Data Nugget September 2024

September 30, 2024

We are delighted to bring the latest edition of the Data Nugget. So grab a cup of coffee and start your Monday with some fresh news from the data management world.?

First, we have an interesting read about the increased popularity of AI strategy versus data governance. Second, we have a quick summary of ways to unlock success with a data governance framework. And last but not least, we have the next episode of?the Season 3 podcast?on data management as a code.

Have a great start to the week and enjoy?reading!

Let's grow Data Nugget together. Forward it to a friend. They can sign up here to get a fresh version of Data Nugget on the last day of every month.

Nugget by Achillefs Tsitsonis

AI "VS"?Data Governance. Or maybe not?

The exponential rise of AI focus and popularity in recent years is mainly attributed to breakthrough developments in Large Language Models (LLMs) and Generative AI. With this shift, there are a lot of discussions around all activities related to these fields.

Terms like AI Strategy and AI Governance have risen and are trying to shape their vision/form and at the same time, more traditional terms like Data Strategy and Data Governance are being put more on the side. Without denying the importance of defining an AI Strategy and establishing AI Governance around all AI-related business goals and activities, a reasonable question arises. How much different is AI versus Data Governance, what is the exact purpose of each one and where should one put their focus?

The two focus areas are not different at all though and are, actually, very much intertwined. I would also argue that AI Governance/Strategy are but subsets of the Data Governance/Strategy and that sole focusing on the AI part will most probably lead to a non-optimal if not unsuccessful, implementation of such endeavors. AI Governance discussions are highly focused on ethical, societal, risk assessment and model performance aspects. All of these areas, however, have always existed within the Data Management sphere and in collaboration with other principles like Data Quality, Master/Reference and Metadata management, they need to work not in isolation but in close cooperation not only to achieve the goals of your AI initiatives but to lift the value of your data throughout the whole organization.

For some further food for thought on the subject, one can read the following article which explains further why Data Governance is the foundation of AI Governance and how the two areas complement each other. ???

You can read the full article here. Credits to Stefaan G. Verhulst and Friederike Schüür at Data & Policy Blog.

Nugget by Nazia Qureshi

Unlock Success with Ultimate Data Governance Framework

The article written by Rui Manuel Pereira?talks about a comprehensive data governance framework that would be required for handling increased volumes of business data. It begins with 'clear ownership' and 'accountability,' whereby clear roles are designated, such as data owners and stewards, who take responsibility for the accuracy and integrity of data. Following that, Data Quality Management ensures consistency in data and cleanliness through audits and automation tools. ?

Cataloguing and classification help organize data, which again helps in classifying sensitive information. Similarly, policy and regulatory compliance serve to meet various legislations: GDPR and HIPAA among many others. The security of sensitive data is protected by encryption and access control that helps sustain these standards for their security. ?

Effective communication and change management focus?participation at all levels within the organization to make it agile, ensuring that the value of governance is explained to all stakeholders. Finally, the framework emphasizes monitoring, metrics, and continuous improvement, maintaining a set of performance indicators that would monitor success and make changes accordingly. This structured approach will ensure an in-depth evolution wherein organizations protect their data assets while meeting regulatory requirements. ?

It is a summary that allows ownership, security, compliance, and flexibility of data at the core to make smart decisions, innovate, and build trust in data.

You can read the full article here.

Nugget?by?Winfried Adalbert Etzel

MetaDAMA 3#3: Lars Albertsson - Data Management as Code

"There should be very little reason to say: Hey, I need a human to look at these operational things for me. They are all defined as code." Lars Albertsson has a long career in Data and Software Engineering, including Google and Spotify. Lars is on a mission to spread the superpowers of working with data, with the vision to?'enable companies outside of the absolute technical elite to work with data with the same efficiency or effectiveness as the technical elite companies in an industrial manner.'

Four types of companies:

1. ?Born digital - Data is the basis of their business model.
2. Born digital in a traditional market - completely natural to use data as a competitive advantage.
3. Traditional industries 'born before the internet'?- big difference whether they handle information or are in the physical world.
4. Information Handlers - Banks, Media, etc. have digitalized their whole activity chain long ago.

The differences

• Significant differences in cycle-time in different industries and businesses.
• The only way to beat this cycle is to try out, fail fast, learn, try again.
• 'Successful companies have been really good at failing fast.'
• Fast-moving cultures are more effective and therefore have a better risk focus, without slowing down.
• To move fast in a slow-moving industry, you need to choose your technology and approach wisely, keeping complexity down.
• Cultural slowness - 'The challenge to change the way people work and people think is extraordinarily difficult.'
• Risk and Governance are addressed by rituals rather than tasks.
• The value chain data to client outcome, needs to be anchored in a company. Have a clear picture of what this means.

Getting close

• Success can be measured by how close you are to the end user. The closer you get to a customer, the better the chances of success.
• 'There is no substitute in value creation than talking to the people you actually want to make happy.'

Automation is Innovation

• You need to find ways to ignite people's domain innovation capacity.
• Automation is a gradual process. People don’t lose their work to machines overnight.
• Human oversight is still really important, and there is a long journey with humans as part of the process.
• The focus on automation now is on knowledge workers, yet those have a different stand in society and can resist better compared to the workforce during the Industrial Revolution.
• 'If it changes quicker than one generation, there won’t be natural attrition that matches the changes in the need of the workforce.'

Automated Data Management

• Automating and industrializing data management processes is lower risk than software development, but still not as common.
• Great value to gain?from delaying simple automation processes to data management.
• You need to build everything from raw data to end product to find ways to automate.
• The raw data is the soul of the end product and the other way around. You need to keep these two outer points of the pipeline in mind when think of data quality and data products.
• The limitations in Hadoop forced to work in a certain way. That way can be adapted to data management.
• Hadoop really pushed people in the functional Big-data patterns?that are still the basis of much of the work we are doing today.
• Workflow orchestration can help to know?which data you choose for a certain computation.
• Data Management as code is an area that is underdeveloped and under-appreciated.
• Minimize the technical barriers from Governance?and focus on the social aspects.

You can listen to the podcast?here?or on any of the common streaming services (Apple Podcast, Spotify,? etc.)?Note: The podcasts in our monthly newsletters are behind the actual airtime of the MetaDAMA podcast series.

Thank you for reading this edition of Data Nugget. We hope you liked it.

Data Nugget was delivered with a vision, zeal and courage from the editors and the collaborators.

You can visit our website here, or write us at [email protected]. I would love to hear your feedback and ideas.

Nazia Qureshi

Data Nugget Head Editor