Patent Eligibility of AI Algorithms:Constraints and Paths to Realization
In the era of intelligence,with the deepening of the innovation degree of intelligent algorithms and the successive generalization of commercial application,the patent protection mode of"algorithm+hard-ware+application field"no longer meets the needs of their development.The exploration of"algorithm itself"seeking patent protection came into being,but it is difficult to promote it substantially because of the dichotomy polarization of"abstract ideas and concrete application",the harshness and anomie of technical elements,and the emphasis of technical standards on formal elements and the neglect of essential attributes.It is found that the AI algorithms based on the"technical empowerment"of AI possess technical attributes,so we should abandon the shackles of formal elements of"hardware"and return to the judging criteria for the essence of knowledge at-tributes.Based on the Stokes model,this paper reconstructs the tetrad theory and establishes three principles for rationally setting the threshold of patent eligibility under the guidance of the concept of value equilibrium,algo-rithmic transparency,and scientific and technological development concepts:adhering to the bottom line of un-patentability of purely theoretical algorithms;restricting the object eligibility of abstract application-driven al-gorithm;and comprehensively liberalizing the application-driven object eligibility of concrete application algo-rithms.With this,the innovative development of China's AI technology and the global patenting of major scien-tific and technological fields will be promoted.
AI algorithmobject eligibilityabstract ideaconcrete applicationpatentability
NETL
NSTL
万方数据
