Social Sciences in China (Chinese Edition)

No. 2, 2024

The Philosophical Basis, Fundamental Features and Implementation Methods of Machine Ethics

(Abstract)

Liao Beishui

With the constant improvement of the autonomous decision-making ability of artificial intelligence (AI) systems, the question of how to endow such systems’ decision-making with sufficient ethical considerations has become an important challenge that has attracted widespread attention. The key approach to resolving this problem is to establish a machine ethics that embeds human ethical values and moral norms into AI systems, enabling them to possess a capacity for ethical alignment. Machine ethics is based on human ethics but its fundamental features are different. First, current AI machines lack agency and experience in the real sense; this is expressed as weak agency in ethical decision-making. Second, the machines’ decision-making should reflect the ethical considerations of the human stakeholders affected by their actions, so their ethical decision-making needs to balance the values of different stakeholders, i.e., they should have the capacity for societal balancing. Third, AI machines are easily affected by cultural factors in ethical decision-making, so should be able to reflect cultural differences. Finally, they need to explain their ethical decisions to human agents; understand emotional expression; and attribute responsibility. This requires good human-machine interaction.