Psychological safety is the missing layer in most AI strategies
“Imagine me going back to my sergeant and saying I spent the afternoon doodling.”
That was one of the first things a military member said to me when I introduced mindful doodling as a stress management tool.
At the time, I was delivering programs for Landing Strong, a non-profit that supports military members, veterans, and first responders living with stress injuries, including PTSD. When he followed that first comment with, “I don’t have a creative bone in my body,” I knew I was hearing more than simple skepticism.
I heard a lot of resistance in that work, but this one stayed with me because I could tell it was not really about the doodling. It was about fear of judgment. It was about identity. It was about the discomfort of trying something unfamiliar without knowing whether it would work, or whether you would look foolish in the process.
So I said, “Fair enough. But for the next 60 minutes, can you give yourself permission to be curious, without judging yourself?”
To his credit, he did. He stayed, he followed along, and even though he dismissed it at the start, he began using the practice almost every day. A few years later, I saw him again at an art show. He had pushed himself far outside his comfort zone and was now proudly sharing his own work.
What he needed was not more pressure. He needed permission. He needed a safe place to try something new without feeling judged or exposed.
That is the same resistance I see every day with AI.
I hear people say things like, “I’m not techy,” “I’m too old to learn this,” or “I don’t want to look stupid.” Most leaders assume this is a skills problem, but very often it is not. It is a psychological safety problem.
People do not resist change because they are lazy. They resist when change makes them feel exposed. And when something feels risky and unclear, the brain will always default back to what feels safe.
What happens when people don’t feel safe
For many employees, AI is the biggest shift they have faced in their working lives, and they are being asked to figure it out while still keeping up with their day-to-day responsibilities. When people feel uncertain and unsupported, they do what most human beings do. They protect themselves.
According to Ivanti’s 2025 Technology at Work Report, 32% of professionals are intentionally hiding their AI use from leadership. Some want to keep it as a secret advantage. Some are worried about job security. Others are concerned that their abilities will be questioned if leadership knows they are using it.
The part many leaders miss is this: nearly half of workers are keeping their AI use private, while 60% of employers believe their teams are being transparent. That tells us there is a gap, and that gap is not really about tools. It is about trust. If your team is quiet about AI, that is not just a usage problem. It is a signal.
The mistake most AI strategies are making
Most leaders I speak to are doing what feels logical. They focus on tools, invest in training, and create policies. Those things do matter, but they often skip the one condition that determines whether any of it actually works: psychological safety.
Harvard professor Amy Edmondson defines psychological safety as a workplace where people feel they can speak up, ask questions, admit mistakes, and share concerns without fear of being judged, embarrassed, or punished. That is the formal definition. In practice, when it comes to AI, it means something even more specific.
It means recognizing what you are actually asking your team to do. You are asking them to try something they do not fully understand, change how they have worked for years, make mistakes in front of other people while they learn, raise risks leadership may not see yet, admit when they do not know what they are doing, and in some cases rethink what their role might become.
Every one of those things carries risk. If your team does not feel safe, they will not take those risks openly. They will either avoid AI altogether or use it quietly where no one can see.
The question leaders need to ask
The real question is not, “Have we rolled out AI?”
The real question is, “Have we created an environment where people feel safe enough to actually use it?”
Because psychological safety is not a nice-to-have. It is the difference between AI being technically available and AI actually changing how your business works.
Where most organizations get stuck
This is also why so many organizations get stuck. Everything in the market is pushing leaders in a different direction. The emphasis is on tools, platforms, implementation, and speed. Very little is encouraging leaders to slow down long enough to think about the human side of change.
So leaders follow the playbook that feels responsible. They buy a subscription that looks credible, bring in a demo, create a policy, and tell their team to start using it. None of those steps are wrong, but they are often happening in the wrong order.
What comes next
If psychological safety is the missing piece, then the next question is where to start. This is where most organizations need a clearer path, and it is where the right structure makes all the difference.
The right order for AI readiness
When I work with leadership teams, I follow a simple sequence. Not because it is complicated, but because most organizations are doing it in the wrong order.
The three steps are simple: make AI use visible and safe, build confidence through real work, and scale what actually works. Most leaders instinctively want to start at step two, and that makes sense because tool adoption is what the market is pushing, what vendors are selling, and what boards are asking about. It feels like progress. But without step one, AI adoption almost always stalls.
Step 1. Make AI use visible and safe
The first step is not about tools. It is about visibility and safety.
Right now, in many organizations, AI use is already happening. It is just not happening in a way leadership can clearly see or guide. Some employees are experimenting quietly on their own. Others are avoiding it altogether because they are unsure what is allowed. Most are somewhere in the middle, trying to figure it out without clear direction.
This creates two problems at the same time. Risk increases, and adoption stays inconsistent.
So the first step is to bring AI into the open in a way that feels safe and manageable for your team. That means being clear about which tools are approved, what kind of data can and cannot be used, and where people can go if they are unsure. It also means being explicit about why AI matters in your organization and how you expect it to be used.
This is not about creating a perfect policy document. It is about removing uncertainty.
When people understand what is safe and what is expected, they are far more likely to engage. And just as importantly, they are far more likely to do it in a way that you can actually support and improve.
Step 2. Build confidence through real work
Once that foundation is in place, this is where things begin to shift.
But not because of more information. And not because of broad training sessions that try to cover everything at once.
Confidence is built through real work.
Most organizations make the mistake of trying to train everyone on everything, hoping that usage will follow. In reality, that approach often overwhelms people and leaves them unsure where to start.
What works better is much simpler. You focus on a small number of meaningful workflows where AI can actually make a difference.
You look at where your team is already spending time, where there is friction, or where quality could improve. Then you choose two or three areas to work on together. You test, you refine, and you learn as you go.
This is where confidence starts to build. Not in theory, but in practice.
It is also where people begin to develop judgment. They learn how to check outputs, how to ask better questions, and when to trust the tool and when to question it. That is what actually creates capability over time.
Step 3. Scale what actually works
At this point, most teams will start to see small wins. That is a good sign, but it is not the end goal.
A few individual successes will not change how your business operates. The real value comes when those successes become repeatable across the team.
This is where scaling comes in, and it is often misunderstood.
Scaling does not mean rolling everything out at once. It means taking what is working and making it easier for others to use.
That might look like writing down what is working in a simple way, creating shared examples or prompts, or building lightweight processes that help others follow the same approach. It also means continuing to support your team as they adopt new ways of working, instead of assuming they will just figure it out.
Over time, this is what shifts AI from being a personal productivity tool into something that actually improves how the business runs.
Why this order matters
The order here is not arbitrary. It is what makes the difference between momentum and frustration.
If you skip the first step, the second one becomes much harder. People are left guessing, which slows everything down and increases risk.
If you rush into scaling too quickly, you end up spreading inconsistency instead of results.
This is why so many organizations feel like they are making progress with AI, but are not seeing meaningful business impact. It is not because they are doing the wrong things. It is because they are doing them in the wrong order.
This is also why I do not treat AI adoption as a tooling exercise. Once the human foundation is in place, the next step is looking at where work can actually be redesigned in a practical way so AI supports the business, not just the individual.
Bringing it back to leadership
At its core, this is not just about AI. It is about how you lead your team through change.
AI is asking people to rethink how they work, how they learn, and in some cases how they see their role. That is not a small shift, and it should not be treated like one.
So the question for leaders is not simply whether AI has been introduced.
It is whether the conditions exist for people to actually engage with it in a meaningful way.
Start with the part most organizations skip
This is exactly where I begin with leadership teams inside the AI Readiness Accelerator.
It is a six-week customized program designed to help you build the foundation that makes everything else work. That includes clarity, alignment, and guardrails your team understands and trusts, not just tools.
If your team is resisting AI, hiding their use, or stuck in a cycle of trying things without real traction, this is usually where the issue sits. It is not in the technology itself. It is in the lack of structure around how it is being introduced and used.
Inside the program, we focus on helping your leadership team get aligned on the role AI plays in your organization, create practical and responsible guardrails, and identify real opportunities where AI can improve work. From there, we build early wins that can actually be repeated and scaled.
The goal is not to leave you with a document that sits in a folder. The goal is to leave you with a clear path forward that your team understands and is far more likely to follow.
If this is something you are trying to figure out right now, you can start with a simple conversation.
Book a 20-Minute AI Clarity Call, and we can look at where you are, what is already happening inside your organization, and what your next step should be.
