An Imperative of AI Adoption: Centralized Sources of Governed Truth
A Q&A with Cody Irwin
As barriers to entry into AI fall and analytics become self-service, data designers are considering the steps to make AI adoption successful with centralized, managed data. Here we speak with Cody Irwin, Domo’s AI adoption lead, to ask him for his insights on adoption strategies for enterprise teams aimed at building a data foundation that takes the institution from AI experimentation to real-world implementation.
Understanding the Shift from Siloed Data to Centralized Truth
Maria Grush: We’ve been hearing about the dangers of siloed data for years. How does this change now with AI?
Cody Irwin: Data access and governance represent some of the biggest hurdles to achieving the promised efficiency gains of generative AI. Executives have been told for years that building a data warehouse or data lake is critical to analytics visibility and decision-making. This need has now become a requirement. The barrier to entry is no longer, “Do you know SQL, data science, and visualization techniques?” It’s simple: “Do you know words?” To strengthen the enterprise, data leaders must create centralized sources of controlled truth.
Tackling the Trust Deficit in Higher Education
Grush: Is AI well understood in the context of data management in higher education? Is there a “trust deficit” to overcome?
Irwin: The trust deficit exists everywhere, but is particularly acute in higher education. Educational institutions rely on data to manage admissions, financial aid, research, publications, accreditations, fundraising, compliance, and operations. Misrepresentation of data is often public and can have serious implications for an institution’s credibility. As data access and analysis becomes more self-service, data controllers have the responsibility to create and manage centralized, certified data.
Qualities of Effective Data Models
Grush: I know these are complex topics and our time is short here, but what qualities of data models should design leaders work toward?
Irwin: We saw value in data modeling that has AI – and flexibility – in mind. In particular, institutions should consider adopting a “medallion architecture” where “gold datasets” are made available for use by decision makers. Additionally, AI thrives on context, which requires more than just providing the data – the data models should expose semantics that provide the organizational context that AI can leverage to provide more meaningful and accurate answers.
Working Effectively with Data Productivity Platforms
Grush: Can you share an example of how designers can work effectively with data productivity platforms?
Irwin: The first step for most data designers is to make the data centrally available through a governed interface. They should provide the ability to retrieve or integrate data from almost any source environment. This centralization should enable the implementation of policies, security, logging and certification. Centralization should empower decision-makers, not restrict them. When it’s not easy, people tend to find a way around it. My company, Domo, provides controlled interfaces on this centralized data structure for self-service analytics and simple AI interactions.
Leadership in a Data-Driven AI Culture
Grush: How can people be strong design leaders in a changing AI culture with enormous data needs?
Irwin: The data basis is crucial. The quicker designers can build a foundation, the quicker their internal customers will feel empowered. We recommend not making perfection the enemy of progress. Design leaders should prioritize what they believe will be most impactful and act quickly to get that out there.
[Editor’s note: Image by AI. Microsoft Image Creator by Designer.]
About the Author
Mary Grush is an editor and conference program director at Campus Technology.
For further insights, you can access the full interview Here.
“`
