About R4D

We consider the positive impact (avoided emissions) generated through Mercari’s business by quantifying the environmental burden reduced when buying secondhand items rather than buying new items. To quantify positive impact (avoided emissions), we develop methods to calculate and visualize the extent to which transactions of secondhand items on the Mercari marketplace app contribute to reducing environmental burden. Mercari publishes these results in its annual Impact Report, which summarizes its efforts and achievements toward realizing a circular economy.

As a further research topic, we are also developing an automatic estimation model for environmental burden by product in collaboration with the University of Tokyo Research Institute for an Inclusive Society through Engineering (RIISE) Value Exchange Engineering Research Program. By utilizing multimodal AI, we are building a system that estimates greenhouse gas emissions per product from unstructured data such as images and text, and presents the results in real time on a user interface. This will enable us to achieve the once difficult task of visualizing the environmental impact of a vast amount of items in the secondhand market. We aim to create a system and culture where consumers naturally make the decision to dispose less and reuse more as a result.

We are working on research to help sellers estimate the size and weight of their package when listing items. This reduces confusion regarding packing methods and shipping costs that arises during listing, leading to a smoother transaction experience. In this research, we focus on the listing process as the subject of study, paying attention to points that are likely to generate confusion or stress for sellers when they select package sizes or attempt to understand packing procedures. To address those points, we are researching approaches to estimate the dimensions and weight of items from images using computer vision technologies and visual language models, such as 3D object detection models that operate on mobile devices. Additionally, we are working on developing a presentation method that connects the estimation results with guidance expressions through augmented reality, allowing users to intuitively understand and execute the packing process. Through these efforts, we aim to create a marketplace experience where anyone can list items smoothly and at an appropriate cost.

Through collaborative research with the Osaka University Research Center on Ethical, Legal and Social Issues (ELSI Center), we have organized and analyzed the ethical, legal, and social issues (ELSI) associated with the real-world implementation of advanced technologies such as AI from the perspective of the humanities and social sciences. We have also translated these insights into forms that can be used in corporate settings, and applied them in the formulation of the Mercari Group AI Usage Policy.

As part of our collaborative research, researchers from the ELSI Center have been contributing advice on revising this policy in response to changes in technological trends, social conditions, and feedback from both inside and outside the company. This policy serves as a foundation for maximizing the potential of AI while appropriately managing the risks, ensuring trustworthy and responsible use of AI.

We have also published academic papers regarding our collaborative research initiatives, contributing practical knowledge on corporate AI utilization and AI governance to society at large.

Furthermore, in anticipation of the rapid spread of generative AI and product development utilizing large language models (LLMs), we have moved quickly to establish and publish guidelines specifically focused on LLM usage to ensure safe and speedy development from an early stage. (These guidelines are currently being reviewed and updated.)