The sentence sounds harmless. It won’t stay that way for long.
Artificial intelligence in the workplace is increasingly influencing decisions that directly affect people. AI is already being used in recruitment, employee evaluations, and workforce scheduling. Soon, it may play an even greater role in deciding who gets promoted, who receives certain opportunities, and whose career moves forward.
Depending on who says “AI suggested it,” the sentence can be an explanation – or the first step toward giving up responsibility.
Because a list that has already been ranked looks like an answer. And people tend to question answers less often than suggestions.
Artificial Intelligence Is Not the Problem
I consider AI to be one of the most exciting developments of recent years.
The numbers also provide little reason for sweeping doomsday scenarios. According to a study by the International Labour Organization, around one in four jobs worldwide is exposed to generative AI, while only about 3.3 percent of global employment falls into the highest exposure category.
The more likely consequence is therefore not the complete replacement of humans, but a profound transformation of work through AI. Jobs will change from within, while the title on the business card may remain exactly the same.
The crucial question is no longer whether AI will change the way we work, but how we choose to shape that change.
The same technology can reduce workloads or increase pressure. It can help correct human bias or embed existing inequalities into software, giving them the appearance of objective decisions.
AI is rarely inherently good or bad. It often amplifies what is already present within an organization.
AI and Efficiency: Faster Does Not Automatically Mean Better
Almost every AI project I work on begins with a question about savings and efficiency.
That is legitimate. Organizations want to understand how artificial intelligence can make processes more efficient.
But efficiency alone is not a compass.
It only tells us how quickly we can get somewhere. It does not tell us whether that is where we wanted to go in the first place.
A recruitment process can be faster and still be unfair. An employee evaluation can appear data-driven while overlooking precisely those contributions that never appear in a system – such as someone who has held a team together for months, resolved conflicts, or quietly taken on additional responsibility.
Technology can tell us what is probable.
What is right cannot be delegated to AI.
Human Responsibility in AI: More Than the Final Click
Many organizations reassure themselves by saying: “A human still makes the final decision.”
But that alone is not enough.
The European Union’s AI Act explicitly addresses human oversight of AI systems and the risk of automation bias: the tendency of people to place excessive trust in the outputs generated by automated systems.
The mere presence of a human being does not automatically constitute meaningful human oversight.
If someone simply approves an AI recommendation without understanding the underlying data, without knowing the potential sources of error, and without having sufficient time to critically review the result, that person may formally be part of the decision-making process.
In reality, they are little more than the final click.
They carry responsibility without having truly made the decision.
Not an attractive position to be in.
In my view, responsible AI implementation in organizations therefore requires three things that cannot simply be purchased as software:
People who are allowed to challenge decisions.
Time to properly review results.
And an organizational culture in which saying, “I don’t understand how this result was reached” is considered a sign of competence rather than weakness.
AI in Organizations: Participation Is Not an Obstacle
An OECD survey involving more than 5,000 employees and 2,000 companies shows that training and employee participation are measurably associated with better outcomes for workers.
That may sound obvious.
In practice, it is anything but.
Participation takes time – and time is often one of the scarcest resources in transformation projects.
However, organizations that involve employees early when introducing AI into the workplace gain much more than an understanding of people’s concerns.
They gain knowledge that often exists nowhere else.
Which tasks consume unnecessary amounts of time?
Which processes could realistically be automated?
And where does human experience remain indispensable?
That last question rarely appears in a process diagram.
Stricter European requirements for so-called high-risk AI systems, which include certain AI applications related to employment and recruitment, are intended to provide additional safeguards.
Four Questions to Ask Before Introducing AI
Before implementing AI systems, organizations should therefore discuss more than technology, costs, and productivity.
In my view, at least four questions belong at the beginning of every AI project.
1. Who Actually Gains Time Through AI?
And who simply ends up with more tasks?
Productivity gains rarely distribute themselves fairly.
2. Which Decisions Will Continue to Be Made by Humans?
This should be clearly defined and documented.
Because what is not explicitly established may eventually be transferred to the system.
3. Who Is Allowed to Challenge an AI Decision?
And more importantly: What happens when they do?
A right to object without consequences or a meaningful review process is not an effective control mechanism. It is a ritual.
4. Who Is Responsible When Something Goes Wrong?
If this question cannot be answered clearly in the project plan, it will be even harder to answer after an incident.
The Future of Work with AI Requires a Clear Stance
The future of work with artificial intelligence will not be shaped solely by technology companies and software developers.
It is being shaped every day – through workplace agreements, implementation plans, leadership decisions, and the willingness of someone in a meeting to say that they do not understand how a particular result was reached.
I firmly believe that we can shape this development together.
But that is also where things become uncomfortable:
Those who have the power to shape the future can no longer hide behind the idea that change is inevitable.
Machines will continue to become more capable. That much is certain.
But responsibility for decisions that affect people should remain human.
About Marlow D. Guttmann
Marlow D. Guttmann is a keynote speaker, people and organizational development expert, university lecturer, and author. His work focuses on the intersection of artificial intelligence, ethics, leadership, and organizational culture.
In his keynotes, he shows organizations and leaders how to use AI not only efficiently, but also responsibly and effectively – while ensuring that human judgment and responsibility remain at the heart of technological transformation.
