节点文献
预测方法对AI的限制
Limitations of Prediction Methods on Artificial Intelligence
【摘要】 今天的预测技术和人类一直依赖的推理一样,正在成为科学研究中重要的独立的驱动力。本文对二者进行比较,并指出了预测的不足,正是这种不足造就了今日AI技术发展中存在的一些天生的缺陷,最终导致实现不了agent建立完备世界模型的目标。但是,不能忽视的是,目前的AI技术已经具有缩小人与人之间智力水平的能力。
【Abstract】 Today’s predictive technologies are becoming as important and independent drivers of scientific research as the reasoning that humans have always relied on. This paper compares the two and points out the shortcomings of prediction, which are responsible for some of the inherent flaws in the development of today’s AI technology, ultimately leading to a failure to achieve the goal of agentically building a complete model of the world. However, it cannot be ignored that current AI technology has the ability to narrow the level of intelligence between people.
【关键词】 AI技术;
推理;
预测;
完备的世界模型;
【Key words】 AI technology; reasoning; prediction; complete world modeling;
【Key words】 AI technology; reasoning; prediction; complete world modeling;
- 【文献出处】 科学·经济·社会 ,Science Economy Society , 编辑部邮箱 ,2024年02期
- 【分类号】B15;TP18
- 【下载频次】12