<< back Page 2 of 3 next >>

Opportunity and Threat: The Intersection of AI and Data Governance

The problem is that generating insight is not the same as understanding it. And generating insight isn’t easy either. Enter AI, which is now being deployed to generate insight about, well, insight. The use of machine learning and natural language processing provides the ability to interact with data organically as well as to surface valuable or unusual trends in an automated fashion.

Today, augmented analytics (aka AI-?driven analytics) is being deployed for a variety of reasons, including the following:

  • Enabling access: Can’t remember which dashboard reports what? Stop hunting and just say, “Show me sales.” Augmented analytics allows users to ask for information and receive responses in an intuitive manner—via text or speech.
  • Promoting understanding: A picture is worth a thousand words. Except when the logic behind the picture may not be clear. Worried your users don’t understand the statistical and logical underpinnings of your findings? AI can generate a visualization and an accompanying explanation of the findings in plain English.
  • Expediting insight (for data scientists): Is this the best model? AI can tell you. From suggesting the best algorithm to auto-tuning parameters and feature selection based on the attributes of the data to executing model bake-offs to confirm the best fit, AI can help expedite insight generation.
  • Generate insight (for everyone else): What do you need to know now? AI-?augmented analytics systems analyze business metrics of interest dynamically, without user prompting. AI can be used to perform ongoing analysis and raise the alarm when unusual patterns or outcomes may require attention—without user prompting. Over time, the systems learn which metrics are most interesting (actually used) and adjust output accordingly.

The Threat: Three Ways AI Heightens the Need for Data Governance

  1. Data Literacy

Conversations regarding data literacy often focus primarily—if not solely—on policies, procedures, methods, and tools. While necessary, true data literacy extends beyond nitty-gritty data management practices or the internal workings of the latest AI algorithm.

Ultimately, the value of data comes not from the amount accumulated but from the value created. In fact, this is the mission of data governance: to maximize value and manage risk. For the business to actively engage and adopt a data-driven mentality, its people need to understand why data matters, which means they must understand how and when AI can—and should—be applied.

For more articles like this one, go to the 2020 Data Sourcebook

To achieve that goal, literacy programs must expand beyond technical and compliance training to include the following activities and processes:

  • Asking better questions: Enable business stakeholders to identify analytics opportunities and embed data awareness into daily decision making. Articulating the types of problems that analytics can address using simple metaphors and demonstrating potential applications is a good way to start.
  • Telling a better story: All insight, and no action? Business and technical stakeholders from the CEO to the engineer on the production floor must be able to effectively communicate results in a compelling manner—one that uses the data as an input rather than the end of the story.
  • Defining good enough: Balancing incremental improvements in data quality or model performance against incremental gains in business value or risk reduction.
  • Having accountability: Today, data is created and consumed by too many parties for centralized control. In the new paradigm, accountability for using data appropriately (i.e., in compliance with external regulations and internal policies) belongs to the data consumer. Likewise, data producers must be accountable for making data readily available to the enterprise. Achieving this requires a distinct mindset shift, clearly defined policies, and pervasive training on the rules of the road.
  1. Digital Guardianship

A mantra for moving from compliance to guardianship is moving beyond the idea of “not doing what is wrong” to “doing what is right.” Digital guardianship speaks to the establishment of an open, well-understood, and agreed-upon use of information in the service of the customer and the company—with the emphasis on the customer.

<< back Page 2 of 3 next >>


Subscribe to Big Data Quarterly E-Edition