Scalable × Responsible: Panasonic’s AI Future

From everyday products and essential social infrastructure to R&D, Panasonic Group AI is making an impact across diverse fields. With over a century of manufacturing excellence, Panasonic advances AI through two core concepts: Scalable AI, which leverages world-class development capabilities to accelerate innovation, and Responsible AI, which ensures safe and trusted AI deployment and use in society. Together, these concepts drive Panasonic’s effort to transform society through AI.

Panasonic Group AI is designed to deliver reliability and trust and seamless integration into daily life. Our goal is intuitive ease of use without special expertise. For example, Panasonic refrigerator AI Camera recognize food products, automate stock management, and help reduce household food waste. But Panasonic AI isn’t only for home appliances. We’re adapting AI to a wide range of social infrastructure and business use cases, from facial recognition systems for airports to AI systems that can read floor plans and calculate construction costs instantly. In the laboratory, Panasonic AI is automating the exploration of new materials, cutting product development turnaround significantly and delivering meaningful results.

Panasonic’s ability to apply AI to diverse contexts comes from our history of being early to integrate AI into product functions that users experience directly. For example, our automated facial recognition systems build on AI developed to automate cameras. Panasonic Holdings AI engineer Kazuki Kozuka explains, “To equip small devices like cameras and home appliances with AI, we had to evolve progressively sophisticated models that responded flexibly even when faced with challenges we hadn’t anticipated. Moreover, we had to achieve this under constraints of limited processing and current capacity. Consequently, we have accumulated technologies and expertise that achieve high performance and safety despite these constraints.”

Traditionally, AI models often needed to be trained separately for each application, whether ingredient recognition for refrigerators, facial recognition for airports, floor plan analysis for housing, and so on. Building each AI requires massive amounts of image data and annotation, with development turnaround and cost as major obstacles.

Another Panasonic AI model, Diffusion-KTO, addresses the shortage of training images for AI development. By incorporating human feedback, the system synthesizes training data. Panasonic’s advanced AI models are now gaining recognition at such top-tier international AI conferences as ICLR 2025.

Panasonic Group prides itself on world-class development speed and technical prowess, but Responsible AI is our highest priority. Ensuring that our AI models can be used safely, reliably, and with trust is a non-negotiable commitment. Because AI models can draw incorrect decisions when facing unknown conditions, it is important that these models don’t pretend to know what they don’t know. To this end, Panasonic’s FlowEneDet is a generative AI model designed to estimate uncertainty in image recognition results and prevent misclassification of objects when an AI model has never seen them before. FlowEneDet helps AI models recognize when objects are genuinely unknown, and safely stop or notify users without malfunctioning.

A traditional dilemma in AI development is the tradeoff between accuracy and processing speed. Panasonic’s SparseVLM technique overcomes this dilemma by focusing only on truly important information, much the way people do. SparseVLM delivers speed with intelligence, which were once considered incompatible goals, and enables fast, precise decision-making.

Panasonic Group isn’t just developing AI models. We are using AI to create products and services. This integrated strength is unique to Panasonic as a manufacturing group. We are further expanding our AI advantage to deliver greater value to customers, and the launch of Panasonic Go is another indication of this direction. Panasonic Go is an important group-wide project aimed at driving business transformation and enhancing customer value through AI.

Technology development based on the dual concepts of Scalable AI and Responsible AI is the foundation of Panasonic Go. Panasonic has long refined AI through products tied closely to daily life. Now we are extending this expertise to major technologies that support core functions of society. With scalable development, responsible integrity, and expertise in implementation honed over a century of advanced manufacturing, Panasonic Group will keep delivering groundbreaking AI models and AI-enabled products and services that justify customer trust.