The Panel discussion was held on 13th March 2023 (Monday) 3:30 pm at NLU Delhi, organised by the Centre for Innovation, IP and Competition (CIIPC) and IPR Chair at National Law University Delhi. The panelists representing the University of New Hampshire Franklin Pierce School of Law and discussing the US approach to AI and IP were Dean Megan Carpenter and Professor Micky Minhas, Executive Director of the Franklin Pierce Center of Intellectual Property. Representing the Indian perspective were Mr Vikrant Rana, Managing Partner, S. S. Rana & Co and Mr Ankit Sahni from Ajay Sahni & Associates.
The event began with opening remarks by Dr Harpreet Kaur, Registrar NLU Delhi and the roadmap for the discussion was laid out by Dr Yogesh Pai, moderator of the session. The discussion was centered around how frontier technology like Artificial Intelligence is impacting the contours of intellectual property law. Prof Pai set the agenda for the discussion stating that, various jurisdictions like USA and UK have floated public consultations regarding the interface of AI and IP, India is yet to conduct any such consultation, making academic discussions like these even more essential in understanding how the law ought to develop to keep up with technological changes.
Mr Rana was invited to commence the discussion, he posed the prevalent IP questions that are at the forefront when discussing AI, i.e. how to utilize the technology without infringing existing copyrights while protecting the original work and if AI-based consumer behavior would make trademark meaningless if only past purchase history dictates future consumer behavior. He mentioned that AI is especially useful for trademark searches and is already being utilized by the trademark office for this purpose. On the issue of consumer confusion and deception, where the rational human consumer could possibly be replaced by AI, he stated that this would not entirely invalidate the necessity of trademarks, as they will continue to serve the purpose of denoting quality, goodwill and a manner to identify goods and services. Prof Pai agreed that there could also be competition implications regarding how AI algorithms could promote only certain products.
Prof Carpenter noted that AI-generated works challenge what it means to be human and creative. Describing the approach taken by US law, she stated that only a human can be an author of a copyrighted work. Thus, rendering AI-created works as uncopyrightable as a photograph taken by an animal or religious texts that are supposed to be of divine origin. She also noted that AI-generated works fall short of the copyright requirements of a “spark of creativity” or “modicum of creativity”. Prof Pai brought into discussion the recent USPTO decision of rejecting the copyright over an AI-generated comic Zarya of the Dawn but allowing protection of the dialogue text authored by the applicant Kris Kashtanova. Regarding this Prof Carpenter discussed that here the AI Midjourney could either be viewed as a tool like software, a commission-based freelancer, or as a collaborative endeavor, but stated that in her opinion that the USPTO decision aligned with the requirement of originality and human authorship.
Prof Minhas discussed how AI differentiated from usual computer tools due to its capacity to process and utilize the volumes of data that our society generates in the form of consumer data from the e-retail sector and social data from social media, but tools like Chat GPT has allowed the general public to access AI technology as well. Moving on to the interface of patent law and AI, he stated that most countries seem to require a human inventor for the registration of a patent, with the exception of South Africa. He shed light on how the focus is entirely on the input and output of AI systems, but the process between them, training models are already patentable under law. He also noted that companies have less incentive to patent these training models and rather protect them as trade secrets, as there is no way to usually detect how another company’s training model operates.
Mr Sahni shared his experience of approaching the Indian Copyright Office to register a copyright over an AI-generated artwork Suryast, with himself and the AI RAGHAV as joint authors. The current approach of the Indian Copyright Office is to allow copyright over AI-generated works but there is some disapproval regarding allowing AI as a co-author. He also noted that while Canada allowed the copyright registration of Suryast with Mr Sahni and the AI RAGHAV as co-authors, USA declined their application on the grounds that the contribution of AI could not be separated from that of the human and thus would not be copyrightable. He noted that the current copyright registration process does not require any declaration regarding the usage of AI tools, thus human and AI-generated works should ideally be treated equally. He reiterated that amending legislation to protect AI technology as well as its outputs was essential otherwise it would be a setback where lack of protection and chances for commercial exploitation would diminish the interest in further developing tools like ChatGPT, DALL-E, Midjourney etc. But in the patent scape, Prof Minhas noted that in practice there is no such incentive required for patent protection of AI-related innovations like training models etc. as those can seldom be replicated by competitors so are better protected as trade secrets.
It was recognised that the advantage of AI over humans is that AIs are able to perceive far more combinations of data in an innovative manner, but this does not ensure non-obviousness or patentability. A recommendation of a sui generis term of protection for AI-generated outputs was discussed, in the manner of copyright protection over photographs. The creation of legal fiction to facilitate intellectual protection of AI-generated works could be proposed, with the person owning the AI system as a trustee. It was also stated that the future issues for trademarks regarding AI would be voice search functions where phonetic similarity of trademarks would be a key challenge. The panelists were unanimous in their assertion that it is only a matter of time before the law would necessarily have to adjust to disruptive technology like AI, and thus it was necessary to discuss what the law surrounding ought to be rather than just discussing what it already is.