Artificial intelligence has quickly become a fixture in executive education. But according to NEOMA Business School Associate Dean for Digital Alain Goudey, most programmes are still missing the central leadership challenge: judgment. In this conversation, he explains why executives need to spend less time learning tools and more time defining the boundaries of delegation.

Here are his words…

How are executives currently being taught about AI, and what important elements may still be missing from that approach?

Walk into almost any executive AI course in Europe today, and you will find a tools tour dressed up as a strategy seminar. The vendors change, the interfaces evolve, the slides are updated every quarter, and participants leave impressed by what the technology can do.

But fluency with tools is not what executives are actually paid for. The real challenge is judgment: knowing when to trust an algorithmic output, when to question it, and when to override it entirely. That is the difficult part, and it is the part very few programmes teach explicitly.

Right now, we are creating leaders who can prompt systems efficiently, but who have never been trained to ask the harder governance questions. That gap is becoming dangerous.

Is there a particular example that captures this challenge?

Absolutely. Last year, a founder joined our Executive Certificate in Generative AI for Business. She runs a small communications agency specialising in wine and tourism, and like many executives, she was overwhelmed by operational work.

Campaign reports and synthesis tasks were consuming one or two full days every week. She recognised immediately that generative AI could help recover time. What she lacked was not technical knowledge, but clarity about delegation: which tasks should be automated, which should remain human, and how to bring her employees into the process without making them feel threatened.

By the end of the programme, she had built AI-supported processes for drafting campaign reports. Tasks that previously took two days now take a morning. But the most important decision she made concerned what she would not automate.

Client relationships remain human. Strategy remains human. And before expanding AI further inside the agency, she decided she would personally train her three employees so the technology would amplify their work rather than replace it.

She told me something I still remember: “When you work with winemakers whose hands are in the soil all day, the human dimension is not a slogan. AI is a crutch, useful, nothing more.”

That, for me, is leadership.

What kinds of risks around AI are emerging in organisations today, and where are leadership and management practices most challenged?

The failure mode is already visible. Boards appoint Chief AI Officers and call it governance. Marketing teams accept AI generated media plans without interrogating the assumptions hidden inside them. Executives approve deployments without considering environmental impact, strategic dependency, or the long-term effect on employees’ cognitive autonomy.

And when the model is wrong, which it will be, perhaps quite often, nobody has been trained to recognise it, let alone challenge it.

Too many organisations are outsourcing judgment quietly, without noticing they are doing so.

How does NEOMA approach this differently?

At NEOMA Business School, one of the most important parts of our Executive Certificate in Generative AI for Business is a half day module called Step Back.

We created it because every AI course risks becoming a product demonstration, and executives do not need demonstrations. They need frameworks for decision making.

In Step Back, we ask participants four questions:

• What is the environmental cost of the model you are deploying at scale?
• What happens to your team’s cognitive autonomy when every first draft becomes machine written?
• What strategic dependencies are you creating, and on whom?
• And when the algorithm is wrong, who inside your organisation has the authority, knowledge, and courage to override it?

These questions are intentionally uncomfortable. But they are precisely the questions that separate executives who govern AI from executives who are governed by it.

NEOMA has trained thousands of people on AI. What have you learned from that experience?

Over the past three years, we have trained more than 12,000 people: 90 per cent of our faculty, half of our administrative staff, around 9,000 students, and nearly a thousand executives through certificates and in company programmes.

The systemic nature of this approach was recognised by AACSB through its 2024 Innovations That Inspire award.

But the number that matters most to me is much smaller. It is the number of participants who arrive looking for productivity gains and leave having drawn a deliberate line around what they refuse to delegate.

That line is the leadership act. Everything else is plumbing.

What distinguishes the most effective AI leaders?

The best AI leaders are not necessarily the people with the cleverest prompts, even if prompt design and context window management are becoming important skills.

What distinguishes them is that they have done the harder work in advance. They have decided what must remain human, and they can defend those decisions, to their board, to their employees, and to themselves.

The fluent executives often outsource judgment without realising it. The thoughtful executives build a perimeter and hold it.

Has this changed the way you assess and teach executives?

Completely. For example, our marketing executives are not assessed on their ability to generate an AI campaign. They are assessed on their ability to dismantle one, identifying attribution errors, training data biases, or reputational risks the model itself will never flag.

We have also built AI negotiation simulators where learners interact with AI counterparts displaying different negotiating styles and profiles. The goal is not technical mastery. It is rehearsing judgment before it becomes necessary in real business situations.

And when we selected a frontier AI partner in 2025, we chose Mistral AI partly because we wanted to apply the same scrutiny to ourselves that we ask executives to apply to their own organisations. If we teach leaders to question environmental and geopolitical implications, we must be willing to do the same.

There is also a broader European dimension here. We believe stronger collaboration between academia and the AI ecosystem is essential if Europe wants to build a credible and responsible alternative model of innovation.

What is the one question every chief executive should ask themselves about AI?

I would ask every CEO to forget, for a moment, what their AI strategy enables. Instead, write down the three decisions in your organisation that no algorithm will ever make on your behalf.

If you cannot finish the list, then your AI strategy is probably not a strategy at all. It is a delegation you simply have not noticed yet.

For additional insights, trends, and perspectives, visit the conversation on Artificial Intelligence in Business Education.