Governance gaps between traditional data and analytics (D&A) practices and AI’s requirements are emerging with organizations accelerating AI adoption. Without clear accountability and role delineation, organizations risk compliance failures, duplicates or gaps in governance, and a loss of trust in AI use cases.
D&A leaders must proactively negotiate, define, and map AI-ready data responsibilities while strengthening specialized model governance.
Clarify Roles Across D&A and AI Governance
Historically, governance of traditional analytics, like business intelligence, has fallen under the mandate of data governance, making “data” versus “D&A” governance interchangeable.
However, AI changes this by bringing in new technology, and additional stakeholders and risks.
Further, AI adoption is exposing gaps in traditional D&A governance.
D&A leaders can mitigate these gaps by connecting D&A governance to relevant aspects of AI governance and by taking the lead in clarifying the roles of key stakeholders including data management, data governance, AI teams, cybersecurity, assurance, and others.
In D&A governance, decision rights are often shared or distributed among business, IT, and data leaders.
However, AI introduces new technical and operational complexities that traditional D&A governance does not typically address such as risk management.
AI governance has a large scope, overseeing the rapid scaling of technology, specialized infrastructure, and ongoing model oversight, as well as an operational governance for many AI use cases, which extends the scope of governance beyond data itself.
D&A leaders must proactively map and assess their organization’s existing governance mechanisms and policies for AI, data, and analytics, based on regulatory compliance, efficacy, and completeness for achieving business objectives. This process should focus on clearly defining accountability and responsibility for the governance decisions, and monitoring and enforcing relevant policies.
Prioritize AI-Ready Data
D&A leaders in particular should prioritize how AI-ready data fits into their D&A governance mandate.
Despite D&A teams making considerable strides in data management over the past decade, many D&A teams are still evolving their governance frameworks and capabilities to address the requirements for AI-ready data.
Few organizations have already implemented the necessary practices, such as observability, lineage tracking, and data labeling, to ensure their data is AI-ready. These foundational activities are essential for supporting the transparency, trustworthiness, and compliance that AI systems require.
According to AI leaders, insufficient data quality or availability is a top failure point for AI.
This leaves D&A leaders with an opportunity to carve out explicit accountability for AI-ready data governance tied to the broader AI governance framework.
Avoid Conflating Data and Model Governance
Many organizations mistakenly treat AI, data, and analytics governance as a single discipline, but data governance and model governance have distinct purposes and requirements.
Data governance is focused on the policies, standards, roles, and processes that ensure data quality, consistency, security, and accessibility throughout the data lifecycle. Its primary aim is to ensure appropriate behavior in the validation, creation, consumption and control of data and analytics.
In contrast, model governance builds trust in AI models by ensuring transparency, fairness, and accountability. This includes bias mitigation, model validation, explainability, ethical oversight, and risk management.
Despite the importance of data and model governance, many organizations are not meeting reasonable standards for either.
Effective data governance requires robust frameworks and practices to ensure data is high-quality, consistent, secure, and accessible throughout its life cycle.
Meanwhile, effective model governance demands specialized tools and skills to address unique challenges such as model drift, fairness, explainability, and ongoing model validation. It also needs to ensure that the data represents the use case.
Without clear differentiation and dedicated attention to each, organizations risk undermining the quality of their data and the trustworthiness of their models.
It is essential to connect oversight of data and model governance through a contiguous framework that supports enterprise outcomes and enables continuous improvement as business needs, technology, and regulations evolve.
Organizations can avoid conflating data and model governance by adopting adaptive governance, clarifying decision rights, and maximizing operational efficiencies and compliance.
The Wrap
D&A leaders must take responsibility for AI-ready data within the broader AI governance framework.
Organizations must intentionally fill gaps in governance structures to achieve responsible AI and operational efficiency, in order to realize the full value of the organization’s AI use.
Trusted insights for technology leaders
Our readers are CIOs, CTOs, and senior IT executives who rely on The National CIO Review for smart, curated takes on the trends shaping the enterprise, from GenAI to cybersecurity and beyond.
Subscribe to our 4x a week newsletter to keep up with the insights that matter.
