You Cannot Delegate Your Way Into the AI Era

For most of my career, seniority came with distance.

The more experienced you became, the less often you were expected to touch the actual tools. Someone else opened the file. Someone else built the prototype. Someone else prepared the deck that explained what everyone else had done.

Leadership meant seeing the larger picture.

That still matters.

But AI has introduced a dangerous new kind of distance: leaders making decisions about a technology they experience mainly through presentations prepared by people who report to them.

They know the budget.

They know the vendor.

They know the phrase “AI transformation.”

They just do not use AI.

This is how companies end up with an AI strategy consisting of 74 beautifully aligned slides, already outdated somewhere around slide 12.

You cannot lead a transformation you only understand from a dashboard.

And you cannot delegate your way into the AI era.

You cannot outsource curiosity

Delegation is one of the basic skills of leadership. It creates focus, distributes responsibility, and prevents the founder from approving every shade of grey in the product interface.

But delegation becomes dangerous when we use it to avoid understanding.

A leader can delegate implementation.

A leader cannot delegate curiosity.

If AI is beginning to influence how your team researches, writes, designs, recruits, analyses, sells, and makes decisions, then using it personally is no longer optional. You do not need to become an engineer. You do need enough direct experience to recognise what the technology changes, where it fails, and what questions your team should be asking.

An hour of real use will teach you more than a week of AI thought-leadership carousels.

Ask it to challenge a strategy.

Use it to analyse a document you know deeply.

Give it a creative brief and watch where the output becomes generic.

Ask for sources and check whether they exist.

Let it impress you. Then let it disappoint you.

Both experiences are necessary.

Without them, leaders tend to fall into one of two camps. They either treat AI as magic, or dismiss it as hype.

Neither position is leadership.

They are both forms of distance.

Fluency is not technical mastery

The conversation around AI literacy often becomes unnecessarily intimidating.

You do not need to understand every model architecture. You do not need to read research papers before breakfast. You do not need to speak in acronyms until everyone in the room quietly stops asking questions.

Fluency means something more practical.

What is this system genuinely good at?

Where is it likely to invent, flatten, or oversimplify?

What information should never be placed inside it?

Which decisions can be accelerated, and which decisions must remain slow?

Who is responsible when the output is wrong?

A fluent leader can have a serious conversation with technical people without pretending to be one of them. More importantly, that leader can protect the organisation from being seduced by confident output.

AI makes production look easy.

Judgment is still expensive.

Your team is watching your behaviour

Companies often approach AI adoption as a communication problem.

They announce a new initiative. They run a workshop. They publish a policy. Someone creates a cheerful internal channel called something like “AI Pioneers,” and three people use it enthusiastically for eleven days.

Then the energy disappears.

The problem is rarely the announcement.

People watch what leaders actually do.

If the leadership team asks everyone to experiment but punishes imperfect results, the real policy is clear.

If speed is rewarded while verification is ignored, the real policy is clear.

If executives talk about empowerment while every private conversation is about headcount reduction, the real policy is painfully clear.

Culture does not listen to the presentation.

Culture studies the behaviour.

A leader who uses AI openly can show the team something more valuable than confidence: honest uncertainty.

Here is where it helped.

Here is where it produced nonsense.

Here is what I nearly believed.

Here is why I rejected the answer.

This removes the theatre from adoption. It gives people permission to learn without pretending they already know.

Fear is not resistance. It is information.

People are not irrational for being worried about AI.

They have seen the headlines. They have heard companies describe employees as costs and automation as efficiency. They know that “this will free you to do more meaningful work” sometimes translates into “we would like the same output with fewer people.”

A motivational speech will not repair that gap.

Honesty might.

Leaders need to say what AI is being used for, what it is not being used for, how decisions will be reviewed, what data is protected, and what happens when a system gets something wrong.

They also need to create visible benefits for the people doing the work.

Remove repetitive administration.

Make research faster.

Help junior talent explore more directions.

Give experienced people more time for the difficult decisions their experience was built to make.

If the only person who benefits from AI is the person reading the quarterly numbers, adoption will always feel like surveillance wearing a friendly interface.

People do not need certainty.

They need evidence that leadership is not hiding the cost of change from them.

Experiment before you declare

The instinct of many organisations is to create the complete strategy before allowing serious experimentation.

That sequence made sense when technology changed slowly.

AI does not.

By the time a perfect roadmap survives the meetings, revisions, approvals, and final font correction, the tools have moved, the assumptions have changed, and the roadmap has become an expensive description of last quarter.

The better sequence is smaller and less theatrical. Choose a real problem.

Run a contained experiment. Measure what improved. Document what failed. Protect the people and the data involved. Then decide whether to stop, adjust, or expand. A pilot is not valuable because it proves AI works. It is valuable because it reveals the conditions under which it works for your organisation. That difference matters.

The goal is not to collect shiny demonstrations. The goal is to develop institutional judgment.

Leadership now includes the courage to remain a beginner

I have spent 27 years across design, creative direction, product, and technology.

Experience teaches you to recognise patterns faster. It also creates a temptation to protect the identity of being the person who knows.

AI keeps disrupting that comfort.

Almost every week, I encounter a new capability, limitation, or working method that forces me back into beginner mode. Some days that is exciting. Some days it is deeply annoying. Usually it is both before lunch.

But staying close to the work matters.

Not because leaders should do everything themselves.

Because direct contact changes the quality of the decisions they make for everyone else.

A leader who has struggled with a model understands why the brief matters.

A leader who has checked an invented source understands why verification cannot be a footnote.

A leader who has seen ten polished but identical concepts understands why taste and context become more valuable as production becomes easier.

A leader who remains a learner is less likely to confuse access with mastery.

The new leadership contract

The AI era does not require leaders to become machines. It requires something more human. Stay curious enough to touch the work. Stay humble enough to admit what you do not understand.

Stay honest enough to address fear without insulting people's intelligence. Create room for experiments that can fail safely. Use speed without allowing speed to replace thought.

And when the machine contributes to a decision, keep a human name next to the consequence. Your team does not need you to know every answer. They need to see that you are willing to enter the unknown with them.

Not from the balcony.

Not through a report. Not after someone else has translated the difficult parts into three convenient bullets. Leadership has always been about going first. This time, that means putting your own hands on the tools. You cannot delegate your way into the AI era.