Artificial intelligence is having a noticeable impact on the service desk. AI can categorize inquiries, provide knowledge, prepare responses, or perform individual tasks on its own. This creates new opportunities for service organizations to reduce the workload on employees and deliver services more quickly.
At the same time , a crucial question arises: Who is responsible when an AI makes an incorrect recommendation, misjudges a request, or takes action on its own?
This is exactly where HUCAISM®, short for Human-Centered AI Service Management, . The approach developed by SERVIEW creates a framework for AI-powered services in which human responsibility, control, and traceability are preserved. After all, even as AI acts with increasing autonomy, responsibility cannot be delegated to a system.
Why AI Is Fundamentally Changing the Service Desk
Automation has long been part of everyday life at the service desk. What’s new, therefore, isn’t just that technology is taking over tasks; it’s the way modern AI systems go about doing so.
Traditional software largely follows established rules. Generative AI, on the other hand, operates on a probability-based model. It can provide different answers to the same question and may even generate results that sound plausible but are factually incorrect.
This changes a fundamental assumption for the service desk. A service does not have to experience a technical failure to cause a problem. The system may be accessible and respond quickly, yet the information it provides may still be incorrect.
The HUCAISM book describes this very problem using the example of an internal service desk wizard. The system provides outdated instructions because an internal policy has changed, but the knowledge base it uses has not. Technically, the wizard continues to function. However, from a business perspective, it still leads employees in the wrong direction.
As a result, the classic question “Is the service up and running?” is no longer sufficient in an AI-powered service desk.
In addition, there are questions such as:
- Is the answer provided by customer service correct?
- Who is responsible for the result?
- Who can tell when the AI's behavior changes?
- When should a person intervene?
- Who decides what AI is allowed to do on its own?
AI in the service desk is therefore not just a technological issue. It becomes a matter of responsibility and service management.
Even a properly functioning AI service can still be wrong
This is precisely what makes AI-powered services unique. A typical error is often visible: a system is unavailable, a process stops, or an error message appears.
With AI, things may be different.
A wrong answer can be phrased in a linguistically convincing way. An inappropriate recommendation may seem plausible at first. A request may be prioritized incorrectly without the system itself reporting an error.
HUCAISM therefore does not focus solely on technical availability. It is also crucial whether the AI-powered service provides accurate information and is helpful to people.
This has a direct impact on the service desk. From HUCAISM’s perspective, an incorrect AI result can constitute an incident, even if nothing has technically failed. This broadens the focus from mere system functionality to actual service performance.
Responsibility cannot be delegated to AI
One of the six fundamental principles of HUCAISM is: Responsibility cannot be delegated.
The idea behind it is deliberately simple. AI can provide information, make recommendations, or perform tasks. However, a human must take responsibility for the outcome.
This distinction is particularly important in the service desk. If an AI misclassifies a ticket, prepares an inappropriate response, or triggers an action, the explanation afterward should not be, “The AI decided that.”
HUCAISM therefore requires that it be clear who is responsible for every AI-powered service. Furthermore, when making decisions with significant implications, it must be explicitly defined what role humans play and when they make decisions on their own.
This shifts the central question from technology to organization:
Not only: What is our AI allowed to do? But also: Who is responsible for it?
The AI Service Owner establishes clear responsibilities
The AI Service Owner therefore plays an important role within HUCAISM.
He is responsible for an AI-powered service from start to finish, and this explicitly includes the AI component of the service. This involves, for example, determining the purpose of the AI, the level of human oversight required, and who is accountable for the service in the event of an emergency.
The AI Service Owner thus answers one of the most important questions in AI-powered service management: Who is responsible for this service?
HUCAISM does not automatically mean that companies have to create numerous new positions. Responsibilities can be assigned to existing roles. What matters is that responsibility, competence, and the ability to take action are actually aligned.
In addition to the AI Service Owner, HUCAISM defines other areas of responsibility:
- The supervisory body monitors the AI-powered service from a technical standpoint and requires expertise, time, and genuine authority to intervene.
- The governance role is responsible, among other things, for policies, documentation, and overall oversight.
- Knowledge managers ensure that the knowledge accessed by generative AI is maintained and kept up to date.
- The operations role is responsible for the safe technical operation of the AI system.
These responsibilities are particularly intertwined in the service desk. After all, a good AI response doesn't depend solely on the system used. It also depends on whether the underlying knowledge is accurate, whether changes are detected, and whether people can step in when necessary.
To what extent should AI be allowed to make decisions on its own at the service desk?
Clear accountability does not mean that every action taken by an AI must be individually approved by a human.
HUCAISM distinguishes between three basic forms of human oversight. In " human-in-the-loop," a human approves a decision. In " human-on-the-loop," the AI acts autonomously, while a human monitors the process and can intervene. In " human-out-of-the-loop," the AI can act autonomously within clearly defined limits.
According to HUCAISM, which form is appropriate depends primarily on three factors:
- How serious would the consequences of a mistake be?
- Can a wrong decision be reversed?
- What are the regulatory requirements?
A simple example from the HUCAISM book is automated password reset. The consequences are relatively minor, and the action can be undone. A decision with major or difficult-to-reverse consequences, on the other hand, requires significantly more human involvement.
For the service desk, this means that not every case requires the same level of oversight. The key is to deliberately choose the appropriate level of automation.
AI Is Intended to Empower Service Desk Staff First
Another principle of HUCAISM is “augmentation before substitution.”
AI should initially be used to support people in their work and enhance their capabilities. Full automation is not an end in itself, but rather a conscious decision to use it in appropriate areas.
This approach opens up many opportunities, especially for the service desk. AI can make knowledge more readily available, prepare information, or take over routine tasks. This frees up employees to focus on situations that require experience, context, and human judgment.
At the same time, HUCAISM warns against removing human expertise from a service too early. This is because unusual or complex situations, in particular, may require skills that are rarely needed in a normal automated process.
The goal, therefore, is not to achieve as much automation as possible at any cost. What matters most is a service in which humans and AI work together effectively.
Knowledge Becomes a Matter of Responsibility at the AI Service Desk
An AI assistant at the service desk is only as helpful as the information it draws on.
At first glance, this may seem obvious, but it takes on special significance in the context of generative AI. Outdated, contradictory, or inappropriate sources of knowledge can cause a system to generate convincing but incorrect answers.
HUCAISM therefore explicitly establishes a responsibility for knowledge. The knowledge sources of generative services must be curated and maintained. Ensuring that information is up-to-date and accurate thus becomes part of the responsibility for the AI-powered service.
For companies, this means that knowledge management does not end with the introduction of an AI assistant. On the contrary, the quality of existing knowledge becomes even more important.
Especially in situations where employees or customers receive answers directly from an AI service, it must be clear that:
- Who maintains the underlying information?
- How are changes taken into account?
- How is incorrect or outdated content identified?
- Who takes action when AI draws on inappropriate knowledge?
This is how knowledge management becomes an important component of responsible AI in the service desk.
Human supervision must actually work
One person on an organizational chart isn't enough.
HUCAISM makes it clear that effective oversight requires three prerequisites: expertise, time, and an actual right to intervene. If any one of these is missing, human oversight quickly becomes merely theoretical.
For the service desk, for example, this means that employees must be able to recognize when an AI result is questionable. At the same time, they need a clear process for overriding an automated decision or escalating an issue.
This is precisely where employee training takes on greater importance. HUCAISM describes the ability to make judgments about AI results in the face of uncertainty as a core competency. People must learn to be able to reasonably distrust a result that seems plausible, to verify it, and to intervene at the right moment.
Human oversight therefore does not mean that humans monitor every action taken by the AI. It means that they remain able to act when their judgment is needed.
Clear accountability builds trust in AI services
For users, the specific AI model behind a service is often not the most important factor. Above all, they want to be able to trust that their request will be handled correctly.
However, trust isn't built solely on quick responses.
People need to be able to recognize when AI is involved. Those in charge need to be able to understand how a result was produced. And if something goes wrong, it must be clear who can take action.
HUCAISM therefore combines responsibility with transparency. Traceability should not be limited to the technical level but should also be present in the service itself. An AI-powered service must be designed in such a way that users, operators, and auditors can understand how AI results are handled.
This is an important step, especially in the service desk. That's because this is where many people experience AI firsthand in their day-to-day work.
Questions Companies Should Answer Now
Organizations that are already using AI in their service desk—or plan to do so—don’t need to immediately rebuild all their structures from scratch. The first step is to establish clarity.
A few simple questions can help with this:
- In what ways is AI already supporting or driving our service processes?
- Who is responsible for each of these AI-powered services?
- What decisions is AI allowed to make on its own?
- When should a person review, approve, or intervene?
- Who is responsible for the AI's sources of knowledge?
- Would we be able to tell if the service were working from a technical standpoint but producing incorrect results from a subject-matter perspective?
- Is there a clear way to hand things back to a human if the AI cannot reliably handle a case?
That is precisely where the strength of the HUCAISM approach lies. The discussion doesn't start with the next AI tool, but with the responsible operation of the service.
Conclusion: AI can act, but responsibility remains with humans
Artificial intelligence is playing an increasingly integral role in the actual delivery of services at the service desk. It answers questions, prepares decisions, and can act independently in appropriate areas. However, this also increases the demands placed on accountability, oversight, and traceability.
HUCAISM provides a clear framework for this. The framework puts people at the center and ensures that responsibility remains clear even as AI takes on an increasing share of the service.
For companies, this doesn’t mean less automation. It means more conscious automation. By determining who is accountable for an AI service, what level of human oversight is necessary, and how to intervene in the event of errors, companies lay the groundwork for a service desk where AI is used not only quickly but also responsibly.
Most Recent
Would you like to know how HUCAISM complements existing service management and what role ITIL plays in a service world shaped by AI? Then be sure to read the previous post:
“HUCAISM and ITIL: How AI Is Redefining Service Management”
Training Tip: HUCAISM AI Service Professional at SERVIEW
If you want to understand how responsibilities, human oversight, and AI-powered services are designed with HUCAISM, this is the HUCAISM Professional Trainingat SERVIEW is the perfect place to start. You’ll learn how to integrate AI responsibly into service organizations and how people can remain capable and accountable even as automation increases.
Learn more:
HUCAISM Training Courses at SERVIEW
