Leadership confidence & adoption
Video · 5 min read
The short answer
Watch: How can experienced leaders become more comfortable with the uncertainty around AI?
Watch the explanation, then explore the practical implications below.
Confidence should come from experience with the work
Uncertainty about AI is easier to address when it is attached to a specific task. “Can we trust AI?” is too broad to guide a decision. “Can this system help identify missing assumptions in this proposal, with review before use?” can be tested.
Experienced leaders bring knowledge of customers, delivery and commercial consequences. That knowledge is valuable when assessing output. It becomes more useful when paired with hands-on testing, rather than relying entirely on someone else’s demonstration.
Start with a problem you can judge
Choose familiar work with a clear purpose and manageable consequences. You might ask for questions to raise about a proposal, a critique of a completed decision or a summary of approved material. Avoid beginning with a live commitment that cannot easily be corrected.
Use the system approved by your organisation and information appropriate for that environment. Define what a useful result would look like before running the task. This gives you something more concrete to assess than whether the response sounds impressive.
Separate different sources of uncertainty
An unsatisfactory result may come from missing context, an ambiguous request, weak source information or limitations in the system. These require different responses. Rewrite an unclear brief, check the evidence and identify what the task actually demands before drawing a broad conclusion.
Keep track of errors as well as successes. A model that helps with one kind of analysis may still fail on another. Confidence should remain specific to the task and conditions you tested, including the review needed to use the result responsibly.
A practical first experiment
Take a completed proposal you know well and remove information that is unnecessary or unsuitable for the system. Ask AI to identify assumptions, unanswered customer questions and commitments that need clarification. Request references to the relevant supplied passages.
Compare the response with your own review. Did it find something useful? Did it invent a concern or overlook an obvious one? Adjust the brief and try a second example. Do not treat knowledge of the historical outcome as evidence that the model would have reached the same conclusion independently.
Learn openly with your team
A leader can say what they are testing, what helped and what was wrong without claiming mastery. That makes it easier for colleagues to report problems and compare practical experience. Pair people with complementary strengths: familiarity with the tool and knowledge of the work.
Avoid assumptions based on age or seniority. Assess what each person can contribute and what support they need. Someone may navigate the interface quickly while needing help to judge the commercial implications; another may recognise a weak recommendation but need help to frame the task.
Move from personal confidence to a shared process
Individual experiments become more useful when the learning is retained. Record the task, approved inputs, review criteria and limitations. Agree which findings should become shared guidance and who can approve changes to it.
Expand gradually into live work with clear responsibility and a route to stop or correct the process. Measure net effort, including review and rework, alongside the quality of completed work. A useful system should improve the task, not simply make people feel more confident about an unchecked answer.
How CSM would approach the work
We would start with real examples and help leaders choose a task they can evaluate. Working with the team, we would define the brief, information boundaries and review process, then use the results to identify a sensible next step.
Training should include how to challenge an answer and recover when information is missing. The objective is informed confidence: enough practical understanding to make decisions about adoption while remaining clear about what has and has not been demonstrated.
Questions for your leadership team
Which familiar task could I test this week?
Choose work where you can judge the response and contain the consequences of a mistake.
What would count as a useful result?
Set criteria before the experiment, including the review effort required.
What caused the errors we observed?
Distinguish a poor brief, missing context, unreliable evidence and a capability limitation.
Are we making it safe to report weaknesses?
Share concrete successes and mistakes, and show how feedback changes the implementation.
What evidence supports the next step?
Decide whether to refine, expand or stop the trial based on completed work rather than enthusiasm.
Full video transcript
The full video transcript is included below for reference.
I think there’s a danger of making this too much of a generational question. We assume younger people understand the technology and older people need to catch up. That doesn’t tell you very much about who can actually use it well in a business.
Knowing where the buttons are is useful. Knowing whether the answer makes commercial sense is useful too. A senior leader may have a great deal of knowledge that helps them ask better questions and recognise a poor recommendation.
But they do have to use it. Experience doesn’t help very much if you stay completely outside the process and wait for someone else to tell you what it means. I’d start with something you know well.
Give it a problem where you can judge the response. Ask it to analyse a decision you’ve already made, identify weaknesses in a proposal or prepare questions about an issue you’re dealing with. You’ll find things that are useful and things that are wrong.
Both matter. You’re developing a feel for where you can trust it and where you need to push back. And I think leaders need permission to be learners in front of their teams.
There’s a temptation to wait until you have a confident position before you say anything. But on something moving this quickly, that can mean waiting for a level of certainty that never arrives. You can say, “I’m testing this too.
Here’s something it helped me with. Here’s something it got wrong. What are you finding?” That creates a much more useful conversation than pretending either that you already understand everything or that this belongs entirely to someone younger.
The experience is an advantage if you bring it into the work. The problem is when experience becomes a reason to stand outside it.
Put these questions to work in your business
Bring a real decision or workflow to a conversation with CSM. We can explore the evidence, responsibilities and practical steps needed to make AI useful for your team.