This past quarter, a leadership team spent the better part of a meeting working out how to tell an executive-level colleague that his strategy document had obviously been written by AI, and they didn’t want to shame him. This is hours of assembled seniority, spent on the diplomacy of an open secret. We haven’t yet figured out the manners for doing this politely.
When I was a girl, I never understood the fairy tale about the emperor having no clothes. Now I get it and there are lots of naked people running around. Every organization runs an honesty threshold. Below it, someone can still call a thing absurd out loud. Above it, the same thing gets implemented with a straight face. The threshold moves with fear, and fear is rife in organizational life at the moment. So the question I have started asking the executives we work with, across completely disparate industries, is, “What does organizational ludicrousness look like now?”
Boards and industry bodies hand the mandate down as an instruction to be executed, and they hand it down believing the work will yield to a search and replace. It does not. AI takes maybe twenty percent of this job and thirty percent of that role, and people do not come apart in those proportions. You cannot carve the twenty percent out of a person and bank it somewhere else. Someone models the headcount reduction as one-to-one replacement, and everyone below a certain altitude can see the model is wrong while everyone above it cannot.
The strategic and economic pressures are real, and a board ignoring those would be failing at its job. Your board is right to push. The ludicrousness is more in the modeling, and in how mandates get handed down in ways that feel out of touch with reality, and then AI is used to meet those demands with work that feels good but is often insubstantial.
If you give a person an instruction they know is unfulfillable, and from a leader whose judgment they are supposed to be able to rely on, they stop relating to you in ways that matter. Social psychologists who study manager and subordinate pairs have found that the trust between them, and their capacity to influence one another, is a key predictor of workplace satisfaction and performance1, and anyone who has had to work for someone they don’t respect gets that. Anyone who has done change work in a system knows that change moves at the speed of trust. You are being asked to move faster than trust can carry you, and trust is an account that is notoriously difficult to replenish once it’s been spent down.
Then, when the pacing and implementation fails, someone has to hold the failure. Organizations are currently shoving accountability into a role — a person titularly becomes the accountable head of some department with AI in the name, there is no definition of progress, and lets everyone else off the hook. Since my days as the Wikimedia Foundation CHRO, I have long thought the HR function is a system’s liver, meant to process its toxicity with varying degrees of success. That job has now been handed to whoever has AI in their title.
And often the first thing to go with headcount cuts is Learning and Development, along with the entry-level roles where people build the discernment to tell slop from quality. Cut those and the discernment tax is disproportionately paid by executives who already put in the time on their own domain expertise, which is how your most expensive people end up spending an impossible day doing QA.
So, back to the threshold, and asking, “What is our ludicrousness tolerance right now?” I have to ask it of myself first, since I lead AI integration at Cultivating Leadership (with AI in my title), and we work with these same organizations, in this same market, under the same pressure to have a confident answer about this technology which is so rapidly evolving. We are in the middle of a reorganization and we have told our clients we are experimenting with it. The bet is that it makes us more adaptable and shapes where our attention goes. We don’t know yet.
My colleague Heidi Brooks, who also teaches at Yale, points out that nearly all our AI conversations live in a future nobody can picture, while almost nobody is asking how we might be with each other in the context of the confusion that is happening now.
AI has gotten incredibly good at pattern recognition but this is not wisdom and the two should not be confused. Pattern recognition is built on predictability and plausibility, but neither of those actually hold particularities —- the particularities of a moment, a team, the particular fears of individuals. Without the latter, actual responsiveness with wisdom isn’t possible. This is where the discernment work is so important and does not scale and pace well with organizational mandates.
What we know from the field of complexity, and years of working with it before it became a buzzword, is that you can’t just manage your way out. You have to raise the honesty threshold, make it a little more possible for people to name what is honest and true and possibly absurd, and moving at the speed of trust. This speed probably won’t feel gentler, but it probably will be a pace that allows change to actually take hold.
On October 7th, I’m hosting a conversation with a few leaders who are navigating these circumstances across, very different industries, about what they’re seeing and what’s actually helping. If any of this is familiar, come and be with us as we grapple together for awhile because that’s frankly what navigating this world well takes: https://www.cultivatingleadership.com/courses/leading-through-ai-transformation
- Kurt T. Dirks and Donald L. Ferrin, “Trust in Leadership: Meta-Analytic Findings and Implications for Research and Practice,” Journal of Applied Psychology 87, no. 4 (2002): 611; Charlotte R. Gerstner and David V. Day, “Meta-Analytic Review of Leader–Member Exchange Theory: Correlates and Construct Issues,” Journal of Applied Psychology 82, no. 6 (1997): 827. ↩︎
