To walk through Ray and Charles Eames’s house is to walk among thousands of objects they chose one at a time because they deeply understood craftsmanship in response to human necessity, and their careful and beautiful curation reflected their commitment to functional, beautiful, and aesthetic design as part of their legacy. Another legacy is the idea “never delegate understanding”, which is even more important in an age of AI. You can hand off some drafting, searching, summarizing. Keep the understanding, because once understanding is delegated to AI, the ability to check the quality of anything else you delegated goes with it.
I was once in Cambodia in the midst of floods, and while I really wanted to see trek out to these ancient temples, I remember being very dubious as I waded hip-deep through the water about what was floating beneath the surface, especially as the floodwaters overwhelmed the sewer systems that were supposed to contain it, and separate safe drinking water from distinctly unsafe water. I feel like we’re now wading through floods of plausible content — plausible decks, plausible memos, plausible strategy documents, and the most dangerous thing is plausible contracts and legalese. Every leader I talk to is drowning in work that looks finished. AI has decoupled the sheer volume of production from effort, so plausibility is now cheap at every level of an organization, partners included. What that produces is what I’ve been calling a “discernment tax,” paid by the reader who receives the output, and with increasing irritation. Most of the cost is borne by the middle layers, and it is absorbed in the working hours of the few people with the experience to pick apart nuance.
And unfortunately, that group of folks is a minority rather than a majority in many organizations. Expertise at its most useful sits at unconscious competence, the stage where you no longer know exactly how you know because you’ve already moved through conscious incompetence and conscious competence, both stages of which take a lot of energy. You glance at a document, and you just know something is wrong in the third paragraph, and you couldn’t explain why for another minute. It’s the hardest kind of knowledge to teach or disseminate because it’s so in the bones, and the people who have it sometimes can’t fully articulate it either. You can only distinguish excellent output from plausible output if you have done the thing yourself.
Which is also why it gets cut. Unconscious competence is the art of making something look effortless again after years of effort, the way great Olympian gymnasts defy gravity. Someone reads a deck for four seconds, says this won’t work, and from the outside it might look indistinguishable from not having done the work. The four seconds are the twenty years of experience.
And the discernment tax is not paid only in hours. Making something can be creative and generative, just like writing is for me. Most people who are good at their work know the particular tiredness that arrives after a day of building, and how unlike the other kind it is. Crafting is one thing, proofreading to try to tell dross from gold is a different kind of exhausting, and that’s not the thing that you want your most experienced people spending their time doing, shifting them from makers into checkers. Some of what those folks report as overwhelm is the indignity of checking work they never should have had to check and grief at the disappearing things.
The exhaustion isn’t in the checking itself. Reading a colleague’s draft closely is one of the more intimate things you can do at work, and people who do it well can find it generative. Part of what has changed is that checking used to be part of the context of a relationship and an opportunity for learning. Now it is a queue, and no one I’ve ever known has loved a queue. Well, maybe the British…
And then there is this phenomenon called answerability. When you put something out into the world that you’ve created, it gets tested by reality. When a model has a mishap, you give it a thumbs down and it generates other plausible things without real consequences to it. When humans misjudge, there are consequences and you learn and adapt or learn to live with them, and the learning and living is what becomes discernment over time. Falsifiability (letting reality give you disconfirming evidence), enactment (the actual doing of the thing), and consequence are ways humans are “answerable” to reality in ways that models aren’t. Expertise comes from that refinement against reality and the capacity to be wrong and to pay the price for it, especially the professional or social cost of it. And while I understand a world in which any reasonable person would want to diminish any opportunity for public exposure or humiliation, there is actually a beautiful point to it. Humiliation and humility share a root — humus, the ground. Being brought to earth and staying close to it are the same word.
So the thing becoming scarce is the thing being cut. Expertise gets more valuable while the conditions that grow it are dismantled. Entry-level work is where every profession has ever grown its people, and these are the jobs at most risk for being cut.
We have been here before at a smaller scale. A colleague of mine had this great metaphor that this is like when email came on the scene and “reply-all” with it, and it suddenly spawned work before protocols or etiquette existed for it, and we had to find ways to deal with it. (Years of my retired parents forwarding chain-mail emails and bad jokes on the internet…) This is like that, with the floodwaters rising faster than levees can be built.
Underneath the overwhelm is something else. Right now people are subject to things faster than they can make those things make sense to them. Meaning-making at speed is not what our wetware is built for, and no amount of tooling changes the rate at which a person can metabolize a change to their own work. So disruption arrives as something happening to you, and it becomes a teacher only when somebody accompanies you through it long enough for your nervous system to metabolize what is happening.
What is becoming scarce is judgment, discernment, decision-making, and frankly, ethics. If all of those are becoming scarce, so is wisdom. None of them can be delegated, and all of them are built the same slow way, including the capacity to hold more than one timeframe at once — moving at speed while also, necessarily, slowing down.
On October 7th I’m hosting a conversation with leaders across very different industries about exactly this — what’s being lost, what’s worth protecting, and what’s actually helping. If it resonates, come join us: https://www.cultivatingleadership.com/courses/leading-through-ai-transformation