In our last post, we talked about data governance, which companies that want to utilize AI and big data should prepare for. Data governance means establishing the rules that must be followed in the data collection process and then performing activities such as data collection based on those rules.
To create a well-organized data governance, you need to decide what level of data will be maintained and specify the R&Rs that will control this level. You'll need to establish metrics for measuring whether data is being managed well, and you'll also need to decide who will define the aforementioned terms when new data arises. Having these rules in place, along with any security or legal interpretations around data collection, will go a long way in helping your organization leverage its data assets in the long run.
There are two high schools located in Seoul and New York. If you were to average the grades of the students in both high schools, would the students in the high school with the higher average score do better? It doesn't make sense to make an absolute comparison based on the scores of students who were taught different things by different teachers, because at least it would be fair to give students in both schools the same test questions and measure their scores.
But as a set of rules for managing an intangible asset, data governance still feels abstract. How can you make data governance work for you?
In recent years, many companies have been making bold investments in AI and big data. Despite this, it is not easy to find the performance that matches the investment. There are many reasons for the failure of this challenge, but the most common ones can be summarized as "lack of competence of data managers", "lack of analyzable data", "low reliability of analytics infrastructure",and "immaturity of related technologies".
Experts point out that accurate diagnosis of the causes is necessary to prevent investment in AI and big data from failing due to the above reasons. In particular, the lack of competence of data managers and the lack of analyzable data can be minimized by establishing data governance.
In the past, data was more about collecting. The downside of mindlessly collecting data was that it didn't have much utility. It's like a squirrel storing acorns, but forgetting where they were stored over the winter. But if we want to use data to uncover new value and help us make decisions, we need to move beyond collection and storage to a structure that takes utilization into account.
You also need to have an accurate diagnosis of the current state of your data. Next, you need to clearly define the value you want to get from your data. By defining the purpose of data utilization, you can understand the current state of your dataand plan for the future. In the past, it was enough to diagnose the current status of the informationization strategy and monitor the improvement points, but if the information collected is not to be used once and then destroyed, it is necessary to think about how to operate the data assets on an ongoing basis.
To summarize, the success or failure of the AI and big data industry depends on designing a sustainable data business. The two risks described above can be best managed through a data governance framework. A data governance framework defines the goals, policies, principles, standards, and procedures that should be addressed with data, respectively. It makes the operational and production management of data more stable and organized.
With a data governance framework, we've been able to address two of the four causes of failure in AI and big data. But how do we address the other two? We've explained that successful data governance requires setting some rules and compensating the people who do the R&R.
However, due to the nature of data activities, which deal with intangible values, invisible hard work is likely to be 'slapped down' in the face. In a similar case, a company fired their entire IT team because they were not experiencing any IT-related issues, and then a series of incidents occurred at the same time, causing chaos. If you want to leverage data for your business in the long run, you need to have a good understanding of the people who will be managing it.
In particular, you need a fluid view of data. Even when you have solid data governance in place, it's not forever. Your goals and the business environment around you change.This means that the definition of the data we need will continue to change as well, so your data stewards, including your data governance framework, must be fluid enough to respond to these changes.
But it's not easy to rebuild an underperforming or aging analytics infrastructure. Even if you have a long-term plan to utilize your data, if your data is incomplete due to a sloppy infrastructure, the AI trained on that data will be unreliable. There's nothing more frustrating than investing heavily in a platform only to have it not perform well.
That's why experts recommend hiring data management talent internally or partnering with a trusted company, especially if you need specialized skills to effectively utilize the data sets you've collected. If you're looking to leverage your data assets for your business, and you're looking at the long-term future, it's important to have an expert partner that you can rely on for the long haul, because no matter how big and specific your goals are, you're not going to get good data if you don't have the right technical solutions or if you don't have the right people.
In the AI-ML industry, data management has traditionally been considered a low-level task.However, as the value of data is becoming more and more important to businesses, the importance of people who can create good data is being emphasized every day. This is why data processing is just as important as data governance.