JDMC's efforts

First, let's introduce the efforts by the compliance study group for AI and data utilization at JDMC. The theme of our activities is to discuss from the perspective of compliance the issues that have arisen as the scope of data utilization expands due to the acceleration of the digital economy, as well as points of focus for business expansion.

While collecting and properly managing data will lead to full-fledged data utilization, mishandling of data poses a major risk to management. In order to promote AI and data utilization, we believe that it is necessary to strictly handle security, compliance, and personal information protection through internal regulations and contracts with external parties. The concept is to consider practical measures and conduct research that forms the basis for strategic utilization.

Mr. Sato of SBI Holdings says, "The purpose of the study group is to look at the stumbling points from the perspective of compliance and ethics as the use of data and AI becomes commonplace in various companies." The study group is now in its third year. In the first​ ​year, we organized the overall picture of points from a compliance perspective. "This year, I would like to clarify what is necessary to find a concrete solution," said Mr. Sato.

Things to be aware of when using data with AI

Regarding the things to be careful not to cause problems in data utilization by AI,8I will mention one point. Contents of acquired data, storage location of acquired data, content of derived data, destination and number of stages of derived data, how to utilize derived data, data deletion operation, security measures, terms of use and policies in data utilization We will introduce it while referring to the diagram below based on the flow of data.

again,NECMr. Yasui ofAIIn order not to cause problems in data utilization usingAIIt is necessary to understand that using , may produce results that exceed human expectations." On top of that, to be able to answer why it happened,AIWhen considering the results of the above, "it is necessary to consider that some people may point out discrimination and prejudice" (Mr. Yasui).

What is the study group's definition of a "stumbling point"?

As we leverage new technologies in our business, we often stumble because we don't see it yet. What could happen with AI?

Let's take a look at the details of the stumbling points defined by the study group.

The reason why unexpected problems arise in data utilization is that there are cases where it is not necessarily against the law, but it is subject to social criticism. In addition, legislation has not kept up with derived data generated through data integration and processing, and it is a repeated situation where laws and regulations are developed to deal with problematic cases after they occur. Kamishima says that the entire study group shares the idea that ``to find stumbling points, we need a methodology based on a broader ethical perspective that goes beyond legal compliance.''

An ethical framework identifies stumbling points

How can we identify stumbling points in the field of data utilization? The research group is creating a tool called "Ethics Framework".

倫理フレームワークでは「実践可能であること」と「有用であること」の2つを目指しています。実践可能であることを担保するために、チェックリスト形式を採用し、活用ガイドとともに提供することで、データ活用に携わる実務者自身が利用できるようにしました。また、データ活用の想定事例も用意し有用性を高めました。用意した仮想のデータ活用事例に対し、ツールを用いることで、過去の事例だけでなく、まだ指摘されていない、潜在的なつまずきポイントを事前に抽出できることを検証しています。

In creating the ethics framework, we have made the following efforts.

First, set up a "checklist" item to prevent situations like the example. Next, the stumbling points are derived by extracting factors common to multiple checklist items. Further find the ethical perspective behind the stumbling points. On top of that, we will supplement the stumbling points and checklist items again from discussions based on actual experience. By repeating this cycle using induction and deduction, we brush up the framework.

Structure of the ethical framework

Regarding this ethical framework, the ethics that will not change in the future are the legitimacy, accountability, fairness, safety and security, and information protection on the left side of the diagram above. On the other hand, the right side is the stumbling block that occurs in relation to each ethics, and Mr. Yasui says that this will change as the times progress.

So, what exactly should we check for each ethic?

Accountability and fairness

This is whereAIcomes into play. The key is to be able to convincingly explain the scope of the original data, the presence of bias, algorithms, etc. It is necessary to understand the limits of derived data, the risk of malicious data being mixed in, the possibility of discriminatory acceptance of good and bad expressions, and the fairness of data provision methods.

Information protection

In terms of information protection, it has become necessary to pay attention to the spread of SNS, especially recently. This is because while you can write comments as you like, if you compare characteristic data when publishing analysis results, you may be able to identify people.

What kind of businesses do the members of the research group who took the podium use AI for? "Many financial institutions have expressed interest in using it. At the research group, we tell them to 'become a member and try it out,'" says Mr. Sato. Based on the collected information on AI utilization, we have created a virtual example of a monitoring service using smart meters so that we can see the benefits and precautions of AI utilization.

Actual use of data by AI

We asked them how they felt when they actually applied AI.

“Certainly, when you fill out the checklist of the ethics framework, you can see a lot of things. We were worried that it would not lead to more sales," said Mr. Sato. Considering the relationship with the family, it is said that simply obtaining consent is not a good thing. In addition, due to the misuse of information obtained by AI, employees may half know the bedtime of each household, which may lead to a deterioration in public security. will be

Technology is evolving quickly, the amount of data we handle is increasing exponentially, and new suggestions are being discovered. This does not mean that you can use anything as long as it does not violate the law. Understanding this will lead to the realization of a prosperous society. Macnica hopes to utilize the knowledge gained through participation in study group activities to implement technology in society in a safe and secure manner.

Plan from now on

Lastly, regarding the future of AI utilization, Kajima said, “I am in charge of the pairing of accountability and fairness in an ethical framework. It is also true that the results of data utilization can make some difference, but we still need to explain that it is fair."

In addition, Mr. Sato is in charge of the validity of this year's ethics framework, and said, "In many cases, ethics are qualitative and not quantitative. rice field. Mr. Yasui is in charge of safety and security and information protection in the ethics framework. He says that the underlying theme is data utilization. "Since personal information and personal information may be included, I would like to be aware of what is important even in the foundation."

Macnica offers implementation examples and use cases for various solutions that utilize AI. Please feel free to download materials or contact us using the link below.

▼ Business problem-solving AI service that utilizes the data science resources of 25,000 people worldwide
Click here for details