and, One with artificial intelligence (TO) created invention can generally be patented, However, there are some legal challenges and requirements, that must be observed.
Fundamentals of patentability
For patenting an invention, Unabhängig davon, whether it was developed with AI, the following general requirements apply:
- Newness: The invention must not be known or publicly accessible before the registration date.
- Inventive step: It has to stand out clearly from the state of the art and not be obvious for a specialist.
- industrial applicability: The invention must be practically usable and make a technical contribution.
Special challenges for AI generated inventions
- Inventor:
- According to applicable patent law (from. B. Article 81 EPÜ) a "natural person" must be given as inventor. AI cannot currently be named as inventor, Because it is not legal.
- In practice, man must therefore, who programmed the AI, trained or exploited your output, be specified as inventor.
- Subject to the invention:
- It is often difficult to prove, How the AI contributed to the result and how great the human influence was.
- The patent authorities could request additional evidence, To confirm the innovative character of the invention.
- Technical contribution:
- The invention must have a technical effect or contribution, which goes beyond a mere mathematical model or an algorithmic method.
- If the AI was only used as a tool (from. B. To analyze data or generate prototypes), the technical performance of the invention is checked, not the AI process itself.
- Disclosure obligation:
- The patent annealer must describe sufficiently, As the AI produced the invention, So that a specialist can understand her. This can be complex, Since AI systems often work as a "black box".
Examples of patentable AI findings
- Inventions by AI support:
- A new chemical connection, which was developed by a AI, If the technical contribution and the purpose are clearly described.
- Optimized algorithms or new AI architectures, solve the technical problems.
- AI-based processes or systems:
- Anwendungen, in which AI is used to control physical devices (from. B. autonomous vehicles, medical technology).
Limits of patenting
- Protection of AI models itself:
- AI models or algorithms are often not patentable, Since they are seen as abstract mathematical methods. Protection is more possible here through copyrights or business secrets.
- Lack of inventive activity:
- If the invention is based on nearby solutions, that were created by the AI, the patent application could be rejected.
Summary
- An invention developed with AI can be patented, If the requirements for novelty, inventive activity and commercial applicability are met.
- Legally, a human inventor must be named, Since AI is not recognized as a legal person.
- There are special challenges in the disclosure of the AI development process and the attribution of the invention.
In the future, patent law could develop further, To explicitly create regulations for AI generated inventions. Until then, the human contribution is decisive.
Leave a Reply