Five ways of knowing AI cannot reach
A few weeks ago I wrote about six new cognitive abilities that AI made possible. The argument was that AI doesn’t make knowledge workers dumber. It shifts the valuable work from production to judgment.
This piece is the other side of that ledger. Five ways of knowing that AI structurally cannot reach. Not “cannot yet.” Cannot. Each one requires something AI does not have. A body. A stake. A life.
Some of these will look less like knowing and more like character or feeling. That’s the point. The kinds of knowing AI cannot reach are the ones that require being a person. We’ve split knowing off from the body, the stake, and the life for so long that we forgot they were ever the same thing.
Here they are, with a story for each.
1. Meaning-making from suffering
This is Viktor Frankl’s territory. Turning loss, illness, and failure into purpose. AI cannot suffer. It cannot find meaning in pain. This is the most uniquely human thing we can do.
The year my mother died, my business grew 81%.
For a long time, I could not hold those two facts in the same sentence. I took a sabbatical and sat with grief. I learned that grief is unexpressed love. And I formalized the Learning Culture Lotus, which became the spine of what we now sell. The work that mattered most came out of the year that hurt the most.
AI cannot lose a mother. It cannot find meaning in a life because it has no life.
2. Tacit knowing
Michael Polanyi called this “we know more than we can tell.” It runs in two directions.
- Inward: Antonio Damasio’s somatic markers, the gut decision your body reaches before your brain catches up.
- Outward: the sense that “I’m fine” isn’t fine, the heaviness in a pause, the unsaid thing under the said thing.
Both come from a body that has lived a life and stands to lose something. AI has neither. It cannot get a knot in its stomach, and it cannot feel the temperature change when someone walks into the room.
I think about this most with my six-year-old son.
A six-year-old cannot mask. When Leo walks into a room, his body tells you what is true before he has the words for it. The slumped shoulder. The eyes that won’t quite land. The half-second pause before “I’m fine” that says he isn’t. Adults learn to bury this. Small children haven’t yet. Reading them trains a part of you no curriculum teaches.
That is what Polanyi meant. We know more than we can tell because we live in bodies that pick up signals long before language catches up. AI reads the sentence. It cannot feel the room change when the kid walks in.
3. Moral courage under cost
Knowing what’s right when it costs you something, and doing it anyway. AI can recite every ethical framework ever written, but cannot care. It loses nothing by being wrong. Humans bleed for principles.
My best friend in high school is one of the most morally courageous people I know.
Back then, he was the kind of student you wouldn’t have noticed. Quiet. Head down, working on the yearbook.
His friend Camille submitted a quote: something like “the pot at the end of the rainbow isn’t quite what you expect it to be.” The teacher in charge flagged it as a hidden drug reference, “coming from that kind of girl.” My friend heard her say it.
He confronted her in front of the rest of the yearbook staff. Told her, loudly, that she did not get to make assumptions about a student she did not know. He was a quiet kid going against an adult with authority over his recommendations, his grades, his daily life. He spoke anyway.
The move was small, and it was everything. That is the part AI cannot do. It can argue every side of an injustice. It cannot stand in the room and pay for the side it picked.
4. Falling in love with a question
Darwin spent eight years on barnacles. Einstein chased light for a decade. Sustained obsession across years and seasons of life. AI doesn’t wake up at 3am still thinking about something. It has no unfinished business with the universe.
I have spent 20 years now on one question: how do adults actually learn?
200 days of in-person facilitation at KPMG before I ever taught anyone online. Each stage stacked one more skill. Facilitation. Curriculum. Sales. Video. Entrepreneurship. By the time I launched a $350K online course, I had been chasing the same question for 15 years.
Credibility comes from putting in the reps for years and building patterns you understand well enough to decode for others. AI can read every learning science paper ever written in an afternoon. It cannot stay with a question for 20 years.
5. Taste
Steve Jobs picking the typeface. The chef knowing when the sauce is done. Aesthetic judgment built from a singular life of noticing. AI averages across training data. Taste comes from specific, concrete details; from having been somewhere with someone at some time.
My 20 years of exploring the same question have built something I can’t quite name. A felt sense of whether a learning experience is actually transformational or just looks that way.
When my team brings me a workshop draft now, I don’t read the deck first. I drop into the participant’s seat and walk through the room with them. The flat moments show up in my body before they show up in my notes.
That’s taste. Years of accumulated noticing, compressed into a judgment that runs faster than analysis. AI can grade a deck against a rubric. It cannot sit in the participant’s chair and feel the energy drop.
Where does this take us?
AI is going to do a lot of work. It will write better than I can on most days. It will out-research me, out-summarize me, out-pattern-match me.
It will not love my children. It will not grieve my mother. It will not stand up in a classroom for someone it loves.
That is where the next ten years of human work lives. Not in racing the machine on what the machine is best at. In going deeper into the parts of us the machine cannot reach.
