They Called Them Soft Skills. Turns Out They're the Ones AI Can't Touch.

The New York Times ran a piece this week with a headline that stopped me: "That Meeting You Hate May Keep A.I. From Stealing Your Job."

The argument is pretty simple. AI can speed up almost everything — the memo, the pitch, the deck, the code. What it can't do is the thing that actually makes work move. It can't read the room. It can't notice that someone across the table is nervous and needs reassurance before they'll say yes. It can't walk into a space full of competing interests and help everyone find their way forward.

As AI handles more of the production side of work, the irreplaceable thing turns out to be the person who can do exactly that.

I've been thinking about this for a long time, and it's strange to watch it finally show up so clearly in the data.

What We Rewarded

For a long time, we rewarded output. Speed. Volume. The person who could write the fastest memo, build the cleanest model, close the most tickets. We called that productivity, and we built entire performance systems around measuring it.

And then there were the other skills. Empathy. Listening. The ability to tell a story that lands. The capacity to make someone feel heard even when you're disagreeing with them. The talent for walking into a tense room and changing the temperature of it.

We called those soft skills. Which always struck me as an odd label, because the people who had them were doing some of the hardest work in any organization. Getting a skeptical stakeholder to trust you. Helping a team that's stuck find a way through. Convincing someone to take a risk they're afraid of. None of that is soft. It just didn't fit neatly into a spreadsheet.

What's Rising

The New York Times piece profiles a fractional executive who uses AI to do the work of an entire team. Strategy in hours instead of weeks. Websites in a month instead of six. He thought he might keep adding companies to his roster indefinitely.

Then he counted his meetings. Ten a week across two companies. Stakeholder check-ins, executive one-on-ones, project alignments, investor prep. Add a third company and that number jumps fifty percent. Add a fourth and he's in meetings almost literally all day.

The AI made him faster at everything it could touch. What it couldn't touch is what's filling his calendar.

A consulting firm mentioned in the piece used to encourage younger employees to specialize in a subject area or a technical skill. Now they're looking for generalists who take initiative and are excellent at building relationships. Another company said their hiring process, which once rewarded people who could code, now focuses on whether candidates can identify good ideas and persuade colleagues to back them.

This isn't a soft trend. It's a structural shift in what organizations actually need from people.

What This Means

I work at a law school. We're training people who will argue cases, counsel clients, and make decisions that affect people's lives. The most important thing we can give them isn't information — AI can deliver information faster than any of us. It's the capacity to think alongside someone, to understand what they actually need, to earn their trust under pressure.

That has always been true. What's different now is that the economic case for it is undeniable. Organizations that spent decades treating human skills as secondary are discovering that those skills are what's left when everything automatable gets automated.

I don't think AI is bad. I think it's clarifying. It's stripping away the tasks that could always have been done differently and leaving the ones that required a human all along. And the humans who will navigate this moment best aren't necessarily the most technical. They're the ones who know how to be in a room with other people and make something happen.

We spent a long time building organizations that undervalued that. We might finally be in a moment where we have to reckon with what that cost us, and what it could look like to build differently.