Using Scalable and Responsible AI to Contribute to Lifelong Health, Safety, and Comfort for Everyone

The release of OpenAI’s ChatGPT in November 2022 generated a lot of interest and excitement. In the wake of the announcement, as people and companies began to explore the technology, attention shifted to making AI safe to use and easy for ordinary people to understand, leading to greater interest in the concepts of “Responsible AI” and “Scalable AI.” According to Deloitte’s Survey on AI Adoption in Manufacturing, 93 percent of companies believe AI is necessary to drive growth and innovation in the industry. As advancements in AI continue to be made, the importance of both Responsible AI and Scalable AI will grow.

Panasonic’s professionals are active in a wide range of business sectors, and look to AI as a way to address and resolve problems while delivering value to customers. The Group focuses on the development of technologies that have a positive impact on the environment and contribute to the lifelong health, safety, and comfort of each customer. Panasonic’s AI technologies include Taimen-Honyaku, a real-time translation solution that allows people to speak face-to-face in different languages; facial recognition solutions that have been introduced in Japanese airports; walking training robots that help the elderly improve their mobility; and autonomous-driving robots. 

Panasonic is committed to becoming a company that understands the lives of its customers, and its goal is to identify the most suitable solutions and deliver valuable products that fulfill customers’ needs while adhering to ethical standards.

What AI technology is Panasonic focusing on?

Right now, Panasonic is focusing on the speedy acquisition and deployment of cutting-edge AI technologies, leveraging its world-class technological and business development capabilities to deliver greater value to customers.

The Group doesn’t just focus on the research and development of AI—it also focuses on developing AI technologies necessary for its business operations and on approaches that can increase the number of people who are able to use AI. Currently, there are over 1,500 AI engineers at Panasonic.

Panasonic can also use AI to respond to new opportunities that are emerging from customers’ ever more diverse lifestyles. To better understand the needs of customers, the company began consolidating product usage data via Panasonic Digital Platform (PDP) in 2014, utilizing the data in a variety of products and fostering cross-organizational collaboration that makes the most of the Group’s knowledge and expertise.

With the establishment of the operating company system, each business is strengthening its ability to understand and respond to the needs of its customers. Frontline staff are being encouraged to make full use of AI to provide products and services that enhance customers’ lives and benefit the environment.

What kind of initiatives are being taken to increase the number of people in the company who can use AI?

Panasonic provides curricula on AI tailored to employee levels and department needs. Courses are also offered by university professors and other experts who have been invited to work with Panasonic. Kaggle Grandmaster* Ryuji Sakata, a Panasonic employee, offers the Advanced Data Analytics Course to Panasonic employees. Additionally, the Group regularly shares information through an online, group-wide AI human resources community with over 1,500 members and holds a multi-day AI event once a year to share initiatives and latest discoveries in the technology.

*A Kaggle Grandmaster is someone who has achieved the highest-level proficiency category in the Kaggle online learning platform

How is AI being used within Panasonic?

In February this year, B2B solutions specialist Panasonic Connect rolled out an AI assistant—ConnectAI—to all employees in Japan. Within three months of implementation, the service has exceeded 260,000 total uses, averaging 5,800 uses per day. In April, using ConnectAI as the model, the company began implementing PX-AI for all Group employees, providing an environment where employees outside of the IT department can interact with the cutting-edge technology. With these conversational AI tools in place, employees are able to learn how to efficiently use AI while keeping up with the rapid advances being made in this technology. As of July, PX-AI has exceeded approximately 700,000 total uses for over 30,000 employees.

Is there a knowledge gap between corporations and academia? How about research and field applications?

“We should narrow the gap between the AI level of actual industrial applications in factories and that of researchers in our laboratories. At the same time, Panasonic should keep up with the world’s top-level studies to ensure that we can effectively make use of the latest AI technologies in our business. For this reason, our motto is: ‘Reaching the top and bridging the gap.’” says Prof. Tadahiro Taniguchi, who joined Panasonic as a visiting general chief scientist in 2017. 

In addition to its collaboration with Prof. Taniguchi, Panasonic has also partnered with researchers such as Prof. T. Yamashita of Chubu University on the company’s internal research group—REAL-AI—and maintains active research partnerships with North American universities like Stanford University and University of California Berkeley.

What is “Scalable AI?”

The process of developing AI models has made it clear that progress is being held back by two specific hurdles—the quality and quantity of data to be trained by AI; and the real world introduction of AI models and their implementation in devices. “Scalable AI” aims to overcome these hurdles and promote practical application of AI.

AI model development requires large amounts of high-quality data, but open-source datasets are limited. If high-performance AI models can be developed even with a modest amount of data, the range of applications will expand. Panasonic has developed technologies that can significantly reduce the amount of data required for training while maintaining performance.

Creating autonomous robots that can actively explore the environment is time-consuming, costly, and requires a controlled environment. Panasonic Holdings and Panasonic Connect are collaborating with Prof. Taniguchi on the development of a “world model” AI that learns about its environment on its own.

Meanwhile, Panasonic’s independent subsidiary Blue Yonder, a supply chain management company, and Panasonic Connect are working together to provide AI/CPS solutions that can be delivered quickly to customers, such as an Autonomous Supply Chain™.

What is “Responsible AI?” How does Panasonic ensure safety, privacy, and human rights?

Throughout AI’s evolution, ethical issues have come to light that researchers did not anticipate when developing their products. “Responsible AI” is the solution to help customers feel at ease and build trust in the product while using the new technology. Panasonic is strengthening its AI ethics governance, and in April 2019, established the AI Ethics Committee. From there, the Group released its AI Ethics Principles in April 2022, began providing AI education for all employees in September 2022, and introduced an ethical risk assessment system for any product or service that makes use of AI technology in November 2022. Every product Panasonic develops must submit to extensive quality checks during testing and is retrained when quality fluctuations are detected.

In May 2023, Yoky Matsuoka, Managing Executive Officer, Panasonic Holdings, delivered a keynote speech on Responsible AI at the IEEE International Symposium on Circuits and Systems 2023 (IEEE ISCAS 2023). Learn more about her approach to Responsible AI, “For Humans, By Humans, With Humans,” here.

• For Humans—Always put consumers first. Check on the actual effect for people and the collateral damages by technology.
• By Humans—Humans remain in control and in the loop. AI used to make humans “superhumans”
• With Humans—Hire, train, and nurture ethical and responsible people. Protect human well-being and community.

To ensure fairness and that no discrimination arises in the use of Panasonic products, the Group offers an AI Ethics Basic Course to all employees and a Practical Course to ethics staff, developers, and facilitators. Panasonic has achieved Responsible AI with reliable technology by approaching AI reliability from both the elemental technology and development process perspectives. Panasonic makes sure that the AI is Explainable AI (AI in which humans can understand the reasoning behind its decisions), utilizes Out-Of-Distribution detection to certify a model’s ability to handle unforeseen situations, and maintains quality assurance MLOps processes to make AI development reliable and productive.

Hiroshi Kutsumi, Director of the Digital and AI Technology Center at Panasonic says, “Going forward, Panasonic will continue contributing to the environment and to people’s lives by focusing on Scalable AI and Responsible AI. It will also work to increase the number of people who have access to and can use AI technology. Panasonic AI is For Humans, By Humans, With Humans, and it will continue to deliver AI-equipped products that tackle its customers’ needs and that they can trust.”