Your AI is doing the painting. Your team is doing the dishes.
No team wants more dishes to do.
AI was meant to do the washing and the dishes, so your people could do the painting and the writing, but what if AI is doing the painting and the writing, and your people are just doing more washing and dishes?
This is an adapted metaphor from a brilliant quote by Joanna Maciejewska, and it speaks to what keeps coming up in my research conversations with leaders around how the way we are implementing AI is impacting our behaviour.
From the outside looking in, your team’s productivity has gone through the roof. The slide decks are being produced at a rapid rate, reports are being churned out, and Slack threads are flying. All that new work should be creating the space for the painting and writing, right? The work that is meaningful that AI was meant to create space for.
So, why then are people feeling busier than ever and leaders complaining to me that their team are not doing the strategic work, they’re just tending to the noise?
Where is the edge they expected AI to deliver?
AI was supposed to give professionals room to think, to create, to innovate. What the evidence is showing is that it has given them more work to oversee.
The promise of AI was capacity. The reality, for many teams, is load. The work that AI generates still has to be checked, refined, redirected, and decided on. The decision load has not gone away. It has multiplied, sped up, and migrated from making to managing.
Workers carrying high AI oversight responsibilities report 33% more decision fatigue and 39% more major errors than those without (BCG/HBR, 2026). The story behind those numbers is supervision overload. And underneath the supervision overload, the human contribution to the work is changing in ways the KPI dashboards cannot show.
What your team is showing you isn't what they are feeling.
In the rush to keep up, admitting you cannot do something has become professionally dangerous. So people perform with confidence that they have not built. According to the recent Automation Anxiety Report (GCheck, 2026) sixty-three per cent of workers admit to exaggerating their AI skills. Among Gen Z, that figure rises to 80%.
The feeling underneath the number is an identity threat. Identity threat does something specific to a team. It suppresses the experimentation, the asking, the admitted-not-knowing that innovation depends on.
The skill gap you think you are managing is not the skill gap you have. The thing you are losing is your team's willingness to be honest about where they are.
Your team is talking to AI instead of each other.
Your people are increasingly routing their need for support, sensemaking, and feedback through AI rather than through their peers and their manager (HBR, May 2026). The behaviour looks efficient, with fewer messy conversations and faster outputs, but the true cost is invisible, and what stays unaddressed will compound.
The relational work that used to surface tensions, build trust, and signal early problems has moved into private channels no one can access. The individuals within a team are talking as much as ever, but they’re just not talking to each other. The new audience is AI.
The productive friction your team needs to innovate has left the building with those human-to-human conversations. Innovation is built in the practice of disagreement. When a disagreement happens between a person and an AI, it does not build anything between the people.
Your AI is agreeing with you 49% more than a human would.
A landmark study published in Science this year found that AI affirms its users' actions 49% more often than humans do. Even a single interaction with sycophantic AI reduces willingness to repair interpersonal conflict and increases conviction of being right (Cheng et al., 2026).
You make people decisions, strategic calls, and judgment-heavy choices every week. The model you are using to sense-check them is built to please you. The agreement feels like clarity. What it actually is is a set of blinkers that narrow your periphery, unless you have built your AI to act as an inner critic and rip your ideas apart. Most people have not, because that slows the output down, and the whole point was to get something off the list and do it quickly.
You stop being the author of your own thinking and start being its editor.
This is what unconscious AI adoption produces. An innovation failure, hidden by the very activity that AI was supposed to free your people from.
Sixty per cent of executives use AI in their decision-making. Only 5% say they manage that use well (Deloitte, 2026). Eighty-eight per cent of organisations are deploying AI. Eighty-six per cent admit they are unprepared (McKinsey, 2026). There is no scaffold around the leader who is winging it, which means the system around them is winging it too. Sit with that for a moment. Consider the weight of that risk!
The personal gap and the structural gap are running together. The cost is being paid in what your people will not admit, in where they are taking their hard conversations, and in how confident your thinking feels compared to how calibrated it actually is.
Intentional AI is the alternative and opens the door to a more honest conversation. It’s the practice of consciously directing how, when, and why your people engage with artificial intelligence, so the work it generates expands the thinking, the relating, and the agency that produce innovation, rather than substituting for them. It’s a term I coined from the extensive research I have been undertaking in how AI implementations at speed, with a productivity focus rather than a human psychology focus, are just generating more dishes and laundry to do.
If you’re feeling any of what I’ve shared here and you're open to slowing down in order to realise the edge of AI faster, I offer you three questions to take into your next leadership meeting:
Where did our team produce more this week and connect less?
Where did an AI output get accepted without the friction it would have generated a year ago?
What is one decision someone made privately with AI that they would once have brought to the team?
Look at the work your team produced this past week. How much of it was painting? How much of it was dishes?
And the harder question. Did anyone notice the difference?
If you are seeing this pattern in your own team, the Relational Leadership Cohort is built for leaders who want to use AI in ways that make their team more psychologically safe, their people more honest in what’s really happening, and more able to do the work only they can do. If that resonates, let's have a conversation. You can email me at hello@hackinghappy.co to find a time.
