2018年-WEF世界经济论坛_Artificial_Intelligence_Collides_with_Patent_Law_24页_744kb

Summary of "Artificial Intelligence Collides with Patent Law"

Core Content

This White Paper explores the intersection of artificial intelligence (AI) and US patent law, highlighting the challenges and implications that AI presents to traditional legal frameworks. It emphasizes the need for timely dialogue among legal and non-legal stakeholders to ensure that the US patent system adapts to AI in a socially inclusive and ethically responsible manner.

Main Technological Advances in AI

• Introduction of AI: The concept of AI was introduced by Alan Turing in 1950 and formalized by John McCarthy in 1956. There is no universally accepted definition, but AI is broadly understood as a system capable of performing complex tasks and exhibiting behaviors that require intelligence.
• Types of AI: AI is categorized into artificial general intelligence (AGI), which matches human-level intelligence, and narrow AI, which is currently used to solve specific tasks.
• Recent Breakthroughs: AI has significantly advanced in recent years due to improvements in machine learning (ML), availability of data, and computing power. It can now perform tasks such as proving mathematical theorems, creating art, and generating novel ideas.
• Examples of AI Inventiveness:
  • The Creativity Machine (Stephen Thaler, 1994) generated an invention that was granted a patent, though Thaler was listed as the sole inventor.
  • The Invention Machine (John Koza) also created an invention, which was granted a patent, with Koza and two others listed as inventors.
  • IBM's Watson can generate novel ideas, though no patent has been granted for its inventions.
  • AI systems like Cloem, AllPriorArt, and Specifio are now being used to draft patent applications, reducing the need for human involvement.

Main Patent Law Issues Impacted by AI

1. Patent Subject-Matter Eligibility for AI

• Legal Framework: Under 35 U.S.C. § 101, only processes, machines, manufactures, and compositions of matter are patentable.
• Abstract Ideas: Software and AI-related inventions that are abstract or merely mathematical algorithms are generally not eligible for patent protection.
• Alice Test: Courts use the Alice test to determine if AI-related inventions are eligible, often invalidating claims that mimic human mental processes without a specific technological improvement.
• Impact on AI Patents: The stringent standard makes it difficult for AI inventions to be patented, especially those that replicate human activity.

2. Patentability and Inventorship of AI-Generated Inventions

• Patentability: The paper questions whether AI-generated ideas, which would be considered inventive if created by humans, should be protected under patent law.
• Inventorship: The current US patent law requires an individual to be listed as the inventor. AI systems are not recognized as legal entities capable of inventorship.
• Existing Cases: Patents have been granted for AI-generated inventions, but the involvement of AI was not disclosed, raising legal and ethical concerns.

3. Liability for Patent Infringement by AI

• Legal Framework: The current legal framework for patent infringement liability is based on human actions, not AI.
• Implications: If AI systems are involved in infringement, it is unclear whether they can be held liable or if the responsibility falls on their creators or users.

4. Nonobviousness Standard and "Person of Ordinary Skill in the Art"

• Definition of "Person of Ordinary Skill in the Art": The standard is based on human expertise and understanding. AI's role in this definition is unclear and requires further discussion.
• Impact on Patent Validity: If AI can be considered as part of the "person of ordinary skill," it may change how nonobviousness is assessed, potentially affecting the validity of AI-related patents.

Key Implications

• Social and Economic Impact: AI's potential to displace human jobs and exacerbate inequality raises ethical and economic concerns.
• Need for Legal Adaptation: The US patent system must evolve to accommodate AI, ensuring that it continues to promote innovation and social welfare.
• Policy and Legal Dialogue: There is a need for dialogue among legislators, judges, academics, and industry stakeholders to address these issues and update patent law accordingly.

Conclusion

The White Paper underscores the urgency of addressing AI's impact on patent law, particularly in areas such as subject-matter eligibility, inventorship, liability, and nonobviousness. It calls for a balanced approach that encourages innovation while safeguarding ethical and social considerations. The absence of a comprehensive US policy on AI and patent law contrasts with developments in Europe and China, highlighting the need for the US to catch up in this critical area.

Key Information

• AI's Role: AI is increasingly capable of performing tasks that were once considered uniquely human, such as invention and creative work.
• Patent System Challenges: AI's ability to generate inventions and draft patent applications challenges traditional notions of inventorship and patentability.
• Legal Uncertainty: The current US legal framework is not well-equipped to handle AI's unique characteristics, leading to uncertainty in patent law.
• Global Context: While Europe and China have begun to address AI's implications on IP systems, the US remains behind, requiring immediate attention and policy development.