Why do we give better feedback to GPT than to people?

With people, we dance. In advertising, it’s practically ballet. Nobody wants to rattle a client, bruise a teammate, or break the fragile peace of a meeting. So we hide behind clichés. 

I’ll know it when I see it. Keep pushing. It’s not quite there. Everyone nods, everyone smiles, and everyone leaves with no idea what actually needs fixing. By the time the team has decoded “keep pushing,” three days are gone and the work is worse.

Now think about how we treat GPT. Brutal honesty. No hesitation. No fear of hurt feelings. Just: This is boring. Try again. Make it funnier. Less cringe. Fact-check it. Research it. And we’ve all seen how much better it gets from the first prompt to the next. The machine gets the truth in seconds. Humans get riddles.

And it’s not just me noticing this. A Gartner survey found that 87% of employees believe algorithms could give fairer feedback than their managers. Think about that. People would rather hear the truth from a bot than from their boss.

I’m not suggesting we bark at colleagues the way we bark at chatbots. Manners matter. Candidness with generosity is the balance. I fall into the same trap as everyone else. It’s easier to sidestep, to smooth the edges, to avoid the awkward moment of saying what needs to be said. 

The paradox is that candor isn’t cruelty. Most of the time, people are more grateful than not. What’s kinder: telling someone your presentation skills need work or letting them spend months wondering why they never get invited to the big meeting? What’s more useful: your research needs depth or you just need more time? Clarity is generosity. Ambiguity is cruelty with a smile.

Pixar made that tool into a ritual. Their Braintrust meetings are famous because they institutionalize brutal, kind feedback without the usual power games. Directors bring works-in-progress, the room gives plain, direct notes, and no one votes; the director listens and decides. 

The magic is twofold: first, feedback comes from peers who know the craft and aren’t protecting egos; second, because the director retains final say, ownership stays intact and teams can iterate fast. The result is not humiliation but acceleration. 

The work improves quickly, people learn faster, and everyone walks away with clearer choices rather than watered-down consensus. It’s honest, instructive, and deeply rewarding because you’re not being shredded for vanity’s sake; you’re being sharpened so the work can actually sing.

The beauty of every ad campaign is that it’s forever. Once it’s out in the world, there’s no recall, no software update, no do-over. The only chance to make it better is in the making. After that, it’s over and on to the next one. 

That’s why feedback, even the kind that stings, is an act of generosity. I don’t know a creative who wouldn’t rather take a tough note that makes the work sharper than a polite smile that leaves it weaker.

AI can draft, edit, and simulate. But it can’t do the one very human thing that moves work forward: look someone in the eye and tell them the uncomfortable truth that actually makes them better.

So maybe stop saving the good feedback for machines. Don’t train GPT to get sharper while your team leaves the room confused. Because eventually, sure, the machines may take our jobs. But if that happens, let’s at least go down knowing we gave each other the good stuff too.

And if you’ve got feedback on this piece, don’t hold back. I even asked GPT to review it, but told it not to hurt my feelings. It responded with a string of compliments. Proof that machines may take our jobs one day, but they still can’t take a joke.

Vini Dalvi is the chief creative officer for Publicis Canada. This article first appeared in Campaign Canada.