我们应该如何在发挥人工智能最大用途的同时，预防那些机器人可能带来的威胁呢？随着人工智能的日益完善和发展，人工智能先驱斯图尔特·罗素正在创造一些不同的东西：那就是具有无限可能的机器人。让我们听听他对人类该如何兼容人工智能的看法，如何才能真正利用人工智能使其利用常识、利他主义以及人类的价值观来解决问题。

人工智能是如何让我们变得更好的

Superintelligent machines could make decisions beyond human capability, yet this rapid progress also brings risk. Giving AI fixed goals—like fetching coffee at all costs—can lead to catastrophic results, echoing the King Midas myth and the cautionary tale of HAL 9000. The solution is not to halt AI, but to design systems that act with humility, uncertainty, and a deep alignment to human values. Rather than blindly following orders, these machines must learn from human choices and remain open to correction.

Smarter Robots Need Uncertainty To Protect Humanity

"Slaughterbots"

code

communication

algorithm

humanity

machine learning

innovation

data

future

computers

robots

technology

science

decision-making

ethics

TED Stage Talk

斯图尔特·罗素: 人工智能是如何让我们变得更好的