Data Governance and AI Governance: Where Do They Intersect?

Artificial intelligence is reshaping businesses and industries worldwide, driving new data products and services. According to a recent Stanford University AI Index Report, about 55% of companies have already integrated AI into at least one business unit or function. To harness the full potential of AI, organizations must align AI and data activities with their business strategy. This is where data governance (DG) and AI governance (AIG) come into play.

Understanding Data Governance

Data governance involves a set of processes, roles, policies, standards, and metrics that ensure the effective and efficient use of information across an organization. By managing data planning and operations, organizations can analyze data to gain knowledge and make decisions. Good data governance builds trust and encourages the use of data.

The core objectives of data governance are to fulfill organizational needs for data-driven insights, enable efficient technology use, and drive digital transformation. Data governance leads are responsible for:

- Allocating resources to support the data strategy and meet business needs.

- Defining data roles and responsibilities.

- Understanding data lineage and ensuring data quality.

- Balancing data access and security.

- Measuring DG performance and driving continuous improvement.

These decisions directly impact AI training and content generation. AI can define, produce, and use data governed by these principles. However, AI is just one aspect of data governance.

Data Governance Beyond AI

Data governance supports the data needs of all corporate technologies and employees handling data. It encompasses compliance with data regulations like the EU’s GDPR and involves various data systems, such as:

- Transaction-based schemas (e.g., relational databases and data warehouses).

- Big data tools (e.g., data lakes and graph databases).

- Sensor data from devices and the Internet of Things (IoT).

- Cloud computing storage and computation.

Data governance primarily focuses on data and its implications, often omitting technical details irrelevant to businesspeople. For example, while DG might address the need for data encryption, it usually doesn’t delve into specifics like encryption algorithms.

Given this broad scope, data governance must support the diverse data needs across the organization, with AI initiatives being just one part. As AI capabilities expand, dedicated AI governance frameworks become essential to manage their unique challenges.

Understanding AI Governance

AI governance (AIG) oversees the processes, roles, and technologies behind AI's cognitive capabilities. It involves system architecture, observation, and risk mitigation to ensure AI systems are profitable, safe, and ethical.

AI governance frameworks vary but share common goals of promoting understanding, accountability, and transparency in AI development and use. Key considerations for AI governance leaders include:

- Aligning AI planning and activities with business strategy.

- Identifying and managing AI technologies and capabilities.

- Assessing and mitigating AI-related risks.

- Defining accountability for AI research, development, and usage.

- Ensuring AI model quality, including strengths, weaknesses, and biases.

- Measuring AIG performance and driving improvement.

AI governance intersects with data governance, ensuring that AI training data and outputs align with business needs. However, AI governance also covers system architecture and performance, expanding beyond data governance's scope.

AI Governance Beyond Data

AI governance must establish best practices for system architecture and data practices. It involves monitoring and measuring AI performance, especially for systems with higher intelligence levels. Different AI application types include:

- Narrow intelligence: AI trained for specific tasks, such as customer chatbots.

- Strong intelligence: AI with general intelligence similar to humans, like self-driving cars.

- Super intelligence: AI surpassing human cognitive functions, currently existing in science fiction.

Higher intelligence levels in AI offer greater business value but also pose increased risks, adding a layer of oversight unique to AI governance. AIG addresses issues like biases, privacy, intellectual property, and technology misuse, ensuring fair and appropriate use of AI.

Similarities and Differences Between AIG and DG

AIG and DG share responsibilities in guiding data as a product created and consumed by AI systems. Both evaluate data integration, quality, security, privacy, and accessibility to meet business needs. For instance, both frameworks would address issues if an AI-powered recommendation engine suggests irrelevant items to customers.

However, their approaches differ. Data governance might audit the data pipeline for inconsistencies, while AI governance could enhance the recommendation algorithm. In some cases, a combined approach from both DG and AIG would be ideal.

Conclusion

Karen Meppen, Director of Client Services at Hakkoda, suggests understanding the business objective to determine the appropriate governance approach. Questions surrounding a business objective can clarify whether data governance or AI governance—or a combination—is needed.

While DG and AIG frameworks overlap, they are distinct with different expectations and outcomes. The key is to understand the data problem or opportunity before deciding the best governance approach.

“There are many actions you can take with your dataset that can have cascading negative or positive outcomes,” says Meppen. Planning for the unknown and learning from emerging data issues is crucial for effective governance.

By aligning DG and AIG, organizations can maximize their AI initiatives' value, ensuring they are profitable, safe, and ethical. Understanding and addressing data and AI challenges through appropriate governance frameworks is essential for achieving business success.