人工智能术语研究阶段性成果-18页_1mb

Report 01: Phase II (Total Issue 37) of Artificial Intelligence Terminology Research

This report details the findings and terminologies developed by the CISS Project on Artificial Intelligence and International Security's Terminology Working Group.

I. Weapons (Systems)

Discusses classification: conventional weapons (offensive/defensive), weapon systems, military equipment, ordnance.

• Defines lethal autonomous weapon systems (LAWS), distinguishing between acceptable, unacceptable LAWS (based on Chinese criteria covering attribution, life/death decisions, predictability), identification, targeting, authorization, decision validation, termination (built upon but distinct from existing Chinese military terminology).zhao Shadong, Ma Yongzhong, Non-Lethal Weapons and Police Equipment, Beijing: Ordnance Industry Press, 2005, pp. 45-46. Also references national standards GB/T41867-2022 for AI terms.
• Covers non-lethal weapons, aiming to incapacitate without causing fatal harm.
• Defines autonomous/robot systems, weapon platforms (airborne, surface, underwater), and unmanned systems.
• Procedure definitions from "指挥关系" supplement the discussion.

II. Intelligent/Autonomous Platforms (Systems)

Defines robots, intelligent weapons, autonomous weapon systems, drone systems (UAV/UA), autonomous platforms (ground, surface, underwater), artificial intelligence clusters (AI clusters).

• Defines machine learning (supervised learning, prediction), deep learning (method, characteristics), neural networks (structure, origin).
• Briefly touches upon AI development stages in China by Zhang Bo and Zhu Jun, referencing first (knowledge/algorithms/computing), second (data/algorithms/computing), and partial third (knowledge/data/algorithms/computing) generation concepts.
• Covers swarm intelligence.

III. Intelligent/Autonomous Technology

Focuses on defining Artificial Intelligence (reliance on digital computation to replicate human intelligence, including true third generation (data, knowledge, algorithms, computing)), machine learning, deep learning, neural networks.

IV. Actions and Control

Defines command chains, command and control (C2 systems), decision points, human control.

V. Human-Machine Relationship

Discusses human-machine interaction, requires AI development guided by "human-centricity," safety, respect for human rights, "intelligence making better," ethical governance. Key aspects: trustworthiness, transparency (algorithmic/decision/data transparency), explainability (clear reasoning), accountability ("human control", ability to supervise, intervene, stop).

• Outlines ethical governance initiatives by China, including transparency, explainability, accountability, traceability, controllability principles for AI development.
• Consequently categorizes AI ethical standards.

Conclusion (Implicit in the term definitions): The report establishes a coordinated set of Chinese concepts and associated definitions related to military applications of AI and autonomy, structured according to terminology standards, aiming for usability, manageability, and safety.