The first and foundational tip sheet in our series for leaders – and anyone – navigating AI and culture change
You can buy the best AI tools on the market, hire the right consultants, and check all the boxes. And it will still go sideways if you ignore culture – and especially the people who create that culture. This is the part most organizations skip. But it’s the part that determines whether everything else works.
Getting AI Culture right – that is, the shared norms, values and practices related to AI – starts with understanding mindsets. Whether they know it or not, leaders and organizations usually have a mix of what we call AI mindsets. These mindsets seep out in subtle (or sometimes not so subtle) “tells” — what gets implemented and how, what gets communicated and how, what gets ignored, and how concerns get handled. Organizations that don’t understand these mindsets will shape their AI cultures by default. This rarely ends well. But if they do understand these mindsets, they can shape their AI cultures by design.
Which Mindset Most Describes Your Organization (Or You)?
In real life, organizations and people may have elements of each of these mindsets. But knowing these distinctions is useful.
The Productivity Purist. For this person, AI is primarily about efficiency and increased productivity. The goal is output, speed, and scale; culture or people concerns tend to register as friction, inconvenience, or distraction. The Productivity Purist isn’t necessarily wrong to care about productivity; they’re just operating with an incomplete model of how organizations actually work. The risk: tools get deployed that damage trust, employee engagement, and ironically, in the end, actually undermine productivity.
The Rudderless Reactor. This person means well but operates without a guiding framework. They may respond to whoever is loudest, adopt whatever solution is most available, and treat AI integration as a series of disconnected problems to patch rather than a system to design. This shows up as quick fixes, inconsistent messaging, and a lot of miscellaneous activity that doesn’t add up to a strategy. The risk: the appearance of progress while the underlying conditions don’t change and confusion abounds.
The Signal Seeker. This person understands that AI lands in a culture and that culture either amplifies or undermines whatever the technology is supposed to do. They listen before they act. They ask “who might be affected and how” rather than just asking “what’s possible.” They treat concerns from colleagues as intelligence, not inconvenience. The good news: this is a learnable mindset that all leaders and employees can adopt. The tips below will help you and your organization to get there, no matter where you sit in the organization.
For Leaders: How to Set the Conditions
1. Know that you are already communicating culture, whether you know it or not.
Employees read your priorities from what you fund, what you measure, what you reward, and what you ignore. They can sense if you are giving lipservice to their concerns and welfare. If attending to their concerns rarely comes up in all-hands meetings, in performance conversations, or in the budget, you have already sent a message. And you can’t try to fake it. You actually have to care and be curious.
2. Get curious before you get moving.
The instinct in most organizations is to deploy first and troubleshoot later. Signal Seekers reverse this. Before rolling out any AI tool that affects people ask who might be negatively affected and how you would know. Get input from different types of employees. This may seem like slowing down but it will prevent having to clean up avoidable messes made at speed.
3. Make psychological safety and employee engagement preconditions, not afterthoughts.
Resist the temptation to focus only or primarily on productivity. This almost always backfires. If you implement practices in ways that foster distrust, decrease employee engagement or make employees feel they will be penalized for speaking up, productivity typically goes down, not up. In contrast, psychological safety has been shown to be a top indicator of increases in productivity, early error detection, and innovative solutions.
4. Explicitly recognize and address the specific anxieties associated with AI implementation.
Change is always difficult. But anxiety is especially rampant when it comes to AI acceleration and implementation (e.g., about job security, about being surveilled by the very tools meant to help, about whether raising a concern will even matter). Leaders need to communicate clearly that they recognize these concerns and explain how they plan to address them. But again, people will know if you only give lip service to these concerns.
5. Ask disaggregated questions.
Aggregate data hides disparities and differences. “Employees are generally satisfied with our tools” tells you very little. “These specific types of employees report significantly lower confidence in how performance is evaluated” tells you something you can actually act on. Build the habit of asking — in surveys, in focus groups, in one-on-ones — to see if experiences differ across groups. Then actually look at what comes back and do something about it.
6. Remember that this Signal Seeker orientation is an ongoing practice not a one-time program.
Return to these questions regularly, build them into how you run meetings and make decisions, and make it visible to your team that you take them seriously and for the long haul.
For Everyone: How to Show Up and Speak Up
You don’t have to be in a leadership role to shape your organization’s AI culture. In fact, some of the most important Signal Seeker behavior happens at the individual and team level — in how people raise concerns, support colleagues, and push back on tools or processes that don’t seem right. These five tips are for everyone.
1. Trust what you notice.
If an AI tool seems to be producing results that feel off – screening out candidates who seem strong, rating some employees differently than their work suggests, or not capturing the things that matter the most. It’s important to trust that instinct and explore questions you have. You may be picking up on a real signal before anyone else has put it into words.
2. Don’t keep it to yourself.
Raising a concern about an AI tool provides the organization with crucial information. And you don’t have to have all the answers or proof of wrongdoing. “This doesn’t seem right and I’d like to understand it better” should be enough. The organizations that catch problems early are the ones where people feel empowered to say these kinds of things out loud.
3. Be sure to “close the loop” every time.
Nothing erodes trust faster than raising a concern and hearing nothing back. If you’re in a position to respond to someone who has flagged a problem — as a manager, a team lead, or a peer — do it. Tell them what was looked into, what was found, and what, if anything, is changing. Few things destroy trust more than silence.
4. Identify contributions that don’t show up in the data.
AI tools tend to measure what is easiest to measure: things like activity, speed, lines of code or limited kinds of output. They routinely miss the work that is more difficult to quantify: mentorship, culture-building, the question in the meeting that changes the direction of a project, and more complex deliverables. When you see that work happening around you, point it out and give credit where due – whether in team meetings, in peer feedback, or in conversations with managers. Remember that what stays invisible stays uncounted.
5. Push back out loud on productivity as the only, or even primary, metric. Don’t underestimate the importance of this one.
AI tools are often implemented primarily on efficiency grounds. Efficiency can, of course, matter. But when productivity becomes the only or primary lens, important things get missed and talented people get sidelined. When a tool or process is being evaluated primarily by speed, activity or limited types of output, it’s fair to ask: What is this not measuring? Whose contributions might this miss? How is this tool affecting trust and employee engagement? Normalizing those questions is how cultures shift, one question at a time.
Amid what can feel like a tsunami of AI acceleration, it is easy to feel overwhelmed. But this moment can be about more than just riding or surviving the wave. The organizations that come out of this turbulent moment stronger will be the ones who invite their people to ask important questions, push back on productivity purism, refuse to drift rudderless, and take the time to seek out the signals that surround them.