The WIPO conversation on intellectual property and frontier technologies: Sharing session – IPO AI tools and beyond
Artificial intelligence (AI) and other frontier technologies have lept on the global stage. While intellectual patent offices (IPOs) have begun tapping into technological potentials, there remain obstacles to be overcome. The World Intellectual Property Organization (WIPO) President opened the session by welcoming IPO representatives from around the world to showcase their experiences with frontier technologies. The President stressed that the interactive nature of such an exercise would encourage advancements and collaborations.
There were four broad themes addressed by the presenters. First, AI could improve, but simultaneously complicate an applicant’s experience with patent filing platforms. IPOs need to investigate how to optimise the user journey. Second, AI could enhance the efficiency of intellectual property (IP) administration in the same way it helped users. Third, there are certain human rights and organisational concerns among IPOs that adopt frontier technologies in their workflow. IPOs must ensure a smooth transition by looking into how AI impacts the work of their employees. Lastly, AI and frontier technologies might prove fruitful in battling intellectual property rights (IPRs) infringements.
Mr Ke Xu (Senior Program Administrator, IT Department, Patent Office, China National Intellectual Property Administration, China) briefed the audience on China’s AI achievements in various aspects. The new smart search system is supported by cloud computing, big data analytics, and a range of AI functions that offer efficient patent cataloguing, database generations, thematic or image searches, and translations for examiners.
Mr Saar Abramovich (Senior Patent Examiner, Israel Patent Office, Israel) spoke of the improving application success rate due to AI tools. In particular, AI helped streamline the application process by having machines preliminarily sort through applications and humans scrutinising the categorised applications. The office’s AI-enabled search tool offers free accessibility to the general public.
Mr Won Seok Huh (Deputy Director of Trade & Cooperation Division, Korean Intellectual Property Office (KIPO), Daejon, Republic of Korea) detailed several AI-applicable areas. At the application filing stage, AI could classify and examine formalities. Then, neural network (NN) tools could improve text and image searches in the database. During this stage, an AI-powered chatbot could provide support. In the final examination, AI could help draft written opinions.
Ms Bethan Curry (Senior Subject Matter Expert, Trade Mark Pre-Apply Tool Project, UK Intellectual Property Office (UK IPO), United Kingdom) identified challenges users have faced during their trademark applications and some solutions. Users typically have limited knowledge as to the success rate of their applications and the technologies used to assess them. This is why UK IPO provides tools like the goods & services (G&S) selector, those for optical character recognition, as well as automated searching with both text and image recognition to boost users’ knowledge.
Mr Rahul Bhartiya (AI Project Manager, Digital Transformation Department, European Union Intellectual Property Office (EUIPO), Spain) emphasised a human-centric approach on both the users’ and the IP examiners’ sides. To reassure fairness, users are given the same set of pre-assessment tools such as G&S, conceptual, phonetic, and image searches that the examiners will apply during examinations. An AI chatbot on EasyFiling allows users to pose questions and get redirected to a human at any time, should they prefer.
Mr Samir Ghamri Doudane (Head of Lab, National Institute of Industrial Property (INPI), Courbevoie, France) used two examples to demonstrate the low cost, low risk, yet high efficacy of utilising AI tools. By adopting deep learning techniques, the French AI lab provides patent preclassification to examiners. In customer data repository consolidation, the lab improves the consistency of SIREN-linked databases.
Ms Ana Arredondo Macua (Director, Information Technology Division, Spanish Patent and Trademark Office, Spain) exhibited a reflexive exercise that the Spanish IPO carried. The three-step exercise started with an internal discovery of which IPO processes could be robotised. Then, each process is prioritised according to a quality-efficiency matrix. Finally, employees undergo training to become familiarised with working alongside automated machines.
Mr Junhyeok Choi (Deputy Director, Cultural Trade and Cooperation Division, Copyright Bureau, Ministry of Culture, Sports, and Tourism, Republic of Korea) and Bhartiya both shed light on the potential hazards AI might bring to IPR, with Choi pointing to the rising copyright infringement issues of non-fungible tokens (NFTs) and the necessity for expert consultative bodies. Bhartiya said that blockchain technologies could be used for verification and authentication purposes.
Mr Bruno Pouliquen (IP Data Machine Learning Development Manager, Advanced Technology Applications Center (ATAC), WIPO) lastly listed WIPO-developed AI tools in text, image, and speech. For texts, there is WIPO Translate, classification tools, and text similarity check. For images, the WIPO database offers similarity searches. For speech, WIPO is developing both speech-to-text and speech-to-translated-text tools.