Do companies need a head of AI?

Monday morning, a new working week begins.

The sales team is planning to trial an AI-powered telephone assistant that can independently call companies, qualify leads and arrange follow-up appointments. The development department is meeting at 10:00 am to discuss the pros and cons of various AI tools for code reviews. And the marketing team wants to draw up a shortlist today of AI image generators to be used in future campaigns.

Perhaps things like this, or similar, are happening in your company right now.

Welcome to the club: according to a Bitkom study, 41% of companies in Germany with more than 20 employees are currently actively using AI. A further 48% are planning or discussing its use. Only around 11% have not yet engaged with artificial intelligence. [1]

The individual initiatives often seem sensible. Every department is looking for ways to work more productively, reduce routine tasks or cut costs.

However, the increasing use of AI raises questions to which many companies do not yet have clear answers:

• Which AI tools are already being used productively?
• Which teams are working with what?
• Which applications actually deliver measurable results?
• Where are duplicate costs or parallel structures arising?
• What standards apply to data protection, quality and governance?
• What skills do employees actually need?

And above all: Who is actually coordinating all of this?

It is precisely at this point that a new discussion is beginning in many companies. It concerns responsibility, priorities and the question of who maintains an overview of AI initiatives within the company.

Some companies currently assign this task to individual teams or managers. Others are already beginning to organise responsibilities in a more centralised manner. And with this, a new question is increasingly emerging:

Do we need a head of AI in our company?

From tool to strategy: Why AI needs leadership today

At first, the use of AI often seems like the introduction of yet another tool. A tool for marketing. An assistant for developers. Automation in sales.

Over time, however, the scope of the issue changes. Individual teams are rolling out new AI applications at a rapid pace. At the same time, key questions remain unanswered that can no longer be addressed within individual departments:

• Which processes should be fundamentally rethought and which tasks do we want to automate?
• Where is human oversight still necessary and what skills will our employees need in future?
• Which decisions may be prepared or influenced by AI, and who takes responsibility if something goes wrong?
• Who actually uses which tool, and how can knowledge about this be shared within the organisation?

Such questions cannot be resolved on the side during day-to-day business. They affect several departments simultaneously, touch on corporate strategy and require decisions that go far beyond the selection of a single tool.

What’s more, at some point priorities must be set. Investments compete with one another. Different departments pursue their own requirements. At the same time, the market is constantly changing. New models, new providers and new regulatory requirements mean that existing decisions must be reviewed regularly.

This is precisely what creates a growing need for coordination:

• Who assesses risks?
• Who prioritises initiatives?
• Who decides on training measures?
• Who defines standards?
• And who maintains an overview of the role AI is set to play in the company in future?

As long as AI is used only sporadically, these issues can often be resolved informally. However, as integration into operational processes increases, so does the effort required for coordination.

And this is precisely where the discussion begins about new responsibilities surrounding AI and where these should be anchored within the organisation.

Where is responsibility for AI anchored within the organisation?

As AI becomes more deeply integrated into operational processes, an organisational question arises: who is actually in charge of coordination?

There is as yet no established answer to this question. Depending on the size, structure and strategic importance of AI within the organisation, different models are emerging, ranging from informal solutions to the first signs of centralised responsibility.

1. Distributed responsibility

Many companies currently still manage AI in a decentralised manner. Specialist departments, IT, HR or individual managers drive initiatives forward independently of one another.

This model often works for a surprisingly long time. At the same time, however, it frequently leads to parallel decision-making, differing standards and a lack of transparency.

2. Coordination by senior management

Particularly in smaller companies, responsibility often lies directly with senior management.

This enables quick decisions and clear priorities. However, as AI becomes more widespread, this model quickly reaches its capacity limits. Alongside day-to-day operations, it becomes difficult to keep a constant eye on initiatives, risks and training needs.

3. Anchoring within IT

An obvious approach is to assign responsibility to IT. Infrastructure, data protection, security and system integration are already within its existing remit.

However, things become more difficult as soon as AI influences processes, roles or decision-making. Such issues lie outside the traditional remit of IT.

4. Integration within the context of digitalisation or transformation

Integration into existing digitalisation or transformation initiatives is also conceivable.

This model is particularly suitable when AI is understood as part of a broader change strategy. It becomes more difficult when AI-specific issues such as model selection, data protection or tool governance lose attention alongside other change initiatives.

5. Consolidation under a central AI responsibility

As the importance of AI grows, dedicated roles, staff units or cross-functional positions are increasingly emerging.

The focus here is less on the technology itself and more on coordination: defining standards, prioritising initiatives, structuring training and providing guidance.

It is currently difficult to predict which model will prevail in the long term. However, one thing is already clear today: the more AI transforms companies, the more difficult it becomes to organise responsibility informally in the long run. It is precisely this development that is making a role which consolidates this responsibility increasingly important: the head of AI.

What would a head of AI actually do?

The discussion about a potential head of AI quickly raises another question: what responsibilities would such a role actually entail?

The answer depends heavily on how AI is used within the company in question. In some organisations, the focus is primarily on utilising existing AI tools. Others develop their own chatbots or AI agents, or integrate AI directly into their products and software solutions. The responsibilities can vary accordingly.

However, one central task remains the same: AI creates a need for coordination between specialist departments, technology, governance and corporate strategy. In many companies, a head of AI would therefore have less responsibility for technology per se and more of a coordinating role between different departments.

Prioritising AI initiatives

As soon as several specialist departments are working with AI at the same time, investments, resources and attention compete with one another.

Not every idea warrants a company-wide rollout. Not every automation actually improves a process. And not every new feature from a provider should be adopted immediately.

This is where a head of AI would be needed: Which initiatives create strategic value? Which remain local optimisations? And where do duplicate structures arise between individual departments?

Especially in larger companies, this quickly becomes a classic management task. Different teams pursue their own goals, budgets and requirements. Without centralised prioritisation, the risk of parallel developments and competing decisions grows.

Managing skills development within the company

AI does not affect all roles within the company equally. A developer requires different skills to a recruiter, a procurement officer or a marketing manager. Some employees must be able to actively operate AI systems. Others must critically evaluate results, assess risks or monitor decisions.

This is precisely where a key task of the role lies: to specifically manage skills development within the company. Which skills are actually relevant in which roles? Where are new requirements emerging? And which training measures really contribute to the company’s future way of working?

Linked to this is also an economic question: training ties up time and budget. Companies must therefore decide which skills should be developed strategically and where basic knowledge is sufficient.

Defining governance, standards and human oversight

As AI becomes increasingly integrated, new demands are placed on governance and decision-making rules.

For example, should AI be allowed to pre-screen applications or recommend decisions? Such decisions cannot be made as part of day-to-day operations. At the same time, specialist departments expect quick solutions, whilst companies require common standards for data protection, quality, documentation and governance.

A head of AI would mediate these tensions: Which tools may be used? Is every team permitted to procure new AI licences independently? Where is human approval still necessary? And which decisions should be made centrally without unnecessarily slowing down individual teams?

Particularly in regulated sectors or when making sensitive decisions, this gives rise to rules that cannot develop informally on the side, but must be consciously defined.

Strategically contextualising market developments

The AI market is changing at a rapid pace. New models, providers and regulatory requirements are constantly altering the operating environment.

A head of AI would not only assess such developments from a technical perspective, but also contextualise them strategically: Which changes are actually relevant to the company? Where is there a concrete need for action? And which trends generate short-term attention without having a long-term impact?

This also includes the question of when companies should consciously respond to new technologies and when stability is more important than constant reorientation.

Acting as a bridge between specialist departments, IT and senior management

AI addresses technical, organisational and strategic issues simultaneously. It is precisely this that often leads to misunderstandings within companies between specialist departments, IT and senior management.

A head of AI would act as a liaison here. The role would need to make technical possibilities understandable, contextualise operational requirements and translate strategic decisions into concrete measures. It is precisely this coordinating function that distinguishes the role from traditional technology positions.

All in all, this results in a job profile that is spread across several people in many companies. And this is precisely where the real organisational question arises: is an informal distribution of responsibility sufficient for this, or will it lead to an independent leadership role in the long term?

At what point does AI governance become a role in its own right?

Not every company needs a head of AI. And it is unlikely that a single organisational model will become the norm. Nevertheless, many companies are increasingly asking themselves at what point informal coordination is no longer sufficient.

This is not just a question of company size. The decisive factor is often the extent to which AI is integrated into processes, decisions and products, and the organisational consequences that result.

In some companies, the coordination burden is growing in particular. Several departments are introducing new applications in parallel. Different teams are testing similar tools. Investments are overlapping. At the same time, there is a growing need for coordination between IT, specialist departments, HR, compliance and senior management.

The more AI initiatives are launched simultaneously, the more difficult it becomes to maintain an overview informally in the long term.

In other companies, the pressure stems more from governance and risk issues. Anyone working with sensitive customer data, mapping regulated processes or integrating AI directly into products must define much more precisely which rules should apply. Here, it is no longer just about efficiency gains, but also about liability, traceability and control.

A further factor overlays both patterns: the speed of technological change.

New models, providers and functions are constantly changing the possibilities in the market. Companies must therefore decide when to experiment deliberately, when to establish standards, and where stability becomes more important than constant adaptation.

When increasing coordination efforts, rising governance requirements and a rapidly changing market converge, many companies are beginning to discuss a greater centralisation of responsibility.

Many companies are only now beginning to determine for themselves how this responsibility should be embedded within their organisational structure.

In the end, there may not be a head of AI at all

The more specifically companies consider issues of responsibility surrounding AI, the more obvious the idea of a dedicated role seems at first. At the same time, it remains entirely unclear whether a standalone head of AI will actually become a permanent fixture.

Many tasks relating to AI already overlap with existing roles: IT departments are responsible for infrastructure, security and system integration. Digitalisation and transformation departments deal with process changes and automation. HR organises training and skills development. Compliance and data protection define the regulatory framework. Senior management sets strategic priorities.

This situation is not new. Companies have already faced similar questions with previous cross-functional issues.

Digitalisation, too, led many companies to temporarily introduce dedicated leadership roles such as the Chief Digital Officer. In some organisations, this evolved into a permanent role. In others, the role disappeared again after a few years because digital issues increasingly became part of existing responsibilities. Similar developments can also be observed in other cross-functional areas. New tasks often initially arise outside established structures. As they mature, they gradually migrate back into existing departments and management roles.

How this will specifically develop into a role such as head of AI therefore remains to be seen. Possible outcomes include permanent roles, hybrid forms, or the reintegration of responsibilities into existing departments. The discussion cannot therefore be narrowed down to the title alone. It is equally possible that the role will emerge temporarily, change again later, or partially dissolve once the use of AI has become more firmly established as organisational practice.

Conclusion: Who takes responsibility?

It is difficult to give a definitive answer today as to whether companies will actually need a head of AI in the future.

In some organisations, responsibility for AI will likely remain within existing roles on a permanent basis. Others will create new roles, AI councils or centralised responsibilities. And some roles may only be temporary, before AI becomes more firmly established as the organisational norm.

This is precisely why the discussion about the job title alone falls short. The real change lies elsewhere: in many companies, AI is evolving for the first time from a single tool into an issue that demands permanent organisational responsibility.

Who prioritises initiatives? Who defines standards? Who assesses risks? Who decides on training, governance and human oversight?

Many companies are only just beginning to find organisational answers to these questions. How this responsibility is ultimately embedded within the organisation will be handled differently by different companies. However, it is already clear today that this responsibility is a reality.

Notes (partly in German):

[1] Bitkom: Digitalisierung der Wirtschaft: Fast jedes Unternehmen beschäftigt sich mit KI

Here you will find an assessment of the question: How good are AI-supported code reviews really?

Here you will find an article on generating unit tests with AI.

Would you like to discuss the merits and drawbacks of a head of AI as an influencer or opinion leader? Then share this post on your networks.

Michael Schenkel has published more articles on the t2informatik Blog, including: