The messiness of people doesn’t really bother me. I’ve spent more than thirty years in human resources circles, and I’m a psychologist. So I guess I signed up for that. A colleague even asked me once, during a challenging period at a particular company, “This is so messy. Doesn’t it upset you?” I said it didn’t, and she basically assured me that I would always have a job. There are few guarantees in life — but I do think the reliability of human messiness is one of them.
We’re a combination of self-interest, bias and interdependency. We’re dreamers that are risk averse. We are social animals with trust issues. That’s messy all in one body, let alone several bodies trying to create something coherent. Organisations are exactly that — trying to corral all that potential and all that instability into something functional.
Leaders can use AI to help improve teams. But the messiness of people must be baked into the solution.
The whole apparatus of leadership development is organised around the idea that the messiness is a defect to be engineered out. We build competency models. We benchmark against best practice. The underlying promise, rarely stated outright, is that with enough frameworks a human team can be made to behave like a well-run machine.
And I think AI has raised the stakes on that promise. And not quietly: we are all a little AI-intense right now. That’s not a terrible thing. And, side note, it’s me as much as anyone — I’ve spent three decades solving people problems, but a lot of that was at tech-forward, Silicon Valley-mindset places — so I am definitely playing with this shiny new toy… a lot. We are all super excited about the idea that this is the thing that breaks open the next level of information and progress and solutions.
That’s the human bias towards innovation, but on top of that, it’s the tool itself. When a tool produces answers that look immaculate we collapse the quality of the form with the reliability of the answer. That’s just the mundane(ish) stuff I have warned about before: if you don’t understand the science of baking, don’t ask AI to create a cake recipe. It will sound amazing. It may be inedible. And you won’t know, because it sounded so perfect when it wrote it.
But when we use the tool to help us identify the right type of person, it’s a bit of a disaster in a different way. People are a compromise. You will give up something in order to get something else. You won’t get high competence without some ego. You won’t get innovation without a little breakage, or a steady hand without a little too much rigour.
And AI explaining how a team “should” operate, or what a leader “should” be able to achieve, has a baked-in problem. Everyone has a pretty low ceiling compared to an AI render of a human. Everyone is complicated, has different motivations, and will act irrationally at times. But AI can make it easy to believe that people can and should be consistent. They can’t. They never could. The models will always be describing a person who doesn’t exist. The problem is baked into the prompt: if you are asking for the “best”, “optimal”, “ideal” version, you’re setting a destination you will never reach.
Our grasp for perfection is much older than AI. It remains beyond our reach.
By the way, we were always doing this. Most business school educations and management consultants are selling how to improve human behaviour. And they both do it — but improving is a long way from perfecting. We’ve a strong bias to getting better at things and that’s worked for civilisation… pretty much always, with spurts like the industrial age and now (and much further back, but let’s stay focused). But we also have a high failure rate. Great companies don’t evolve enough to stay relevant. Terrible companies started off meteoric but couldn’t quite land. Why did Myspace go one way and Facebook another? Humans are messy, and organisations of messy humans are also messy. Some too messy to survive.
But why do the ones that last make it so much longer? The leadership team learns to channel the messiness. Years ago, companies used a military model successfully because most men who became employees had served in the military. (Gen X was the first American generation not to do compulsory military service. Perhaps it’s not a coincidence that the model began to really fall out of fashion by the 1990s, as those Gen Xers entered the workforce.)
So the military model reduced messiness with rules and conformity. That is certainly one way to do it. Turn of this century, we had open-plan offices — organisational architectural psychology: fewer walls, more transparency, more productivity. We also began to build creative outlets for all that restless messy energy: whiteboard walls and foosball tables and canteens stocked with energy bars. This more recent era didn’t have workers with a shared military history. And, the digital era has thrived more on innovation than production scalability that passes QA standards.
AI should help leaders with this even more because the question is still how you lean into messy. And doing so in ways that could create more innovation without abandoning the fixed reality of the mess is possibly the competitive advantage.
And if you need proof that the channel matters more than the perfection of the people inside it, look at what happens to leaders when the context shifts under them. I’ve watched people get quietly written off — yesterday’s results, wrong temperament, or just not the top of anyone’s list — and then the market turns, or a real crisis happens, and now that person is the person. They didn’t change; the circumstances did. (I’ve written before about fast starters versus fast finishers because these pivots are not uncommon.) I’ve also seen the reverse: leaders who look brilliant right up until …they don’t. There is no “ideal leader” sitting outside of circumstance. There is only this person, this team, this moment, and the question of whether you can make the combination work. And then pivot when it stops working. Because it will stop. Messiness can be contained or redirected. But not eradicated.
Using AI to help us corral human messiness, instead of removing it, is the right goal. Maybe the only realistic one.
And that’s the job. Not perfecting people, just matching them, imperfectly, to problems worth solving, and accepting that the friction is where the ingenuity tends to come from. It’s ok that AI can’t raise us above the level our imperfections hold us to. The innovation that comes from the collision of messy people trying to do a thing together is how so many unusually innovative ideas come about. A tech brain and a design brain. A legal brain and a marketing one. The inability to truly reconcile to one perfect machine is the very reason we’ve been able to build something that is beyond the scope of any one person. Maybe AI can only help around the edges. That’s still a lot of space to improve outputs, even if you can’t perfect the people.
