Questions Around Who Should Manage Data and AI Intensify as Adoption Accelerates

The big question of who should actually manage data and AI is getting louder as more and more businesses start using these powerful tools. Simply put, it’s a complicated picture with no single, easy answer. Different organisations, depending on their size, industry, and existing structures, will grapple with this in unique ways, but the core challenge remains: getting the right people with the right skills and authority in place to handle these assets effectively and responsibly.

Why the Fuss Now? AI’s Rapid Rise

AI isn’t some niche tech project anymore; it’s becoming central to how many businesses operate. From automating customer service to optimising supply chains and even driving strategic decisions, AI’s reach is expanding at a remarkable pace. This rapid adoption, while exciting, is also bringing to light some crucial gaps in how we typically manage our most valuable assets. Suddenly, data – the fuel for AI – isn’t just for reporting; it’s the very foundation of intelligent systems, and how it’s handled directly impacts AI’s success and ethics.

The Overlap Problem: Data and AI Are Inseparable

You can’t really talk about managing AI without talking about managing data. AI models are only as good as the data they’re trained on. This means ensuring data quality, accessibility, privacy, and security become even more critical when AI is involved. The traditional lines between data management, IT, and business operations are blurring significantly. This interconnectedness is why the “who” question is so pressing – it’s no longer just about databases; it’s about the intelligence derived from them and the processes built upon that intelligence.

There are several departments and roles that typically put their hand up (or have it put up for them) when it comes to overseeing data and AI. Each has its strengths and weaknesses, and often, the right solution involves elements from several.

The IT Department: The Traditional Gatekeepers

IT has always been responsible for the infrastructure and security of our digital world. It makes sense that they would have a significant role in managing data and AI.

Pros of IT Leadership

Infrastructure Expertise: IT pros are the masters of servers, networks, cloud platforms, and data storage systems. They know how to keep things running smoothly and securely from a technical perspective.
Security Focus: Data and AI security are paramount. IT departments are well-versed in cybersecurity protocols, access controls, and incident response, which are vital for protecting sensitive information and AI models from threats.
Scalability Management: As data volumes grow and AI deployments expand, IT has the experience to manage the underlying technological scalability required.

Cons of IT Leadership

Business Context Gap: Often, IT departments might not fully grasp the strategic business context or the specific ethical implications of AI use cases. Their focus tends to be on technical functionality rather than direct business value or societal impact.
Agility Challenges: Traditional IT structures can sometimes be slower to adapt to rapid changes, which contradicts the often iterative and experimental nature of AI development.
Data Content vs. Container: While IT secures the “container” (the database, the server), they might not always be experts in the “content” (the quality, meaning, and business relevance of the data itself).

The Data Team: The Specialists

This could be a dedicated data science team, a data analytics department, or a chief data officer (CDO) function. They’re all about understanding and leveraging data.

Pros of Data Team Leadership

Data Expertise: Naturally, data teams are experts in data quality, data modelling, data governance, and analytics. This knowledge is fundamental for preparing and maintaining data suitable for AI.
AI Understanding: Data scientists and analysts are often the ones building and deploying AI models. They understand the nuances of model performance, biases, and interpretability.
Business Alignment (Often): Good data teams work closely with business units to understand their needs, which helps ensure AI solutions are relevant and valuable.

Cons of Data Team Leadership

Infrastructure Limitations: While they understand data, they might lack the deep infrastructure knowledge of IT, potentially leading to bottlenecks or suboptimal scaling solutions.
Broader Organisational Scope: A data team, while crucial, might not have the organisational authority or breadth of involvement to enforce data and AI policies across all departments without broader executive support.
Focus on Insights, Not Operations: Their primary goal is often to extract insights or build models, rather than necessarily owning the ongoing operational management and governance aspects of AI systems once deployed.

The Business Units: The End-Users and Value Generators

When AI is designed to serve a specific business function (e.g., marketing, finance, HR), the relevant business unit often wants a strong say in its management.

Pros of Business Unit Leadership

Direct Value Alignment: Business units are acutely aware of their operational needs and how AI can deliver value. They can ensure AI initiatives are directly tied to business goals.
Domain Specific Knowledge: They possess invaluable domain expertise, which is critical for identifying relevant use cases, understanding data nuances specific to their area, and interpreting AI outputs accurately.
Quicker Adoption: When a business unit champions an AI solution, adoption within that unit can be much faster and more seamless.

Cons of Business Unit Leadership

Siloed Approaches: If each business unit manages its own data and AI independently, it can lead to fragmented efforts, inconsistent standards, and redundant work across the organisation.
Technical Knowledge Gaps: Business unit leaders may not have the technical knowledge to assess the underlying complexities of AI, such as model ethics, scalability, or data security risks comprehensively.
Governance and Compliance Risks: Without a central guiding hand, individual business units might inadvertently create compliance risks (e.g., GDPR, ethical AI guidelines) by not adhering to broader organisational policies.

The Role of Centralised Governance and Strategy

Given the complexities, many organisations are realising that a fragmented approach simply won’t cut it. This is where centralised governance comes into play.

The Chief Data Officer (CDO) / Chief AI Officer (CAIO) Evolution

The rise of the CDO in recent years was a direct response to the growing importance of data. Now, with AI, some organisations are considering a Chief AI Officer (CAIO) or expanding the CDO role.

The CDO’s Expanding Remit

Data Strategy: A CDO is responsible for the overall data strategy, ensuring data is collected, stored, managed, and used effectively across the enterprise. This naturally extends to ensuring data is AI-ready.
Data Governance: Establishing policies and procedures for data quality, privacy, security, and accessibility falls under their purview. These governance frameworks are essential for ethical and reliable AI.
Cross-Functional Collaboration: A good CDO acts as a bridge between IT, business units, and data teams, fostering a data-driven culture and ensuring consistency.

The Potential for a CAIO

AI Strategy: A CAIO would focus specifically on the organisation’s AI strategy, identifying new opportunities, overseeing AI development, and ensuring alignment with business objectives.
Ethical AI and Fair Use: This role could be critical in championing ethical AI principles, fairness, transparency, and accountability across all AI deployments.
AI Governance: Developing and enforcing policies for AI model validation, monitoring, risk management, and bias detection would be key responsibilities.

Challenges of New C-Suite Roles

Overlapping Responsibilities: The lines between data and AI are so blurred that separating CDO and CAIO roles can lead to unnecessary overlap and turf wars.
Talent Scarcity: Finding individuals with both deep technical AI knowledge and strong business leadership skills is a significant challenge.
Integration with Existing Structures: Introducing new C-suite roles requires careful integration into existing organisational hierarchies to avoid friction.

Data and AI Governance Committees

Beyond individual roles, a cross-functional committee can be an effective way to bring diverse perspectives to the table.

Composition is Key

Representation: Such committees should include representatives from IT, data science, legal, compliance, relevant business units, and potentially ethics or risk management.
Balanced Perspectives: This ensures that technical feasibility, business value, legal/ethical implications, and operational practicalities are all considered in data and AI decisions.

Remit and Authority

Policy Setting: These committees can define overarching data and AI policies, standards, and best practices for the entire organisation.
Risk Assessment: They would be responsible for identifying, assessing, and mitigating risks associated with data use and AI deployment, including privacy, security, and ethical concerns.
Dispute Resolution: Providing a forum to resolve conflicts or make decisions when different departments have conflicting views on data or AI usage.

Navigating the Ethical and Regulatory Labyrinth

Beyond the pure operational aspects, the ethical and regulatory landscape around data and AI is rapidly evolving and demands dedicated attention.

Built-in Ethics, Not an Afterthought

Managing data and AI isn’t just about efficiency or profit; it’s increasingly about responsibility. Decisions made by AI systems can have profound impacts on individuals and society.

Ensuring Fairness and Reducing Bias

Data Sourcing and Curation: Proactive steps must be taken to identify and mitigate biases in training data, as these will inevitably be amplified by AI models. This requires critical evaluation of data sources and collection methods.
Model Validation and Monitoring: Continuous testing and monitoring of AI models are essential not only for performance but also for detecting unintended biases or discriminatory outcomes once models are deployed.
Explainability: When possible, understanding how an AI model arrived at a particular decision (explainable AI or XAI) can help identify biases and build trust.

Transparency and Accountability

Clear Policies: Organisations need clear internal policies on how AI is developed, deployed, and monitored, making it transparent to stakeholders.
Human Oversight: Even with advanced AI, human oversight remains crucial. There should always be a clear path for human intervention and accountability when AI decisions are questioned.
Stakeholder Communication: Openly communicating about the responsible use of AI with customers, employees, and the public helps build confidence and manage expectations.

Compliance with Evolving Regulations

New laws and guidelines are emerging globally that directly impact data and AI management.

Understanding Regional Differences

GDPR (EU): Principles like data minimisation, purpose limitation, and the right to explanation have significant implications for AI using personal data.
AI Act (EU): This upcoming regulation categorises AI systems by risk level and imposes stringent requirements, especially for high-risk AI, covering areas like data quality, human oversight, robustness, and cybersecurity.
Local Regulations: Beyond major international laws, organisations must also navigate specific national and industry-specific regulations that govern data privacy and AI use.

The Role of Legal and Compliance Teams

Early Involvement: Legal and compliance teams must be involved from the very start of any AI project, not just at the end. Their input is crucial for designing compliant systems.
Risk Assessment and Mitigation: They can identify potential legal and regulatory risks associated with specific AI applications and help develop strategies to mitigate them.
Policy Development: Collaborating with governance committees to create internal policies that align with external regulations and ethical principles.

Hybrid Models and Shared Responsibility

For many organisations, the most practical answer isn’t to put everything under one roof but to adopt a hybrid model where responsibilities are shared and clearly defined.

Federated Approach with Centralised Oversight

This model allows different business units or departments to manage their specific AI applications and data sets while operating within a common framework.

Decentralised Execution

Domain-Specific Ownership: Business units can maintain direct control over AI projects relevant to their operations, fostering agility and direct business value creation.
Empowered Teams: This empowers individual teams to develop and innovate with AI, leveraging their specific domain expertise.

Centralised Guidance

Strategic Direction: A central data/AI function or governance committee provides the overall strategy, sets standards, and ensures alignment with organisational goals.
Infrastructure and Security: Core IT remains responsible for the foundational infrastructure, security protocols, and potentially shared AI platforms.
Policy Enforcement: The central body ensures that all AI initiatives adhere to established governance, ethical guidelines, and regulatory compliance.

The Importance of Collaboration and Communication

No matter which model is chosen, effective collaboration and clear communication are non-negotiable.

Cross-Functional Teams

Design and Development: AI projects should involve cross-functional teams from the outset, including data scientists, engineers, business domain experts, legal/ethics representatives, and IT.
Shared Understanding: This fosters a shared understanding of the project’s goals, risks, and technical requirements, reducing misunderstandings later on.

Clear Roles and Responsibilities

Accountability: Each person and department involved needs a clear understanding of their specific roles, responsibilities, and decision-making authority regarding data and AI. This avoids ambiguity and ensures accountability.
Defined Processes: Establishing clear processes for data access, model development, deployment, monitoring, and incident response helps streamline operations and reduce friction.

In essence, deciding who manages data and AI isn’t just an organisational chart exercise; it’s a strategic imperative. The best solutions will likely be dynamic, involving a blend of specialised expertise, centralised governance, robust ethical frameworks, and a strong culture of collaboration. As AI continues to embed itself deeper into our businesses, finding the right balance will be key to unlocking its full potential responsibly and effectively.