Incident Tech

Introduction: Will AI Replace Managers Too?

Hello, this is Nishihara.

This article is the middle installment of a three-part series, where I’d like to reflect on the relationship between AI and management — and where it’s headed.

In the previous piece, I explored how the center of gravity in work is shifting from creating to deciding in the age of AI. AI can write documents, organize information, and present options. But the judgment of what to aim for and which option to adopt remains on the human side.

So what will happen to management — one of the most prominent roles historically responsible for that judgment?

Management Is Not “Administration” — It’s the Work of Taking Responsibility

When people hear “management,” many picture the operational tasks of project management:

• Tracking team members and their assignments
• Following up on delayed tasks
• Scheduling meetings
• Reporting to superiors and clients
• Progress monitoring
• Task management

These are all important parts of a manager’s job. But the essence of management does not lie there.

The essence of management is making decisions under uncertainty and owning the consequences.

• Insufficient information
• Conflicting interests among stakeholders
• Pressing deadlines
• Residual risk

Even in these conditions, work cannot move forward unless someone decides.

In incident response, for example, there are moments when you cannot wait until every root cause is identified. Do you prioritize a temporary fix? Do you continue investigating the cause? When do you report to the customer? How broadly do you define the scope of impact?

As this illustrates, there is rarely a single best answer in business judgment. The job is to choose the better option from among many. That is precisely why the essence of management is not simple administration — it is taking ownership of choices.

AI Is Becoming Better Than Humans at Explaining and Organizing

When it comes to “explaining things,” there will be situations where AI outperforms humans. AI excels at:

• Rephrasing content to suit the audience
• Reducing technical jargon
• Structuring logical flow
• Explaining without emotional bias
• Responding consistently regardless of condition
• Fast turnaround

More concretely: tailoring an incident report for customers, internal teams, and executives; separating decisions from open items in a meeting; softening instructions so they don’t come across as too harsh. These are significant strengths of AI.

But being able to explain something is different from being able to take responsibility for it.

As noted above, AI excels at organizing and communicating information. However, it cannot make final judgments or bear responsibility for outcomes. What the output is used for, and how — work does not end the moment an output is produced.

Only Humans Can Hold Responsibility — And That Makes It More Important

At this point in time, AI cannot serve as an accountable party within an organization. If a report written by AI is sent to a customer, the responsibility lies not with the AI but with the person or organization that sent it. If a response strategy proposed by AI is adopted, the humans involved must own the outcome.

This is precisely why, in the age of AI, the importance of human accountability is increasing, not decreasing.

Because AI organizes information and supports communication, humans must be even more deliberate about what they ultimately choose and how they handle it. Judgment can be assisted by AI — but how that judgment is reflected in operations and communicated to customers and stakeholders is something that humans and organizations must take on themselves.

In AI-era management, what matters is not just the ability to use AI, but the commitment to take responsibility for the results of using it.

When AI Appears to Hold Responsibility, Humans Let Go of It

What happens when AI advances further — to the point where it appears to offer better explanations and judgments than humans?

This brings to mind the depiction of electronic brains that govern human society in Osamu Tezuka’s Phoenix (Hi no Tori).

In the “Future” arc referenced in a previous article, electronic brains in each nation act as proxies for much of human decision-making. In that world, people grow apathetic, and even clothing and food choices that deviate from the electronic brain’s instructions become restricted. I’ll leave the ending to the work itself — but I’ll add the footnote that the people in that world did not end up there out of laziness.

Even if AI output comes to be regarded as “correct,” if humans stop mediating judgment and stop accepting responsibility, the ones who suffer most are customers. When judgment adapted to the customer’s situation and the context of the front line is missing, AI output may look right but won’t translate into value. The result: the services and responses you believe you’re providing cease to be meaningful to the people receiving them.

When “the AI said so” becomes an acceptable excuse, humans lose the sense that they decided anything — and the awareness of owning outcomes fades with it. The temptation to think “I don’t need to think for myself” or “following AI is the safe choice” is hard for humans to resist. But yielding to it also means losing human diversity and the drive to improve.

Humans and AI have different strengths. It is through role division — not replacement — that better value can be delivered.

The Cost of Humans, the Cost of AI

Deciding whether to entrust something to a human or to AI also requires thinking through cost. Cost is generally considered across three dimensions:

• Time
• Effort
• Money

Comparing humans and AI along these three axes, AI appears to have the advantage:

• Time: Fast output generation
• Effort: Consistent output volume
• Money: Lower cost compared to labor

But there is a cost AI cannot escape: the cost of risk.

Risk is always present in work. Even if issues like understaffing, shortage of managers, over-reliance on individuals, training costs, and the burden of reporting are significant — responding to risk is a human task, one that requires accumulated experience and judgment.

Even if AI is trained with the right context and produces information suited to sound judgment, the people who act on that information are human.

That means the humans involved must understand, judge, and respond to the risk themselves. The cost of AI is not just the subscription fee. It includes the cost of reviewing outputs, applying contextual judgment, and handling them with a clear-eyed understanding of risk.

Conclusion: Should Management Be Done by Humans or AI?

In the end, should management be done by humans — or by AI?

This question cannot be answered with a simple binary. That’s because within management, there are parts that are easy to delegate to AI and parts that must not be.

What AI can handle well:

• Organizing meeting discussions
• Identifying risks
• Drafting reports
• Flagging gaps in perspective

In these areas, AI is a powerful supporting actor.

What humans must carry:

• Setting objectives
• Making final judgments
• Taking responsibility for outcomes

Owning the results and managing risk is the work of humans — and that becomes the core of what managers do.

From those who manage tasks → to those who ensure the quality of judgment. From those who gather information → to those who decide based on it. From administrators → to those who accept responsibility.

In the age of AI, the accountability dimension of management will be scrutinized more sharply than ever.

So how will communication — the core of interactions with AI and team members alike — change in this era? Next time, I’ll explore the ways communication becomes harder in an AI-driven world.