What Work Should Remain Human

The parts of work we shouldn’t rush to give away

I’ve had a few conversations recently with founders that keep sticking with me. Someone will mention using AI to help with hiring, or marketing, or planning, and they’ll say it almost in passing, like, “It just makes things easier.”

And they’re right. Things move faster, tasks are completed, and the day feels lighter.

When you ask what changed, the answer isn’t always about better outcomes or higher quality. It’s usually about relief of not having to revisit the same decision again. Not having to sit in that uncomfortable space where you’re not sure yet, but you know it matters, and no longer having to carry so many loose ends at once. We know how this feels as we’ve all experienced juggling too much.

That’s often the real starting point for automation. The fatigue. The quiet kind that comes from holding too many half-decisions at the same time. Meetings that end without real closure only to repeat the next week. Ideas that are close, but still need someone to stay with them a bit longer. The burden of being the person who keeps things moving even when clarity hasn’t fully arrived.

In that state, any AI tool that speeds things up feels like magic. A draft comes back cleaner than expected or a recommendation sounds reasonable enough to accept. Something finally moves forward instead of getting stuck in paralysis analysis. When nothing breaks, it reinforces the choice. From the outside, it looks like momentum.

But pay attention to how quickly that feeling becomes normal.

Ease has a way of turning into a default before we notice what it replaced.

Not all friction is a problem

There’s a habit, especially among people who build and operate things, to treat all friction as something to eliminate. If a process feels slow or a conversation feels repetitive, the instinct is to fix it. Streamline it. Hand it off. Don’t get me wrong, that instinct has created extraordinary systems. But it has also quietly removed some important ones.

However, to play devil’s advocate, perhaps not everything we do at work exists to produce output. Some parts of work exist to shape how decisions get made. They slow things down just enough for people to notice what they think, where they hesitate, and what they’re willing to stand behind. Those moments often feel inefficient on the surface, but they’re doing work beneath it.

Think about the discussions that take longer than expected because there isn’t enough data to be certain. Someone has to make a call anyway, and everyone in the room knows it. Those moments teach people how judgment actually works inside the organization. They show whose voices carry weight, how disagreement is handled, and what tradeoffs are acceptable.

When those moments are automated away, the decision still gets made. What disappears is the shared understanding of how it was made. Over time, that changes how people participate. Fewer questions get asked and fewer assumptions get challenged. Things move faster, but they also get thinner.

This isn’t about protecting struggle for its own sake. It’s about noticing what kind of work friction was doing before deciding it didn’t belong.

Not everything we do at work exists to produce outputs. Some parts of the work were never meant to be “efficient”.

Where being human still matters

There are a few places where this question becomes less abstract and more personal, especially if you’re the one ultimately responsible for the outcome.

Hiring is one of them. Not necessarily sourcing, screening, or scheduling. The part that should stay human is the decision itself. The part where you ask whether this person will raise the level of judgment in the room or quietly lower it. Whether they’ll make hard calls clearer or introduce more ambiguity over time. AI tools can surface patterns and risks, but they can’t feel the cost of being wrong six months later when the team dynamics have already shifted.

Creative direction is another. Not producing assets, but deciding what the work stands for. Founders often sense when something technically strong doesn’t feel aligned, even if it’s hard to articulate why. That feeling usually comes from having lived with the company long enough to know what compromises accumulate and which ones hollow things out. When that instinct gets overridden because the output looks “good enough,” the brand doesn’t break. It just becomes less distinct.

Strategy lives in this human sphere too, especially the parts that don’t show up in decks. Choosing what not to pursue, and deciding when to stay focused instead of chasing momentum. Knowing when a direction is exciting versus when it’s simply available. These are decisions shaped by learned lessons, consequences, and an understanding of what the organization can actually absorb. They don’t resolve cleanly because they’re not math problems. They’re strategic bets based on collective human experience.

In each of these moments, the work isn’t to produce an answer quickly. It’s to carry responsibility for the answer once it’s made.

Some decisions need a human because someone has to live with the outcome.

The cost of handing these off quietly

What makes these moments easy to delegate is that nothing breaks right away. The new hire may work out well enough. The campaign performs fine. The strategy holds for another quarter. From the outside, the system looks healthy.

What changes is more subtle. Founders start feeling slightly less connected to why decisions landed the way they did. Teams get clearer on execution, but less clear on intent. Over time, judgment becomes something that’s inferred rather than practiced. When hard moments arrive, there’s less shared muscle memory for how to think through them together. We’ve outsourced our cognition too much to the machines, and now judgment has been delegated.

This is how organizations become efficient without becoming wise. It doesn’t happen through neglect but gradually through ease of convenience.

When judgment stops being practiced, it doesn’t disappear. It just gets thinner.

Choosing what stays with us

Keeping certain decisions human doesn’t require rejecting AI or slowing everything down. It requires clarity about who owns the call. Who carries the weight of being wrong. Who decides when the data is helpful and when it’s insufficient.

Founders are already making these choices, whether they name them or not. Every time something feels “easier” to hand off, there’s a trade being made. Sometimes it’s the right one. Sometimes it’s a quiet erosion of judgment that only shows up later.

The work now isn’t to draw rigid lines. It’s to stay awake to where human presence still matters most. Where instinct, experience, and accountability are doing work that doesn’t announce itself as productive, but compounds over time.

What remains human isn’t about control. It’s about authorship. About knowing which parts of the system still require you to show up fully, not because you have to, but because the organization would be different if you didn’t.

That’s not a future problem. That choice is still available to us today, and that is: what parts of us will remain just human?

—

Delightfully human,

Yen