Written by Amit Kumar, a Final-Ysear Student of Campus Law Centre, Faculty of Law, Delhi University, Delhi

The reports from sources like WIPO and various patent offices suggest, that in the ever-changing realm of artificial intelligence, creativity is reaching unprecedented levels. AI has played a crucial role in advancing technological limits, whether through predictive machine learning algorithms for disease detection and receptor imaging, or recurrent neural networks for financial crises prediction. Nevertheless, as the pace of technological progress quickens, our legal frameworks, notably the patent system, encounter difficulties in maintaining synchrony. 

Artificial Intelligence (AI) is a concept that involves machines performing tasks requiring human-like intelligence. AI is implemented through software, which includes algorithms and models. This software, developed for specific applications, enables machines to learn, reason, and make intelligent decisions. Examples include machine learning, natural language processing, and computer vision. AI and software are interconnected, with software development being a key aspect of bringing AI capabilities to various fields.

In India, software is eligible for patent protection if it demonstrates a technical effect or advancement, solving a technical problem. Patents are granted based on novelty, inventive step, and industrial applicability. While mathematical methods and computer programs per se are excluded, inventions with technical contributions may qualify. 

PATENTABILITY

The Alice/Mayo patent eligibility test is used to determine whether a patent claim is eligible under 35 U.S.C. § 101, which governs the types of inventions that can be patented. 

This eligibility test refers to a legal framework established by two significant US Supreme Court cases in the United States: Alice Corp. v. CLS Bank International (2014) and Mayo Collaborative Services v. Prometheus Laboratories, Inc. (2012). This test is used to determine whether a patent claim is eligible under 35 U.S.C. § 101, which governs the types of inventions that can be patented.

The test involves a two-step analysis:

1. Step One (Alice Test): a. Determine whether the patent claim is directed to a patent-ineligible concept, such as an abstract idea, a natural phenomenon, or a law of nature. b. If the claim is directed to a patent-ineligible concept, move to Step Two. If not, the claim passes the eligibility test.
2. Step Two (Mayo Test): 

1. Evaluate whether the claim includes an inventive concept that transforms the patent-ineligible concept into a practical application. 
2. The inventive concept must be significantly more than the patent-ineligible concept itself.
3. If the claim does not add significantly more, it is considered ineligible.

The Court underscored that although abstract ideas or natural phenomena, when considered in isolation, may not be eligible for patent protection, their conversion into a practical application that serves as an innovative building block could qualify for patent eligibility.

However, when applied to AI technologies, the test can yield subjective, inconsistent, and sometimes contentious results. The central challenge arises from the perception that numerous AI innovations may be deemed abstract, given their reliance on algorithms and mathematical processes.. Determining where the line is drawn between a patent-ineligible abstract idea and a patent-eligible inventive concept in the realm of AI can be challenging. 

Differing from the practices of patent offices in Europe, Japan, China, Singapore, the US, and the UK, the Indian Patent Office has yet to establish separate guidelines for the evaluation of AI-based innovations and are examined as per the Computer-Related Inventions  Guidelines 2017 (CRI guidelines). This absence of guidelines often leads to confusion among examiners regarding the appropriate approach to adopt and are examined based on the subject matter exclusions defined in Section 3(k) of the Indian Patents Act, 1970. Section 3(k) prohibits the patentability of “mathematical methods, business methods, computer programs per se, and algorithms.” The CRI guidelines offer clarification regarding what is permissible or impermissible concerning mathematical methods, business methods, computer programs per se, or algorithms.

However, there exists a lack of clarity when it comes to applying these exceptions to AI-related inventions. The general approach applied to inventions tied to computer technology is similarly extended to those centered around AI in India, resulting in routine objections to computational methods, algorithms, and applications at a general level. 

Furthermore, there exists inconsistency in the standards used to evaluate inventive characteristics and determine the specificity of patent descriptions. Currently, accurately predicting the patentability of AI inventions in India is notably challenging.

In the case of Ferid Allani vs Union of India & Ors., the Delhi High Court emphasized that computer-related inventions (CRIs) should not automatically be excluded from patentability. The court stressed the importance of evaluating the “technical effect” or “technical contribution” of the invention. The court’s ruling clarified that the prohibition on patenting is applicable to “computer programs per se” and does not extend to all inventions based on computer programs. Recognizing the significance of digital innovations, including AI and blockchain, that rely on computer programs, the court highlighted the importance of not dismissing them from patent eligibility.

INVENTIVE STEP

The concept of identifying the skilled person under section 2(ja) of the patent act , a critical factor in patent evaluation, remains a contentious issue worldwide. As the capabilities of AI continue to grow, questions arise about the extent to which AI can replicate human expertise and creativity. This raises challenges when establishing the baseline for assessing inventive steps.

In addition, the existing legal framework typically confines the realm of relevant prior art to analogous sources, which is often used to argue against the obviousness of combinations in patent claims. However, the emergence of “Inventive AI” introduces the possibility of expanding the scope of prior art beyond traditional boundaries. The proliferation of AI technologies suggests that the reservoir of prior art could exponentially increase, as AI potentially has access to a vast array of information.

Moreover, as skilled person in various fields increasingly integrates AI into their processes, it may necessitate reevaluating the standards for assessing inventiveness. The changing landscape, where AI is not only a tool but also an active contributor to innovation, underscores the need for a nuanced re-examination of the existing criteria for determining patentability and inventive step

In various legal cases, including the instance of Press Metal Corporation Limited vs. Noshir Sorabji Pochkhanawalla , the court made a ruling stating that when an invention is described in complete specification using unclear and ambiguous language, it becomes susceptible to rejection as a patent.

However, this situation is frequently observed in inventions related to artificial intelligence (AI), where the input and output are understood, but the intermediary AI-driven process remains undisclosed. In such cases, even if the procedural steps are elucidated, there is no assurance of achieving the same outcome upon execution. This complexity is a key reason why disclosing AI-related inventions comprehensively can be challenging.

DRAFTING PATENT CLAIM FOR AI BASED INVENTION

Crafting effective patent claims for AI innovations involves a multi-layered approach, starting with broad claims and progressing to specifics for comprehensive coverage. Emphasizing practical outcomes over intricate algorithms through functional language widens claim scope and simplifies understanding. Clearly specifying data usage, including type, processing, and relevance, enhances clarity and innovation. Avoiding overly broad claims and concentrating on distinct aspects of the invention highlights its value. Staying updated on AI and patent trends is crucial for aligning strategies with legal standards, ensuring robust protection.

The swift evolution of AI technology poses complexities for existing legal frameworks, especially within domains like patent law. Below are key issues and considerations:

1. Novelty and Inventiveness:
   • Patents traditionally recognize innovations that are novel and non-obvious. In the realm of AI, determining the source of inventive ideas becomes intricate. AI-generated inventions involve multiple contributors—algorithm developers, dataset creators, and users refining the model—raising questions about attributing “inventorship.”
2. Ownership and Attribution:
   • The intricate nature of AI systems, incorporating algorithms, data, and contributors, complicates determining patent rights ownership. With multiple contributors involved in AI-generated inventions, challenges arise regarding ownership and the equitable distribution of benefits.
3. Disclosure and Description:
   • Patents demand a clear and comprehensive description of inventions for replication by skilled professionals. Describing AI-generated inventions proves challenging due to their intricate and sometimes opaque nature, posing difficulties in meeting patent law’s disclosure requirements.
4. Proliferation of Prior Art:
   • The rapid AI development transforms cutting-edge technologies into common knowledge swiftly. Consequently, prior art—existing technologies or solutions—can emerge, potentially invalidating patent claims.
5. Exponential Growth:
   • The surge in AI-related patent applications can overwhelm patent offices, leading to examination delays. This situation may result in the potential issuance of patents not meeting novelty and inventiveness criteria due to inadequate prior art searches.
6. Ethical and Policy Concerns:
   • Beyond technical aspects, ethical considerations like bias in AI algorithms or potential misuse impact patent decisions. Striking a balance between technological advancement and ethical and societal concerns remains an ongoing challenge.
7. International Harmonization:
   • AI innovations transcending national borders necessitate international harmonization of patent standards and regulations to prevent inconsistencies and disputes. Achieving a cohesive approach to AI patents on a global scale becomes crucial.

In summary, navigating the legal landscape at the intersection of AI and patents involves addressing these multifaceted challenges, recognizing the intricate dynamics involved in AI-driven innovation.

CONCLUSION

n conclusion, the acquisition of patent protection holds immense significance for safeguarding AI-based innovations within the ever-evolving landscape of startup ventures. Given the widespread influence of artificial intelligence across diverse sectors, startups are increasingly leveraging AI technologies to pioneer revolutionary products and solutions. The process of obtaining patent protection for these cutting-edge innovations offers startups a myriad of critical advantages.

Chief among these advantages is the competitive edge conferred by patents, providing startups with exclusive rights to their AI-based inventions. This exclusivity not only facilitates market differentiation but also positions startups favorably to attract investors, secure funding, and establish a robust market presence. Patents serve as invaluable assets, contributing to the overall valuation of startups and fortifying their negotiating position in business dealings.

However, the journey through the patent landscape for AI-based inventions is fraught with challenges, including the intricate task of determining inventiveness and staying abreast of the rapid technological advancements. Consequently, startups must adopt a strategic approach by closely collaborating with legal experts specializing in both AI and patent law. Through this collaborative effort, startups can meticulously craft patent applications that not only meet stringent legal requirements but also effectively encapsulate the intricate technical nuances of their groundbreaking creations. This strategic and detailed approach significantly heightens the likelihood of successfully securing robust patent protection, thereby strengthening the startup’s competitive standing and ensuring a resilient position in the dynamic business landscape.

REFERENCE

1. Intellectual Property Rights in India, V K Ahuja, ISBN : 9789351433880, 2nd Edition 
2. Law Relating to Intellectual Property, Dr B L Wadehra, ISBN : 9789350350300, 5th Edition 
3. Copyright Act 1957
4. Patent Act 1970
5. Paris Convention, 1883 
6. TRIPS Agreement 1995
7. Guidelines for Examination of Computer Related Inventions (CRIs), Office of the Controller General of Patents, Designs and Trade marks 2017
8. Paragraph 4.4.4 Guidelines for Examination of Computer Related Inventions (CRIs), 2017
9. WIPO Technology Trends 2019 – Artificial Intelligence
10. Ahsan MM, Luna SA, Siddique Z. Machine-Learning-Based Disease Diagnosis: A Comprehensive Review. Healthcare. 2022; 10(3):541. https://doi.org/10.3390/healthcare10030541
11. Predicting systemic financial crises with recurrent neural networks, Journal of Financial Stability, Volume 49, 2020, 100746, ISSN 1572-3089
12. Alice Corporation Pty. Ltd Vs. Cls Bank International (19.06.2014 – USSC) : MANU/USSC/0068/2014
13. Mayo Collaborative Services, Dba Mayo Medical Laboratories, Et Al. Vs. Prometheus Laboratories, INC.: MANU/USSC/0038/2012
14. MANU/DE/4323/2019
15. MANU/MH/0221/1983