Your organization can use data governance for AI/ML to build the foundation for innovative data-driven tools.
Data Management ensure that data is available, consistent, usable, reliable, and secure. It’s a concept that organizations grapple with, and the premise will be heightened as big data and systems like artificial intelligence and machine language come into the picture. Organizations quickly realize that AI/ML systems work differently from traditional, fixed-record systems.
With AI/ML, the goal is not to return a value or state for a single transaction. Instead, an AI/ML system sifts through petabytes of data in search of an answer to a query or an algorithm that might even be a bit open-ended. Data is processed in parallel with streams of data being fed to the processor concurrently. Large volumes of data that are processed concurrently and asynchronously can be pre-removed by IT to speed up processing.
UNDERSTAND: Hiring Department: Database Engineer (TechRepublic Premium)
This data can come from a variety of internal and external sources. Each source has its own way of collecting, managing and storing data – and it may or may not conform to your organization’s governance standards. Then there are the recommendations of the AI itself. Do you trust them? These are just some of the questions that companies and their auditors face as they focus on data governance for AI/ML and look for tools that can help them.
How to use data governance for AI/ML systems
Make sure your data is consistent and accurate
If you are integrating data from internal and external transaction systems, the data must be normalized so that it can be communicated and combined with data from other sources. Application programming interfaces are made available in many systems so that they can exchange data with other systems facilitating this. If APIs are not available, you can use ETL toolconvert data from one system into a format that can be read by another system.
If you’re adding unstructured data like photo, video, and audio objects, there are object linking tools that can link and link these objects together. A good example of an object linker is a GIS system, which combines photos, diagrams, and other types of data to provide full geographic context for a particular setting.
Confirm your data is usable
We often think of usable data as data that can be accessed by users – but it’s much more than that. If the data you retain has lost its value because it is out of date, it should be deleted. IT users and end businesses must agree on when data should be deleted. This will appear as data retention policies.
There are also other cases when AI/ML data should be deleted. This happens when the data model for AI is changed and the data no longer fits the model.
During an AI/ML governance review, testers would expect to see written policies and procedures for both types of data deletion. They will also check that your data deletion operations are in compliance with industry standards. There are many data purification tools and utilities on the market.
Make sure your data is reliable
Circumstances change: AI/ML systems that used to work fairly well can start to lose their effectiveness. How do you know this? By regularly testing AI/ML results against past performance and against what is happening in the world around you. If the accuracy of your AI/ML system is slipping out of your hands, you have to fix it.
Amazon’s hiring model is a great example. Amazon’s AI system concluded that it was best to hire male candidates because the system was looking at past hiring practices and most of the hires were male. What the model failed to adjust to move forward was the larger number of highly qualified female applicants. The AI/ML system has moved away from the truth and has instead begun to instill recruitment bias into the system. From a legal point of view, the AI did not comply.
UNDERSTAND: Ethical policy on artificial intelligence (TechRepublic Premium)
In the end, Amazon perfected the system – but companies can avoid these errors if they regularly monitor system performance, test it against past performance, and compare it to what’s going on in the outside world. If the AI/ML model is out of sync, it can be adjusted.
There are AI/ML tools that data scientists use to measure model deviation, but the most direct way for business professionals to check for bias is to cross-compare AI/ML system performance with historical performance. For example, if you suddenly find the weather forecast is less than 30% accurate, it’s time to check the data and the algorithms your AI/ML system is running.
