探索与争鸣 ›› 2023, Vol. 1 ›› Issue (6): 91-98.

• 技术与文明 • 上一篇    下一篇

人工智能与未来社会:ChatGPT 专题|GPT 的现实挑战与未来风险——从人类的观点看

何怀宏   

  • 出版日期:2023-06-20 发布日期:2023-07-25
  • 作者简介:郑州大学哲学学院特聘首席教授,北京大学哲学系教授。(北京 100871)

The Real Challenges and Future Risks of GPT——From a Human Perspective

He Huaihong   

  • Online:2023-06-20 Published:2023-07-25

摘要:

人与技术的一个基本矛盾就是人类飞速发展的控制能力和落后的自控能力之间的矛盾。这一矛盾可能以这样一种方式结束:随着人的控制能力发展到极致,一种新的智能可能会突然超过人的智能而使其归零。随着 GPT 越来越容易上手,人们可以随时随地用自然语言提出各种各样的问题,这将刺激技术飞速发展并扩大应用的领域。GPT 会变得越来越通用,直到囊括几乎所有人类智能和体能的领域。那时,机器语言将战胜人类的自然语言,转变的关键在于通用人工智能是否会获得自我意识进而获得自主意识,从而成为一种全面超过人的超级智能。

关键词:

GPT, 通用人工智能, 超级智能

Abstract:

One fundamental contradiction between humans and technology lies in the stark contrast between the rapid development of human control capabilities and the lagging self-control abilities. This contradiction may eventually lead to a scenario where a new form of intelligence suddenly surpasses human intelligence, rendering it obsolete. As GPT becomes increasingly accessible, people can easily use natural language to ask a wide range of questions, which will drive the rapid advancement and expansion of technological applications. GPT will continue to evolve and become increasingly universal, encompassing nearly all aspects of human intelligence and physical capabilities. Eventually, machine language may supersede human natural language. The key lies in whether general artificial intelligence will attain self-awareness and autonomous consciousness, thus transforming into a superintelligence that surpasses humans comprehensively.

Key words:

GPT,  general artificial intelligence,  superintelligence