The three postures of AI adoption: resist, ignore or enable

I’m deeply concerned about organisations whose response to AI is to ban it.

I understand the worries about confidentiality, accuracy and accountability. Those are real issues that need proper answers. But a blanket ban can leave people substantially less capable, discourage learning and basically opt you out of the future economy.

As a leader, you’re making a decision about how your people develop and how your business improves. Resisting AI has arguably much bigger consequences than embracing it and learning how to adapt and manage the risks.

I’d group the choices available to leaders into three postures: resist, ignore or enable.

1. Resist: “We don’t allow AI”

This is the stance that worries me most.

You don’t want confidential information pasted into an unapproved tool. You don’t want a client receiving a report full of invented facts. You don’t want people handing over their judgement to a chatbot.

But “therefore, nobody can use AI” is a poor response to those concerns.

It closes off the opportunity to learn how to use the technology properly. People need practice recognising where AI helps, where it fails and how to check its work. Preventing that practice leaves them less prepared for the next decision.

I’d be very uncomfortable leading a business where curious, capable employees had to hide their attempts to improve how they work.

Imagine someone discovers that AI can help turn an hour of messy notes into a useful first draft in ten minutes. If management’s only answer is “stop doing that”, what happens next?

Some people will stop. You lose the potential improvement.

Others may keep using it on their personal phone or laptop. You lose visibility.

Neither is a good outcome.

My bigger concern is what happens over time. While one organisation is debating whether employees should be allowed to experiment, another can be building skills, sharing useful approaches and improving its processes. That learning accumulates.

A business that bans AI may eventually change its policy. Building the capability it discouraged will take longer.

There are good reasons to restrict particular tools, types of data and uses. Make those restrictions specific, explain them clearly and provide an approved way forward.

“Don’t upload client information to an unapproved service” is a useful boundary. A blanket ban needs a much stronger justification, an owner and a plan for what comes next.

I don’t think “we’re being cautious” is enough to justify standing still indefinitely.

2. Ignore: “Don’t ask, don’t tell”

This posture avoids the argument.

People experiment. Managers suspect it’s happening. Perhaps the managers are doing it too. Nobody asks too many questions as long as the work gets done.

That gives people freedom, but it leaves some basic questions unanswered.

Which tools can handle company information? Who checks the output? What happens when someone makes a mistake? How does a useful discovery spread beyond the person who found it?

Consider two people doing the same job. One has worked out a reliable way to use AI to prepare a first draft, check it against the source material and finish the task faster. The other is still doing everything manually.

If AI use is something people quietly get away with, that improvement may never become a team practice.

You also make it harder for people to ask for help. An employee who is unsure whether something is safe should be able to raise it before they do it. Ambiguity makes that conversation harder.

3. Enable: “Let’s work out how to use this well”

This is the posture I’d aim for.

Leadership takes an active interest. People have access to suitable tools, understand the boundaries and can show each other what works.

The first step is for leaders to use AI themselves.

You’ll make better decisions about training, access and expectations if you’ve experienced both a useful result and a convincing wrong answer. Pick a real task, give it a go, and check the result carefully.

Then make it practical for the team:

• Provide approved tools. Make the approved route convenient enough that people have a reason to use it.
• Explain the boundaries. Be clear about what information can go where and which uses need approval.
• Keep responsibility clear. The person delivering the work remains responsible for checking it.
• Make questions welcome. People should be able to discuss uncertainty and report mistakes promptly.
• Share useful examples. Show the task, the approach, the checking involved and the result.

I’ve previously written about how AI can help employees contribute beyond their formal role. Leaders have a corresponding job: give that experimentation somewhere useful to go.

The trap: forcing usage

There’s another response worth watching for: mandating AI usage through quotas.

It can look like enthusiastic adoption.

You buy the licences, set a usage target and ask why some people aren’t hitting it. Soon, people have an incentive to put AI into tasks whether it helps or not.

A higher prompt count tells you very little about whether the work improved.

I’d ask people to demonstrate a useful application instead. What did they try? Did it save time after checking and corrections? Did it improve quality? Could someone else repeat it?

Sometimes the answer will be that AI made the task slower. You want people to be comfortable saying that too. It’s useful information.

Start with a conversation

At your next team meeting, ask:

“Where is AI already helping you, and what’s getting in the way?”

Then listen. You may find a workflow worth sharing, a boundary that needs clarifying or a tool people need access to.

If your organisation currently bans AI, I’d make reviewing that decision a priority. Identify the specific risks, decide what would make a limited trial acceptable, and give someone responsibility for moving it forward.

You don’t need every answer before people can start learning. You do need to take responsibility for helping them progress.