China's Ministry of Science and Technology released an ethical guideline for the research and development of driving automation technology on July 23, which proposes corresponding ethical norms for different types of driving automation systems.

Life safety first

"Driving automation systems possess both technical and social attributes. Their technological innovation and industrialization must fully consider the public's values and ethical concerns," said Gong Ke, a member of the Artificial Intelligence Ethics Subcommittee of the National Science and Technology Ethics Committee.

The guideline applies to entities involved in driving automation technology, including technology developers, manufacturers and users. It is meant for vehicles with driving automation functions, covering both the advanced driver assistance function and automated driving function.

The policy emphasizes driving safety, stating that the technology's development and application must take safety fully into account. This includes not only the safety of human participants in automated driving, but also road traffic environment safety and the security of data, algorithms and other technical aspects.

"In terms of risk prevention, the guideline places high importance on respect for and protection of life safety. It explicitly states that in unavoidable traffic accidents or extreme driving conditions, harm to humans should be minimized to the greatest extent," Gong Ke stated.

He added that the guideline also focuses on preventing risks to relevant parties from factors such as data privacy, algorithmic discrimination and algorithmic interpretability.

Clear rules for different types

Drawing on international standards, national policies and local practices, the guideline categorizes driving automation into three types and assigns responsibilities accordingly.

For vehicles with advanced driver-assistance systems, or Level 2 and below driving automation, human drivers are responsible for the driving process.

For conditional automation, encompassing Level 3 and Level 4 driving automation, human drivers, system developers or manufacturers are liable, depending on specific operational scenarios.

At the full automation stage, or Level 5 driving automation, the automated driving system generally serves as the primary responsible entity. However, the human driver actively intervenes in special situations.

"Due to limitations in professional knowledge, the public can easily fall into a cognitive gap regarding driving automation technology, leading to misconceptions and misuse of related products," Gong said. Under the pressure of commercial competition, technology developers and car manufacturers may exaggerate technical capabilities or conceal risks, potentially causing accidents that trigger public panic and a crisis of trust.

Therefore, to promote healthy technological development, build extensive societal trust and ethical consensus, the guideline also lists public education and communication norms that the relevant parties must follow.