We give AI better feedback than we give our colleagues
Maybe one of the more useful things AI can teach us about working with humans has nothing to do with AI. It's about feedback.
Recently I came across a question that stuck with me:
Do we treat AI better than our colleagues? And what is AI teaching us about our relationships with actual humans?
My immediate reaction was:
Feedback.
Think about how quickly we have learned to give feedback to AI.
"That's too long."
"You missed the point."
"Keep the first paragraph. Rewrite the rest."
"No. That doesn't sound like me."
"Better. Now make it more concrete."
We don't spend twenty minutes wondering whether ChatGPT will take it personally.
We don't sandwich the criticism between two compliments.
We don't save it for the annual performance review.
We say what isn't working, explain what we need instead and give it another go.
And interestingly, most of us have also learned something else very quickly:
The better the feedback, the better the result.
"That's bad" doesn't help much.
"This is too generic. Keep the argument, lose the jargon and give me one concrete example" does.
AI needs context. It needs precision. It needs to know what good looks like.
Guess what?
So do people.
Humans are not machines. Obviously.
Before someone gets very excited about that comparison: no, I am not suggesting we should start managing people like ChatGPT prompts.
Humans have feelings. History. Ambition. Bad days. Great days. Things happening at home you know nothing about. Empathy matters. Timing matters. How you say something matters.
But somewhere along the way, many organizations have confused being human with avoiding clarity.
We soften.
We hint.
We postpone.
We tell someone they are "doing great" for eleven months and then surprise them with a performance rating that suggests otherwise.
Or we say nothing because we like them.
Which brings me to something I wrote years ago and still believe.
Your company is not your family. Good.
I apparently thought this controversial enough back then to introduce it with "here it goes."
I don't anymore.
At work, we are there to collaborate, solve problems and drive business results. We should absolutely do that with people we respect and trust. Ideally, we even like quite a few of them.
But getting along cannot be the goal in itself.
And "we're a family here" is not a substitute for trust.
Family is, at least in theory, unconditional.
Employment isn't.
Companies change strategy. Roles disappear. Leaders change. Performance matters. People leave. Organizations restructure.
Someone you genuinely like can still be wrong.
Someone you regularly disagree with can be extraordinarily valuable.
None of that is a failure of the relationship.
And disagreement isn't disloyalty.
The people who challenge you might be the people you need most
I've worked with people I trusted enormously and disagreed with regularly.
Those two things aren't opposites.
In fact, some of the people who challenged me most were exactly the people I wanted in the room when things got difficult.
I wrote years ago that the best teams aren't those that always agree. Different perspectives, complementary skills and constructive disagreement are what move things forward. I still believe that.
The best teams can say:
I don't think this will work.
You missed something.
I disagree.
Try it this way.
And nobody has to spend the next three days repairing the relationship.
That requires trust.
It also requires respect.
And enough confidence in the relationship to know that challenging an idea is not the same thing as challenging the person.
Perhaps that's one thing AI can actually remind us of.
Feedback isn't the opposite of respect. Done well, it's evidence of it.
You bothered to explain what isn't working.
You gave me enough context to understand why.
You were precise about what needs to change.
And you gave me another shot at it.
That sounds like a pretty decent working relationship to me.
Care. A lot.
None of this means companies should become cold, transactional places where everyone clocks in, delivers an output and disappears.
Quite the opposite.
Care about people.
Know what they are good at.
Give them room to grow.
Notice when they are struggling.
Celebrate when they succeed.
Fight for the people you believe in.
Help when things go wrong.
And build relationships that survive the org chart.
Some of my closest friendships and most important support networks came through work. My original take on this was that building a network across countries and organizational hierarchies gives you people you can actually call when you need them.
Those relationships matter enormously.
But they became personal because we built them.
The company did not make us family by putting it on a culture slide.
Trust beats family
There was another slightly cheeky line in my original take on this:
You don't choose your family. You are rather stuck with them.
Colleagues, you can at least partially choose.
So choose wisely.
I still like that.
Surround yourself with people who will tell you when you're wrong.
Tell them when something isn't working.
Give context.
Be precise.
Listen when they push back.
And don't confuse disagreement with disloyalty.
Maybe AI is teaching us something about being more human at work after all.
Not because we should treat people like machines.
But because we've become remarkably comfortable telling a machine exactly what we need while still dancing around the same conversation with the person sitting next to us.
Care enough to give people useful feedback.
Your company doesn't need to be your family for that.
It needs trust.