AI gives us a lot. Speed. Access. Drafts. Summaries. Answers. Automation. Scale.
But what is it taking away?
The risk is not that AI makes work easier. The risk is that it makes it easier to skip the part of work that forms us.
Start with the friction. The slow part of thinking a problem through yourself. The hour spent rewriting one paragraph because you couldn’t quite say what you meant.
That friction wasn’t waste.
It was the work.
The work was where you figured out what you actually believed. You’d start writing one thing and discover, three drafts in, that you believed something else. You’d draft a proposal and realize halfway through that the deal didn’t make sense. You’d write an angry email and notice, before you sent it, that you were angry at the wrong person.
Skip the friction and you skip the discovery.
You get an output.
You don’t get the version of yourself that producing it used to build.
That’s the first thing AI takes.
The reps.
The second is judgment. Not whether something is correct. That’s the easy version. Whether it matters. Whether it’s worth saying at all.
When everything is cheap to produce, everything gets produced.
We’re already drowning in content nobody asked for. Polished proposals from people who didn’t think about the project. Decks for meetings that should’ve been a phone call.
The effort required to produce something used to be a brake.
Now there is almost no brake.
Then there’s memory.
The real kind, where you carry the deal in your head because you had to.
You knew the names. You knew the history. You remembered what a client said in March and connected it to what they said in November, and the pattern wouldn’t have been visible to anyone who hadn’t been holding both pieces all along.
When you outsource memory, you don’t just lose recall.
You lose the connections only one mind, holding everything at once, can make.
The last thing, and maybe the biggest, is trust signals.
For most of business history, polish was a proxy for effort. A well-written proposal meant someone took time. A clean deck meant the work was real. An organized response meant the operator was serious.
Now polish is free.
Everyone’s output looks competent, which means polish tells you nothing.
The signal got cheap.
And the people who actually did the work (the ones who studied the problem, lost sleep on the details, got it right because they cared) look the same as everyone else.
That’s a real problem.
Trust runs on signals.
When the signals stop signaling, trust gets harder to build and harder to extend.
Here’s the part nobody wants to say out loud.
AI doesn’t replace the people who do the work.
It just makes them harder to find.
So what do you do with that?
You go the other direction.
You put your name on things. You take positions an AI wouldn’t. You show your reasoning, including the parts you’re not sure about. You return the call. You sit with a problem until you actually understand it.
You build a reputation for being the human who showed up, not the one who automated.
That’s not a Luddite stance.
Use the tools.
Use all of them.
But know what they cost.
In the built world, this matters even more. A polished proposal doesn’t tell you who will protect the client. A clean email doesn’t tell you who will own the mistake. A good-looking profile doesn’t tell you who will show up when the job gets sideways.
The more artificial polish fills the market, the more valuable real reputation becomes.
AI helps us move faster.
But speed isn’t judgment.
Output isn’t understanding.
Polish isn’t trust.
The fastest path to a deliverable is rarely the path that turns you into someone worth trusting when it matters.
Randall Yang Co-Founder & CEO, BuildLogIQ