Why AI Implementation Cannot Be Delegated Entirely to IT

Artificial intelligence implementation involves more than selecting technology, developing applications, and connecting systems. It also requires organizations to make decisions about business priorities, operational processes, resource allocation, and how employees will work with AI.

IT departments play an essential role in these activities. They provide the technical expertise needed to evaluate infrastructure, manage integrations, establish security controls, and ensure that AI applications function reliably within existing systems.

However, technical expertise alone cannot determine which business problems should be addressed, how operational processes should change, or whether an AI investment is delivering sufficient commercial value.

An organization may successfully deploy an AI application without achieving its intended business objectives if the teams responsible for the affected operations have not defined the necessary requirements, performance expectations, and approval procedures.

The challenge extends beyond technical execution. Organizations must ensure that decisions about AI implementation reflect both engineering requirements and business priorities.

This requires shared accountability, with business leaders defining expected outcomes, operational teams establishing process requirements, and technical specialists developing the capabilities needed to deliver them.

Why AI Implementation Requires Business and IT Collaboration

IT departments traditionally manage many of the systems supporting business operations. Their responsibilities include software deployment, infrastructure management, data security, system maintenance, and technical support.

These capabilities remain essential when organizations introduce AI applications. However, AI implementation also influences how employees perform their responsibilities, how business decisions are made, and how operational processes function.

Integrating AI into customer service, for instance, may require connections to customer records, internal knowledge systems, and communication channels.

IT can establish these connections and ensure that the application meets technical and security requirements. Customer service leaders, however, must determine which inquiries AI should handle, when employees should intervene, and what level of service quality is acceptable.

Finance teams may also need to assess whether the expected operational improvements justify the implementation and ongoing operating costs.

These decisions require input from the departments responsible for business performance because technical requirements do not independently determine commercial priorities.

A McKinsey analysis of CEO leadership in AI emphasizes the responsibilities of chief executives in establishing organizational priorities, influencing how work is performed, and shaping the culture surrounding AI adoption.

This does not mean senior leaders must manage engineering decisions themselves. Their responsibility is to ensure that implementation supports business objectives and that the departments affected by AI participate in defining how it will be used.

What Happens When AI Implementation Is Treated Solely as an IT Project?

Assigning full responsibility for AI implementation to IT can create gaps between technical delivery and operational expectations.

These gaps may not become apparent during development, particularly when applications are evaluated primarily against technical specifications. They often become more visible when employees begin using the technology in daily operations.

Business Requirements May Be Defined Too Narrowly

Technical teams develop applications based on established requirements. However, those requirements may not fully reflect how employees handle exceptions, interpret information, or make operational decisions.

A finance department requesting AI-assisted reporting may specify the information an application should retrieve and the reports it should generate.

The resulting solution may function correctly but require significant modifications if reconciliation procedures, approval requirements, or reporting dependencies were not considered during planning.

These omissions can lead to additional development work, delayed deployment, and higher implementation costs.

Direct involvement from operational stakeholders helps ensure that technical specifications account for the processes and responsibilities the application is expected to support.

Technical Delivery May Become the Main Measure of Success

When AI implementation is managed primarily as a technology project, success may be evaluated through deployment completion, system availability, integration performance, or technical accuracy.

Although these indicators are important, they do not independently establish whether the investment has improved business performance.

An AI application may operate reliably while producing little improvement in productivity or service delivery.

The organization may continue to incur implementation and maintenance expenses without achieving the expected operational or financial benefits.

Business stakeholders must therefore establish success criteria that connect technical performance with measurable outcomes, such as processing efficiency, operating costs, service quality, and employee productivity.

Employee Adoption May Receive Insufficient Attention

Employees are responsible for incorporating new applications into their daily activities, but their operational requirements may not be fully addressed during technical development.

An AI application may require unnecessary manual inputs, provide information in an unfamiliar format, or introduce additional verification work.

These challenges can reduce employee adoption even when the application functions correctly.

IT can address technical limitations, but operational leaders must determine how existing responsibilities should change and ensure that employees understand the revised processes.

Successful implementation therefore requires attention to both application performance and the practical conditions that influence how employees use the technology.

Investment Decisions May Lack Commercial Accountability

AI projects require expenditure on software development, infrastructure, integrations, testing, employee training, and ongoing maintenance.

IT teams can assess technical costs and resource requirements, but determining whether an investment is commercially justified involves broader business considerations.

An application may deliver its intended functionality while exceeding the expenditure supported by its expected benefits.

Finance and business leaders must therefore participate in evaluating implementation costs, expected financial returns, and changes that could affect the original investment assumptions.

This helps organizations make decisions based on business value rather than technical completion alone.

How Responsibilities Should Be Shared During AI Implementation

Effective AI implementation requires clear responsibilities across the departments involved in business planning, technical development, operational execution, and risk management.

Shared accountability does not mean every department must approve every decision. Each function should have authority over matters within its responsibilities, while important decisions involving multiple departments should follow established review and approval procedures.

1. Executive Leadership: Establish Business Priorities and Accountability

Senior leadership is responsible for determining how AI investments support the organization's objectives.

This includes identifying business priorities, approving significant resource commitments, and resolving competing demands when projects affect multiple functions.

Executives must also determine which initiatives warrant investment based on their expected outcomes, operational importance, and implementation requirements.

Without clear direction, departments may pursue separate AI projects that compete for resources or deliver improvements with limited relevance to broader business priorities.

A well-structured approach to prioritizing AI initiatives helps organizations compare opportunities and allocate resources toward projects with clear business objectives.

Leadership should also establish who is accountable for the results expected from each implementation.

2. Business and Operational Teams: Define How AI Should Work

Operational departments understand the business processes that AI applications are intended to support.

They can explain how work is performed, identify recurring inefficiencies, document exceptions, and establish the conditions required for successful use.

Their involvement is particularly important when AI implementation affects existing responsibilities or introduces changes to established workflows.

For instance, introducing AI-assisted shipment monitoring requires logistics teams to define which delivery exceptions need attention, what information employees require, and how unusual situations should be managed.

These operational requirements influence application functionality and determine whether the solution is practical for daily use.

Operational teams should also participate in testing and acceptance decisions to ensure that applications meet actual business requirements before deployment.

3. IT and Engineering Teams: Deliver Reliable Technical Capabilities

IT and engineering teams remain responsible for the technical foundations of AI implementation.

Their responsibilities include evaluating system architecture, developing integrations, managing permissions, addressing security requirements, and testing application performance.

They must also identify technical constraints that could affect project feasibility, implementation costs, or long-term reliability.

Integrating AI with older business systems, for example, may introduce challenges involving data accessibility, system compatibility, and existing infrastructure limitations.

Technical specialists must evaluate these dependencies and communicate their implications to the business stakeholders responsible for investment and operational decisions.

This ensures that commercial expectations are informed by technical feasibility.

Engineering expertise remains central to successful implementation, even when accountability for business outcomes rests with operational and executive leadership.

4. Finance Teams: Validate Costs and Expected Financial Returns

Finance teams contribute to AI implementation by evaluating investment requirements, operating expenses, and expected financial benefits.

Their involvement helps organizations distinguish between operational improvements and actual financial returns.

Reducing the time required to complete a process, for instance, does not automatically produce direct cost savings.

Financial benefits may arise when recovered employee capacity supports additional output, reduces expenditure, or helps the organization avoid future staffing costs.

Finance teams should therefore establish how expected benefits will be calculated and validate whether the application delivers sufficient value relative to its total costs.

This supports greater financial accountability throughout implementation.

5. Risk, Compliance, and Legal Teams: Establish Appropriate Controls

AI applications may introduce risks involving data privacy, regulatory obligations, contractual requirements, and the reliability of business decisions.

Risk, compliance, and legal specialists should participate when projects involve sensitive information, regulated activities, or significant operational exposure.

Their responsibilities may include defining acceptable risks, establishing approval requirements, and ensuring that relevant obligations are addressed during development.

The NIST AI Risk Management Framework identifies documented responsibilities, cross-functional coordination, and executive accountability as important elements of AI governance.

Addressing these requirements during implementation helps organizations avoid introducing applications that function technically but create unacceptable business risks.

How to Establish Clear Decision-Making Authority for AI Projects

Cross-functional collaboration is most effective when organizations establish who has authority to make important decisions.

Involving multiple departments without defining their responsibilities can create delays, conflicting requirements, and uncertainty about accountability.

A practical approach is to assign an accountable business sponsor to each AI initiative while identifying the technical and operational stakeholders responsible for execution.

The following framework provides a basis for distributing decision-making responsibilities.

The precise allocation of responsibilities will depend on the organization's structure, regulatory requirements, and the risks associated with the application.

For example, a business sponsor may approve the operational scope of an AI initiative, while the technical lead determines whether the proposed architecture meets reliability and security requirements.

Neither responsibility should automatically override the other. When essential business or technical requirements cannot be satisfied, the relevant stakeholders must resolve the issue before implementation proceeds.

Organizations should also establish when decisions require escalation, who approves material changes to project scope, and which conditions must be satisfied before an application enters production.

This creates a clear framework for accountability without requiring senior leaders to become involved in every engineering decision.

Conclusion

AI implementation requires technical expertise, but its success depends on decisions extending beyond the responsibilities of IT departments.

Business leadership must define investment priorities and expected outcomes. Operational teams must establish how AI should function within existing processes, while IT and engineering specialists develop the technical capabilities required for reliable execution. Finance, risk, and compliance teams provide additional oversight where necessary.

The objective is not to reduce IT's involvement, but to ensure that technical execution is supported by clear business ownership and appropriate decision-making authority.

Forward-deployed AI engineering can strengthen this collaboration by connecting engineering expertise with the teams responsible for operational requirements and implementation outcomes.

For organizations investing in AI, shared accountability helps ensure that applications meet technical requirements while contributing to the business objectives that justify their development.