How Leaders Can Build Trust During AI Change

Part two of our four-week series “AI and Human Potential”, a practical guide to managing change with clarity, empathy and confidence.

Trust during AI change is not a speech. It is a repeatable pattern people can rely on while roles, tools and routines evolve. Most colleagues are not really asking about models or prompts. They want to know whether they still matter here. If you want AI to help your people do better work and feel better about their work, cultivate steady habits that make intent visible. Leaders earn trust during AI change through visible routines, not slogans.

Start by naming the shift with precision. We are using AI to augment people, not to sideline them. That single line sets the tone for many small choices about what you automate, what you protect and how you support the learning curve. A large field study in a Fortune 500 support centre is a useful reality check [1]. A conversational assistant improved productivity on average and supported less experienced colleagues in particular. Encouraging, and still something that needs coaching and good job design around it.

Make the change charter visible and repeatable.

Clarity grows trust because silence creates stories. Do not wait for perfect information. Share what is known, what is undecided and how choices will be made. A simple rhythm helps people stay oriented.

• Purpose. Why we are doing this, for example to reduce toil so people can focus on work only humans can do.

• Principles. What will not be compromised, including accuracy checks, bias testing, privacy and a human in the loop for sensitive decisions.

• Path. The sequence we will follow, such as pilot, learn and scale, with checkpoints to pause or adjust.

A 60 second cadence you can start this week

Monday 9.05 team huddle. One known, one unknown, one next step. Friday 16.00 note. Five things we learned, five open questions and who owns them. That rhythm keeps people informed even as decisions evolve.

Treat this as your change charter. Publish it, revisit it in retros and adjust it in public so people can see the learning curve at work. For expectation setting, the Stanford AI Index is a helpful touchstone. Effects are task specific and tend to be stronger where quality can be verified quickly, which keeps you away from all-or-nothing stories that erode trust [2].  

One reason clarity matters now is structural. Senior tech and operations leaders told the Wall Street Journal they are rewiring org charts around skills, with fewer layers and broader spans of control. If you flatten the structure, you need higher transparency so people understand how work and decisions will flow. Publish who owns what, how AI assisted work is reviewed and where escalation lives. That is how a leaner shape does not turn into confusion [3].

Care is essential because AI change is first an emotional experience and only then a workflow change. If anxiety spikes, people either rush in without thinking or freeze. Name the human experience before you redesign processes. Your Emotional Growth Journey: A Practical Guide [6] offers a quick way to spot your stage and pick one stabilising practice. Pair it with Embracing Emotions as Catalysts for Change and Personal Transformation [7] to turn emotion into energy for action when uncertainty rises.

Credibility is where words meet visible fairness. Publish evaluation criteria before a pilot. Keep them in plain language so anyone can see how accuracy is checked, how bias is tested, what data is allowed and what is out of bounds, and who approves final outputs. Then show your working. Report what improved and how you reinvested time saved. Recent analysis from the St Louis Fed suggests typical users who adopt generative AI save roughly two hours a week on average [4]. Modest, and real. People notice when leaders turn those hours into coaching, customer listening or creative exploration, rather than folding them quietly into the next efficiency target.

It is also wise to be honest and open about limits. Human and AI combinations do not automatically outperform either alone. A recent review found that hybrids can underperform the best human or the best AI on decision-heavy tasks, and are often stronger on content creation [5]. The lesson is to pick the right jobs and keep ownership clear.

A tiny AI use policy you can publish today

We use AI to reduce toil and raise quality. We do not enter sensitive or personal data. We check accuracy and potential bias before sharing externally. For significant decisions, a human reviews and is accountable. When in doubt, ask. Keeping this visible prevents confusion as roles evolve.

To keep things practical, two light routines help.

• A short Friday update that shares five things now known about your AI use and five open questions with the plan to answer them.

• A learning ledger where each person notes one AI assisted improvement every fortnight and one human conversation that the time saved enabled.

Something for you to reflect on. Where will you show, not only say, that the hours saved through AI this month have been reinvested in people and purpose.

Coming up next in the series: Emotional Intelligence in the Age of AI. We will explore the EQ habits that protect dignity and performance as you adopt new tools.

Learn more about Coaching for the AI Era and how I support leaders to build trust through change.

FAQ

Is it honest to say “we don’t know yet”. Yes. Trust grows when leaders state what is known, what is undecided and how choices will be made. Your cadence covers this.

Do flatter structures mean more burnout. Not if you raise transparency and support. Publish ownership, review points and escalation paths so workload is clear.

References

[1] Brynjolfsson E, Li D, Raymond L. Generative AI at Work. The Quarterly Journal of Economics. 2025. Oxford University Press.

[2] Stanford Institute for Human-Centred AI. AI Index Report 2025. 2025.

[3] The Wall Street Journal. AI Is Turning Traditional Corporate Org Charts Upside Down. 16 Sep 2025.

[4] Federal Reserve Bank of St. Louis. The Impact of Generative AI on Work Productivity. 27 Feb 2025.

[5] Vaccaro M, Almaatouq A, Malone T. When combinations of humans and AI are useful. Nature Human Behaviour. 2024. Also summarised by MIT Sloan.

[6] CO Coaching. Your Emotional Growth Journey: A Practical Guide. 23 Dec 2024.

[7] CO Coaching. Embracing Emotions as Catalysts for Change and Personal Transformation. 20 Dec 2024.