A featured contribution from Leadership Perspectives: a curated forum reserved for leaders nominated by our subscribers and vetted by the Construction Tech Review Advisory Board.
External Comms Manager EMEA
Ethical Data Management in the Age of AI and ML
Muhammad possesses extensive experience in designing and delivering data solutions for large blue-chip clients. His skills and expertise encompass setting up data management capabilities to manage data effectively and deliver quality information to both human and system decision-makers. Presently, he serves as the Head of Data Architecture at BAE Systems Digital Intelligence, managing the data architecture capability and leading the development of a diverse, multidisciplinary team to assist and advise clients across industries to achieve strategic objectives.
Could you explain Black Data?
Black data refers to unused data stored within an organization's database system. This surplus data carries potential risks, particularly for cloud-based organizations, incurring storage costs without understanding its purpose or existence. Moreover, retaining such data poses compliance risks, especially if it pertains to individuals who are no longer part of the organization or have any relationship with the organisation as a customer.
How should an organization effectively manage data footprints in terms of data management?
A data footprint reflects an organization's data volume and value. Effective management involves establishing a clear data lifecycle, encompassing data collection, storage, management, utilization, archival, and eventual deletion.
Organizations must define this lifecycle and establish policies to retain data only for operational purposes. Data should be archived or deleted when it outlives its value, except when regulatory obligations mandate retention. This approach streamlines data management and reduces unnecessary accumulation.
How can an organization make data discoverable and accessible without compromising security and privacy standards?
Data discoverability and accessibility do not imply open access for all within the organization. Instead, it means cataloging data internally. Users seeking to utilize data should search, locate, and understand its value. Access can be granted if appropriate. For restricted access, a defined process should facilitate access requests, maintaining security and privacy while enabling legitimate use.
Data is important in the world of the latest technologies. Analytics, AI, and business processes all need data
"An organization should ensure its data use comply with both general and industry-specific obligations." How? Effective data governance requires a robust framework involving policy creation, execution by dedicated personnel, and adherence to internal, external, and regulatory standards. A data governance policy should outline compliance requirements, guided by relevant internal, external standards, and regulatory obligations. A functional data governance team should identify applicable policies and standards, ensuring relevance. This structured approach includes policy definition, compliance management, and strategic decision-making.
What measures should companies take to ensure ethical data use, especially with AI and ML implementation?
Ethical data use, crucially in AI and ML, requires a comprehensive framework encompassing principles, considerations, and governance. Preventing biases when creating outcome-guiding algorithms is essential. Establishing transparent, unbiased principles, disclosing data used in algorithms, and addressing biases ensures responsible data management. Project launches should follow an ethics checklist, addressing ethical questions. Any gaps lead to actions or adjustments. Projects implement changes, collect compliance evidence presented to a Governance Forum for review. This confirms alignment with ethical standards in data processing and usage.
By integrating principles, thoughtful considerations, and robust governance, organizations establish a foundation for ethical data management, especially important in AI and ML contexts.
