INTRODUCTION
Artificial intelligence is rapidly transforming industries worldwide, enhancing efficiency, and creating unprecedented opportunities. With such inventions, there is a need to protect such technologies through Intellectual Property Rights specifically in India. Further these Artificial Intelligence are marked by certain unique challenges due to the nature of AI inventions as well as the limitations in the Patent laws of India.
Therefore the bigger issue that arises is how to examine patent applications of inventions related to CRI (Computer Related Inventions) and AI. Like any other inventions these inventions must also meet with the fundamental legal requirements that is
- Novelty
- Inventive Step
- Industrial Application
All these must be fulfilled for the application to be patentable. Further the IPO does not specifically define what an invention is rather it gives out activities that are excluded patentability to the extent that they are related to the subject matter given in the Indian Patent Act, 1970
Section 3(k): “a mathematical or business method or a computer programme per se or algorithms”; and
Section 3(m): “a mere scheme or rule or method of performing mental acts or a method of playing a game”.
Further to define whether in claims of patent application is considered an invention, it shall demonstrate the presence of any “technical contribution” and “technical effect” in the said subject matter.
- Now the question arises as what is considered to be ‘Technical’
To identify what is technical let us further see in the definition of a CRI invention. Any invention demonstrating some kind of positive effect on the resource of the device, such as reducing memory use or processing time, etc., is considered as technical in nature.
PATENTIBILITY OF AI INNOVATIONS
Understanding Patentability Criteria
The patentability of AI innovations involves meeting specific legal criteria that determine whether an invention is eligible for patent protection. The three core requirements for patentability are novelty, inventiveness (non-obviousness), and utility. These criteria ensure that AI inventions contribute something new and useful to the existing body of knowledge. Let’s delve deeper into each criterion, particularly in the context of AI-based innovations:
Novelty
Novelty refers to the requirement that an invention must be new and not previously disclosed or made publicly available before the patent application is filed. For AI technologies, novelty is often a challenging aspect, especially as the field evolves rapidly, with new algorithms, applications, and methodologies emerging regularly. To meet the novelty requirement, AI inventions must demonstrate that they represent something that has not been disclosed in prior art (which includes previous patents, research papers, products, and public use).
The challenge arises in determining what exactly constitutes a “new” idea in AI, given that many AI innovations are built on existing machine learning models or previously known algorithms. For instance, the introduction of a more efficient version of an existing AI algorithm may be novel in certain aspects (e.g., speed, accuracy, or scalability), but it may be difficult to prove that it is entirely new in every detail. As AI evolves, patent examiners must carefully evaluate the extent to which the core principles and techniques of an AI invention have already been disclosed.
Inventiveness (Non-obviousness)
Inventiveness, or non-obviousness, is a standard that requires the invention to not be obvious to someone skilled in the field, based on prior art. This criterion presents specific challenges when applied to AI technologies, as many AI innovations are incremental in nature, building upon existing frameworks and datasets. AI systems often improve upon or optimize existing algorithms, which raises the question of whether the modification or application of an algorithm can be considered “non-obvious.”
For example, an AI-powered solution that improves a pre-existing deep learning model may appear obvious to an expert familiar with similar machine learning systems, especially when the advancements are modest. Patent examiners must therefore assess whether the invention involves a sufficient level of innovation that it would not have been apparent to a skilled professional in the field at the time of the invention’s creation. This becomes increasingly difficult in an environment where AI is based on iterative improvements and fine-tuning, rather than completely novel breakthroughs.
Utility
The utility requirement mandates that an invention must be useful and capable of providing a practical benefit or solving a real-world problem. In the context of AI inventions, this criterion is often met relatively easily, as many AI applications have clear, practical uses across various industries such as healthcare, autonomous vehicles, finance, and cybersecurity. For example, AI algorithms that predict patient outcomes in healthcare, optimize supply chains, or enhance autonomous driving systems clearly provide utility by addressing important societal challenges.
However, the application of the utility standard becomes more complex when dealing with abstract AI models or algorithms that may not yet have a defined use case. The utility of an AI-based invention must be clearly demonstrated, especially in fields where the technology is still in early development stages or its practical applications are not yet fully realized. This is particularly true for foundational AI technologies, like new types of neural networks, where their utility may not be immediately apparent outside of theoretical research.
CHALLENGES IN PATENTING AI INVENTIONS
Patenting AI innovations presents unique challenges due to the rapid pace of technological advancement, the complexities of intellectual property law, and the nature of AI itself. Here are the key challenges in patenting AI:
1. Defining AI Innovations
AI encompasses a wide range of technologies, from machine learning to neural networks. Determining what qualifies as novel or inventive is challenging, as many AI systems are built on existing algorithms, making it difficult to define the boundaries of a new invention.
2. Patentable Subject Matter
AI inventions often involve abstract concepts like algorithms and mathematical models, which are typically excluded from patentability. Determining whether an AI-based innovation is eligible for patent protection can be difficult, especially when AI relies on abstract ideas rather than tangible technologies.
3. Non-Obviousness and Incremental Innovation
AI innovations often build incrementally on existing models, raising questions about non-obviousness. Small adjustments to AI algorithms may not meet the inventive step requirement, especially when they improve upon already known technologies.
4. Ownership and Inventorship
Determining the ownership and inventorship of AI-generated inventions is a complex issue. In most jurisdictions, only humans can be recognized as inventors, but AI systems increasingly create new innovations autonomously, prompting debates about whether AI should be considered an inventor.
5. Demonstrating Utility
AI models, especially in their early stages, may not have an immediately clear practical application. The utility requirement for patents demands that inventions provide a specific, useful benefit, but the abstract nature of some AI technologies can make it difficult to demonstrate their utility early in development.
6. Speed of Innovation vs. Patent Approval
The rapid pace of AI innovation often outpaces the slower patent approval process. Innovations may be overtaken by new developments before they are granted patent protection, reducing the value of the patent.
7. Patent Thickets and Infringement Risks
The growing number of AI patents can lead to patent thickets—overlapping patents that may hinder innovation and create infringement risks. Navigating these complex patent landscapes can be challenging, especially for smaller players.
8. Jurisdictional Differences
Patent laws vary by jurisdiction, and different countries have different approaches to AI patentability. These differences create uncertainty for AI innovators filing patents in multiple regions, making international patenting more complex.
9. Open Source vs. Patents
The open-source nature of many AI projects conflicts with the exclusive nature of patents. Open-source contributions often cannot be patented, leading to tension between collaborative development and the commercial benefits that patents provide.
CONCLUSION ON PATENTS AND AI
The intersection of patents and artificial intelligence presents a dynamic and evolving challenge for both legal systems and the tech industry. As AI continues to advance rapidly, traditional patent frameworks are being tested, particularly in areas like patentability, inventorship, and the balance between innovation and accessibility.
AI innovations often push the boundaries of conventional patent law, especially when it comes to defining novel inventions and addressing the complexities of abstract algorithms and incremental improvements. Issues such as determining patentable subject matter, ownership, and demonstrating utility are further complicated by the evolving nature of AI technology.
Despite these challenges, the patent system plays a crucial role in encouraging innovation and protecting the intellectual property of AI creators. However, as AI continues to permeate diverse industries, there is a growing need for reforms in patent law to accommodate these new technologies and ensure that they foster rather than stifle innovation. The development of clearer guidelines, international harmonization of patent laws, and better alignment between AI innovation and legal protections will be vital for ensuring that the patent system supports the ongoing growth and application of AI.
In conclusion, while AI brings both opportunities and challenges to the patent system, the future of AI patents will depend on how effectively legal frameworks evolve to address these issues. By adapting to the unique characteristics of AI, the patent system can continue to encourage technological progress while balancing the interests of innovators, consumers, and society.
Contributed by Karan Bhalla (Legal Intern)