A Roadmap towards Machine Intelligence
Tomas Mikolov1, Armand Joulin1 and Marco Baroni1,2 1Facebook AI Research 2University of Trento
Abstract
The development of intelligent machines is one of the biggest unsolved challenges in computer science. In this paper, we propose some fundamental properties these machines should have, focusing in particular on communication and learning. We discuss a simple environment that could be used to incrementally teach a machine the basics of natural-language-based communication, as a prerequisite to more complex interaction with human users. We also present some conjectures on the sort of algorithms the machine should support in order to profitably learn from the environment.
中文速览
让机器真正听懂人话、并通过对话自主学习，是人工智能领域长期悬而未决的核心难题。这篇论文提出，一台实用的智能机器必须具备两种基本能力——用自然语言交流，以及从交互中持续学习——并围绕这两点设计了一套循序渐进的"幼儿园式"模拟训练环境（simulated ecosystem），让机器先在受控环境中掌握语言沟通的基础，再逐步扩展到真实世界任务。作者还探讨了机器需要具备哪类算法能力，才能像人类学习者一样仅凭少量示例就提炼出正确的规律。这项工作的意义在于，它不试图一蹴而就地"解决AI"，而是提供了一条可操作的路线图，把通用智能的宏大目标拆解为一系列可验证的小步骤，为研究者指明了一个兼顾理论严谨性与工程可行性的方向。
原文 arXiv:1511.08130;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1511.08130v2