 织（OECD）是各国政府合作商讨应对经济、社会和环境全球化挑战的特别组织。OECD还是促进各国相互理解，帮助各国政府应对新局势、 新问题的前线力量，例如企业管治、信息经济以及人口老龄化的挑战。OECD的出版物广泛传播该组织收集的统计数据，内容涉及经济、社会和 环境问题的研究结果，以及其成员国达成一致的协议、指导方针和标准。

纲要······················································································1 方法论·············································································2 主要发现··········································································2 结论················································································4 第一章 搭建舞台：评估人工智能影响的方法 ································7 测量人工智能能力和影响的已有研究 ··································· 10 人工智能能力探索性测评的目标 ········································· 15 本报告的结构·································································· 18 参考文献········································································ 18 注释·············································································· 20 第二章 人类技能与人工智能能力的演变 ···································· 23 技能供给的变化······························································· 25 人工智能能力的最新发展 ·················································· 36 测量人工智能能力的重要性 ··············································· 46 参考文献········································································ 48 附录2.A　补充表格 ·························································· 52 注释·············································································· 62 第三章　基于“成人技能调查”的人工智能能力测评的方法论 ·········· 63 “成人技能调查”概述 ························································ 64 确定一组计算机科学家 ····················································· 70 收集专家判断·································································· 72 问卷开发········································································ 74 构建人工智能读写和计算能力表现的综合度量 ······················· 76 困难挑战和经验教训 ························································ 77 参考文献········································································ 79 第四章　专家对人工智能读写和计算能力的测评 ··························· 81 人工智能在读写领域的能力评估 ········································· 82 人工智能在计算领域的能力评估 ········································· 96 参考文献·······································································111 附录4.A　补充表格 ·························································112 第五章　2016—2021年人工智能读写和计算能力的变化 ················113 人工智能读写能力随时间推移的变化 ··································115 人工智能计算能力随时间推移的变化 ··································123 参考文献·······································································131 附录5.A　补充图表 ·························································131 注释·············································································135 第六章　人工智能能力的持续进步对就业和教育的影响 ·················137 测评结果概要·································································138 不断发展的人工智能读写和计算能力对政策的影响 ················145 测评人工智能的新方法 ····················································158 参考文献·······································································159 注释·············································································162 译后记 ················································································163