Held "2026 Life Science IP Forum The Future of the Pharmaceutical Industry: The Advancement of Digitalization and the Routine Use of New AI

On February 3, 2026, the JPMA hosted the "2026 Life Science IP Forum" at the Sola City Conference Center (Chiyoda-ku, Tokyo). 2026 featured lectures and panel discussions by five experts under the title of "The Future of the Pharmaceutical Industry: Advances in Digitalization and the Routine Use of New AI Applications. As in 2025, the event was held in a hybrid format with an online session, and was a great success with more than 500 participants including visitors. This report provides an overview of the lectures and panel discussion.

Introduction

Since 2018, the Life Science IP Forum has so far held four discussions on "Utilization of Medical Data," "AI-based Drug Discovery," "Digitization of Information and Platforms," "IP Landscape," "Digital Therapy," "Application of AI and Human Resources," etc., through discussions on intellectual assets and IP rights related thereto. We have been discussing the handling of intellectual assets and intellectual property rights, as well as business and human resources.

It is said that the recent utilization of generative AI in each task of the value chain in the pharmaceutical industry is not limited to mere "work efficiency improvement," but has entered the realm of business automation with the advent of AI agents, which can now process multiple tasks autonomously. It is no exaggeration to say that the pharmaceutical industry must promote the use of AI to transform its business itself, or else it will be difficult to survive as a company.

In this IP Forum, we will take a bird's-eye view of what digitalization and AI utilization will bring about in the value chain of the pharmaceutical industry, and then discuss the current state of innovation in business through AI utilization in pharmaceutical companies, especially in drug discovery R&D, to improve R&D efficiency through actual AI utilization. The current state of AI utilization in the pharmaceutical industry is discussed. Explanation was also given on the handling of AI products and how they should be protected, as a legal system that should be taken into consideration when utilizing AI. He also introduced issues of data privacy, bias, and reliability of AI products in the utilization of AI. Based on the content of the first half of the session, the panel discussion in the latter half of the session was a lively exchange of views on a wide range of important issues, including the improvement of AI autonomy and its impact on intellectual property (patents), utilization of medical data, AI governance and ethics, legal regulations, AI-induced changes in intellectual property work and human resource development.

Change with AI Change with AI

Nomura Research Institute, Ltd.
Senior Principal, Healthcare and Service Industry Consulting Department, Consulting Division, Nomura Research Institute, Ltd.

The overall picture and direction of DX/AI utilization in the pharmaceutical industry, and the need for people to change first in order to make AI utilization routine.

In an overview of the current situation for Japan's pharmaceutical companies, we see the same emphasis on growth through innovation and globalization as in the past, as well as the increasing importance of productivity and cost management. In particular, the cost and timeframe of drug development is increasing. In particular, the increasing cost and timeframe of drug development are serious concerns that make it difficult to say that the current pharmaceutical industry is sustainable. In this regard, the key is whether productivity improvements using AI can be realized at the level of reform rather than improvement and with increasing speed.

The progress of DX can be categorized into three stages.

DX1.0 includes process reform for internal corporate activities and end-users; DX2.0 includes business model transformation as a stage of providing unprecedented digital services; and DX3.0 includes paradigm transformation aimed at solving social issues. While DX2.0 activities are being undertaken in many industries, most of the activities of pharmaceutical companies remain in DX1.0.

The following are the results of a corporate survey on AI conducted by Nomura Research Institute (NRI) in its annual survey of IT utilization. Generated AI is rapidly spreading throughout Japanese companies, and in particular, about 85% of chemical and pharmaceutical companies responded that they have already introduced it (September 2025). as targets of AI application, its use for "information seeking and acquisition of knowledge and insights" and "document creation, summarization, and elaboration" were conspicuous. AI was mainly used in back-office operations such as personal document creation and information gathering, and its use in front-line (customer-facing) operations and new businesses was few and limited. In addition, the challenges in the diffusion of AI due to a lack of literacy and difficulty in seeing the effects of AI have become apparent.

Next, we will introduce the maturity stage of AI-based business automation (efficiency improvement). The maturity stage can be divided into three stages: (1) automation of content creation, (2) automation of work procedures (processes) and workflow, and (3) Agentic automation for autonomous achievement of strategic goals. (2) Automation of work procedures (processes) and workflow is the level at which data variability and exception handling can be handled automatically. (3) Agentic automation for autonomous achievement of strategic goals is aimed at reading changes in the external environment and automatically deciding on strategies and Y-rows to respond to those changes, and is currently in the research stage.

As AI matures, the roles of people working there and the way they do their jobs will change dramatically, and the era of the "Agent Boss," in which all employees become AI bosses, is approaching. The role of people will be replaced by that of supervising and evaluating the performance and behavior of Agents, leaving routine tasks to AI (Agents). People will be intentional and concentrate on more creative and strategic tasks. In addition, the organizational structure will also change, shifting from the current function-based organization to a purpose- and mission-driven organization.

Finally, there will be two phases of productivity improvement through DX: one is to improve the productivity of the value chain by using AI as a tool (change with AI), and the other is to dramatically improve productivity by changing people and organizations themselves along with AI (change with AI). The former is a continuous transformation phase of improving the efficiency of the current situation, while the latter is a discontinuous transformation that redefines functions and processes. The trend in the use of AI in the future will be to reorganize the latter functions and processes themselves.

Coexistence of Technology and Human: New Opportunities and Challenges in the Pharmaceutical Industry

Mr. Takao Suzuki, Counselor, Head of Digital Transformation Unit, CHUGAI PHARMACEUTICAL CO., LTD.

In the pharmaceutical industry, which deals with pharmaceuticals and services that affect human life and health, there is a strict business culture that requires close checking, evaluation, and collaboration with related departments and external organizations from the perspective of ensuring quality, efficacy, and safety. In addition, it takes approximately 9 to 16 years and hundreds of billions of yen to research, develop, and launch a new drug, and the probability of success is one in more than 20,000, making it a difficult business environment. Expectations are growing throughout the industry for AI as a means to break through this challenge, significantly shorten the time and cost of drug discovery, and improve the probability of success. Chugai has set its sights on "true personalized medicine" as its vision for 2030, and is promoting business transformation through the use of AI and data. We have already completed the "DX1.0" phase of laying the foundation, and are now in the "DX2.0" phase of transforming the entire value chain. The entire company is working to achieve the "TOP I 2030" growth strategy of doubling R&D output and launching its own global products annually.

With the evolution of AI, we have redefined AI as a "partner" to work with, not just a "tool. This strategy consists of three pillars, the first of which is "AI Everyday. The second is "AI Everywhere," in which business processes themselves are restructured on the premise of AI. The third is "AI Transformation," in which we are working with partners to create a system to enhance our competitive advantage in themes directly related to management issues, such as AI drug discovery. The third is "AI Transformation," which aims to create new value, including co-creation with partners, in order to strengthen the company's competitive advantage in themes directly related to management issues such as AI drug discovery.

While the use of AI is spreading throughout the company, it is also necessary to review AI governance, including the leakage of copyrights and confidential information during AI input, infringement of intellectual property rights of other companies during output, and halting of halting of Halcination. Halcination However, in an age when AI is used by all employees in large numbers on a daily basis and many AI agents promote their work autonomously, this approach will no longer be viable. As the governance of the future, there is an urgent need to establish a system in which humans can focus on more sophisticated and ethical decision-making by having AI take charge of routine business processes.

  • *
    Halcination: Phenomenon in which AI generates information that is not based on facts.

Lastly, let me talk about the human capital required in the AI era. In other words, "Business" is to have a deep understanding of one's own strengths and business, "Technology" is to constantly absorb the latest technology, and "Creativity" is to generate ideas and human empathy that cannot be replaced by AI. Integrating these three areas and cultivating human resources capable of creating new value in collaboration with AI will be the key to maintaining the competitive advantage of JPMA companies in the future.

Development and deployment of drug discovery AI platforms linked to simulation

Mr. Mitsutaka Homma, Team Director, RIKEN Center for Biomedical Research and Innovation

In addition to efficacy (activity), the profile required for pharmaceutical products is ADME ADME It is particularly difficult for small molecules to maintain good profiles for multiple items at the same time, and the optimization burden is significant. From a hit compound to a candidate molecule for development, it is necessary to optimize approximately 20 to 30 indicators simultaneously, and this is a difficult point in the drug discovery process. Since the 2000s, efforts have been made to streamline the design process by having AI learn compound structures and evaluation data to predict metabolic stability, etc. In recent years, rapid progress has been made in AI technology.

  • *
    ADME: The four processes of drug absorption, distribution, metabolism, and excretion.

On the other hand, the compound space is extremely wide (it is estimated that organic compounds with molecular weights of 500 or less have about 10 to the 60th power of variation), and actual measured data is limited, so the problem of "Applicability (AD)," where prediction accuracy drops in areas far from the training data, is inevitable. Drug discovery AI must use this point as a premise.

Against this background, from FY2020, the Japan Agency for Medical Research and Development (AMED) will play a central role in the "DAIIA *1 was launched to expand the study data by industry, academia, and government, including 17 JPMA member companies. On/Off Target DAIIA *2 Approximately 15 million data points for 487 species and over 500,000 data points with structural formulas for 30 ADMET items were accumulated, and it was confirmed that the use of corporate data improved prediction accuracy. The fact that companies provided a large number of data with structural formulas is a feature of DAIIA that is rarely seen anywhere else in the world. Furthermore, we used a combination of federated learning that is compatible with data confidentiality.

  • 1
    DAIIA: A five-year project to develop next-generation drug discovery AI through industry-academia collaboration (to be completed in FY2024)
  • 2
    On-target: In drug discovery, a drug is effective against its intended biological target.
    Off-target: Phenomenon in which a drug acts on molecules or receptors other than its on-target target

In addition, we developed a structure generation method that is not influenced by "apparent high scores" in consideration of AD, while performing multi-objective simultaneous optimization using a generation AI (ChemTS) that uses the prediction AI as a reward function. In addition, using the docking prediction AI (Boltz-2), screening of J-Public (440,000 compounds) in about 16 days and narrowing down assay candidates to 10,000 to 30,000 compounds became feasible.

Finally, to ensure that the results of DAIIA can be utilized even after the project is completed, a results version platform is provided and a framework for commercialization is in place. Participating companies can use the results version free of charge, and if they need updates or support, they can continue to use the paid version. DAIIA has been able to establish a framework that enables continuous commercialization, whereas national projects tend to stop utilizing their results after their completion.

Issues in the Patent System in Light of the Development of AI Technology

Mr. Junsuke Chimoto, Planning and Research Officer, General Affairs Division, Japan Patent Office

In recent years, the rapid development of AI technology has had a variety of effects on the intellectual property system. As AI plays an increasingly important role in the process of invention creation, we would like to introduce some of the points being discussed at the Patent System Committee, to which we have invited experts, regarding how to define inventions and inventors when AI creates the majority of inventions.

The occasion for this discussion is a case called the "DABUS case. This is a case in which a patent application was filed in various countries with the artificial intelligence "DABUS" listed as the inventor. The case was litigated in various countries, and in each country, including Japan, it was decided that the inventor must be a natural person.

The Patent System Committee mainly focuses on three issues: (1) invention, (2) inventor, and (3) eligibility of cited invention.

(1) Invention is defined as the creation of a technical idea using natural laws in patent law, but whether AI can be an "idea" or "creation" is an issue.

(2) In the case of a natural person using AI, the question arises as to what kind of contribution the natural person must make to obtain the status of "inventor" under the Patent Law, among other issues. In relation to this issue, the U.S. Patent and Trademark Office (USPTO) issued guidance on AI-assisted inventor recognition in February 2024, but withdrew it in its entirety in November 2025 and issued a revised version. The new guidance adopts the conception standard, which states that even if AI is used as a tool, if a person conceived the idea, he or she is the inventor. Since the guidance has not yet been judicially decided, it is important to keep an eye on developments. The State Intellectual Property Office of China (CNIPA) has also published a simple guideline, but its content is abstract and the details of practical application have not yet been decided. In addition, many countries, including Japan, Europe, and South Korea, have not issued clear guidelines at this time.

(3) Eligibility of cited invention refers to whether AI-generated information can be cited as a cited invention in relation to the examination of other applications. （If we take the position that AI-generated information is not an "invention" under (1), then AI-generated information cannot be cited as a cited invention as a basis for rejection of other applications under (3).

In general, there are still no internationally established rules on various issues related to AI-based inventions. Each country is discussing the issues based on provisional guidelines and court precedents, but as AI technology develops further and more and more situations involving deep involvement in creative activities increase, it will be necessary to think carefully about how the system should work. The JPO will continue to examine how the intellectual property system should work in the new era, taking into account international and technological trends.

Changes in the Pharmaceutical Industry and Intellectual Property Practice Caused by the Routine Use of AI

Mr. Hiroya Okumura, Chairperson, Intellectual Property Committee, JPMA

The first is Japan's "New AI Law," which is a basic law that aims to both promote the use of AI and address risks, and is characterized by the fact that it does not specify regulations or penalties. The new AI law is significant in that it defines for the first time the country's stance of operating flexibly based on soft law and aiming to be "the country where AI can be most easily developed and utilized.

The second is the so-called "Japanese version of EHDS. As a prerequisite for AI application in the medical field, a platform that can utilize real world data (RWD) is necessary, as is the case with Europe's Health Data Space (EHDS), but Japan lags behind in this respect, so a "Japanese version of EHDS" is expected to learn from European examples. The concept is to accelerate next-generation drug discovery and medical care through the use of EHDS by companies. After the provision of pseudonymous processed medical information to third parties becomes possible under certain conditions in 2024, the government policy will be finalized in the summer of 2026, and the bill is scheduled to be submitted in 2027.

In the value chain, we are in the process of using DX/AI throughout the company, from R&D to marketing, to first increase speed and efficiency. Compared to the financial and other industries, the pharmaceutical industry lags behind, and we understand that we are still far from the stage of "change through AI," as I mentioned earlier.

Privacy is another important issue with data. Anonymization is a trade-off between usefulness and reliability, and while the more personal information you have access to, the better it performs, it also has the potential to malfunction important principles such as purpose limitation, minimization, and consent.

Finally, regarding the use of AI in IP work, we have been surprised by the dramatic evolution of generative AI over the past year, but its use in IP work is still generally limited to partial use in daily operations. However, in the IP departments of advanced companies, AI is now being used in so-called core business operations, and in such companies, issues such as ensuring the security of confidential information are being tackled and overcome. In general, we believe that the improvement of individual skills and the establishment of an organizational structure are necessary for the widespread use of generative AI in business operations.

Panel Discussion

Moderator: Hiroya Okumura

Panelists: Hironaga Kudo / Takao Suzuki / Mitsutaka Homma / Junsuke Chimoto

 　

1. Improvement of AI autonomy and its impact on intellectual property (patents)

Moderator: With the increasing autonomy of generative AI, how do you see the possibility that AI can autonomously invent (e.g. pharmaceutical actives)?

  • Since pharmaceuticals have a high risk of halcination, etc. due to the large number of required items and insufficient amount of data, it is difficult to invent in the short term (2-3 years), but can occur in the long term (10-20 years).
  • Currently, human involvement is essential in setting issues, setting conditions for AI, etc., and cannot be "thrown in the towel.
  • As experiment automation and data accumulation advance, autonomy can be increased.

Moderator: Will the potential for autonomous inventions increase in manufacturing, formulation, medical devices, etc.?

  • Compared to drug discovery, the manufacturing and drug formulation processes have clearly defined conditions and can be easily automated and digitized step by step. However, human monitoring in quality control and safety assurance will remain important.
  • Manufacturing autonomy can be increased through simulation and digital twinning.
  • Formulation design has a high affinity for predictive models of physical properties and automated facilities.
  • Medical devices may become autonomous faster than drugs because of the ease of data acquisition.

Moderator: The AI Scientist Big data research group in the medical and health fields Can research process automation AIs such as The AI Scientist and others be applied to drug discovery?

  • Although there is potential for application, drugs administered to humans require a high level of safety, so it will be difficult to achieve this with AI's "seemingly" output alone.
  • In order for AI to move from the stage of automating relatively simple research to finishing it into a drug product with appropriate consideration of many items, a higher level than the current system is needed.
  • Drug discovery can be broken down into subtasks and there is room for partial application.
  • *
    The AI Scientist: A new AI system that automatically carries out the scientific research cycle of idea generation, execution of experiments and summary of results, and writing and peer review of papers.

Moderator: Is there any progress in inventor recognition and international harmonization (harmonization) of AI autonomous inventions?

  • Five Agencies (IP5) The AI Scientist Information is being exchanged among the five IP5 agencies, and international collaboration is highly important.
  • However, the reality is that each country is still searching for a solution, and is still in the pre-harmonization stage.
  • It is necessary to cooperate with other countries to study the system, taking into account the opinions of users.
  • *
    Five IP Offices (IP5): Japan Patent Office (JPO), United States Patent and Trademark Office (USPTO), European Patent Office (EPO), China National Intellectual Property Office (CNIPA), Korean Intellectual Property Office (KIPO)

Moderator: Is there any discussion on the possibility of different standards for inventor certification in different industrial sectors?

  • We are not aware of any public debate on the need to change the criteria for recognizing inventors in different fields.
  • However, the progress of autonomization and the significance of the contribution of "experimentation" differ depending on the field.
  • The view is that while the basic criteria should be the same, the characteristics of each field should be taken into consideration.

Moderator: If AI-autonomous inventions become a reality, how should the protection of pharmaceutical values be considered?

  • The value of a drug is established through the accumulation of clinical value, fulfillment of regulatory requirements, RWD, and scientific evidence.
  • A substance patent alone is only an idea or expected value, and it becomes valuable only when it is accompanied by information, reliability, and clinical support.
  • The value formation process and the positioning of patents should be discussed again.

2. Utilization of medical data and Ai governance logic

Moderator: In the utilization of generative AI, how do you think the quality and consistency of training data will affect the reliability of AI?

  • Differences in performance and halcyonization of generative AI depend greatly on the quality of training data.
  • General-purpose generative AI learns a large amount of net information, and performance such as reliability and consistency is ensured by quantity.

Moderator: In IP work, can we assume that what kind of data is used for AI is important for the quality and reliability of the output?

  • The source of a company's competitiveness is its internal data, and the key to differentiation is how well it can utilize data lying dormant within the company.
  • The utilization of unstructured data is particularly important, and the quality of unstructured data must be improved to enable AI learning and search augmentation generation (RAG). * and use it as inference using AI learning and search augmented generation (RAG)* to improve the quality of output.
  • There are routine and non-routine tasks in IP work.
  • Routine tasks such as patent search and application preparation can be made more efficient by using AI generation.
  • Data on past applications, decision logic for obtaining rights, IP strategies, etc. should be stored as unstructured data, and how to utilize such data by AI is an important issue for the IP division in the future.
  • *
    RAG (Retrieval-Augmented Generation): Technology to improve response accuracy by combining text generation using large-scale language models (LLM) with retrieval of external information.

Moderator: How should training data be maintained and discarded in terms of halcination and bias in generative AI?

  • Quality is particularly important in the drug discovery field due to the limited amount of data.
  • While thesis data is of variable quality and curation is limited, pharmaceutical company data is of high quality and bias can be reduced by integrating data from multiple companies.

Moderator: How should privacy risks be addressed in the growing use of generative AI?

  • Generative AI performs better with higher quantity and quality of access to personal information, but this upsets traditional privacy principles such as purpose limitation and data minimization.
  • As a countermeasure, it is important to think in terms of separating the data itself from the learning results, and the use of federated learning, non-personal data, and synthetic data can be effective.

Moderator: As AI agents become autonomous, how should governance and risk management be conducted?

  • Since highly autonomous AI may have the risk of circumventing safeguards, the key is the design of the implementation architecture.
  • AI Orchestrators. * requires that AIs be monitored under the same governance as humans, and that accountability for AI actions and where responsibility and authority lie should be clearly defined.
  • *
    AI orchestrator: A system that combines multiple AI models and related services, interconnects them, and serves as a command post to perform more advanced and complex tasks.

Moderator: How should users and organizations address copyright and intellectual property infringement risks associated with the use of generative AI?

  • The legality of training data and the availability of AI outputs are issues.
  • The question is how to control problematic behavior as an organization.
  • Discussions are underway at the government level, and a Principles Code has been published.
  • In the future, each company will need to consider its own implementation based on the government discussion.

Moderator Question: Based on what generative AI can and cannot do, how should the roles of humans and AI be divided?

  • It is important to actively utilize AI in areas where AI excels, and to divide roles where human judgment and evaluation are required, with humans taking responsibility for those areas.
  • The ultimate responsibility and authority lies with the human side, and the human-in-the-loop (HITL) HITL It is important to design the human-in-the-loop (HITL)* without excesses or deficiencies.
  • *
    Human-in-the-loop (HITL): Systems and processes in which humans are actively involved in the operation, supervision, and decision-making of automated systems.

Moderator: How do you see human roles and decision-making changing in the future?

  • There will always remain jobs that AI cannot do.
  • The role of humans is to have a will or philosophy, make decisions and evaluations, and take ultimate responsibility.
  • This role is similar to that of a CEO in a company, and people will shift to more BOSS-like tasks.

3. Transformation of IP work itself by AI and human resource development

Moderator: What are the skill sets required for intellectual property personnel in the age of AI, and what is human resource development?

  • The ability to utilize AI is essential to achieve both quality and speed.
  • People will continue to be responsible for tasks for which communication with people is indispensable, such as fulfilling accountability as a patent office and gaining the applicant's understanding.
  • Examiners will continue to need to acquire conventional legal and technical knowledge.
  • In the examination process, there are many cases where new insights can be gained through the use of AI. This will lead to improvement of the quality of examination.

Moderator: What skill sets are required of human resources in the age of AI?

  • Both the ability to use AI as a tool and the ability to appropriately consider a large amount of information to determine a direction are important.
  • Because pharmaceuticals require strict profiling, people will continue to be responsible for setting issues. People will be required to have the ability to do this.
  • It is difficult to learn the above skills at educational institutions. Companies will be responsible for human resource education.

Moderator: What do personnel who integrate disciplines and personnel who can lead integrated collaboration across divisions have in common?

  • Human resources with the ability to correctly position IP in the company's business strategy will be required. In addition, the ability to take on new challenges is also necessary. To achieve this, it is important to improve human ability as a basic skill. These cannot be replaced by AI.
  • The pharmaceutical industry is an industry that requires a great challenge to reform conventional business practices. Business transformation through the fusion of technology and business, which is evolving day by day, is fraught with difficulties. It is necessary to have human resources who can think about it and promote it on their own.

Moderator: What are the goals of IP departments in the age of AI?

  • In the future, the IP division will be the engine of corporate value creation. I expect the IP division to lead the strategic use of IP and become an organization that improves the value of the entire value chain.
  • We expect the IP department to be an organization that visualizes, manages, and enhances the company's management resource called IP.

Moderator: What is the ecosystem in the pharmaceutical industry where AI will serve as a base?

  • There must be non-competitive areas where the industry as a whole should cooperate, and AI can be a starting point for efforts to unravel the distortions that exist in the industry.



(Intellectual Property Committee, Intellectual Property Forum TF)

Share this page

TOP

Site Search