The honesty gap

The chatbot didn’t take that job. It took the vacancy.

Dorka Nagy-Józsa, Horizontal Thinking (2027)

Organizations don't run on lies. They run on softened truth, rounded toward safe one defensible adjustment at a time. Horizontal Thinking calls the space between what everyone saw and what the system was willing to write down the honesty gap - and explains why people now tell a machine what they won't tell their manager.

01

Softened truth

The meeting is called calibration. Twelve managers, one grid on the projector, a long November afternoon of deciding which human was which number this year. I sat in as an external observer - a polite word for witness.

We calibrate instruments so they tell the truth. We calibrate people so they don’t.

Four names in, a clear 4 - the strongest year of her career - meets a grid that frowns: too many high scores looks generous. Not a single person lies. The manager softens an adjective. A colleague recalls a wobble from February. Eleven small, defensible adjustments later, the 4 is a 3, and in January she'll be told, warmly, that it was “a really strong year.” She'll know. People always know.

That space is the honesty gap. Organizations don't run on lies. They run on softened truth: truth rounded toward safe at every interface. And rounding compounds. Take feedback, the ritual with honesty in its job title. Three requests on a January morning, three different ways to lie:

  • The colleague you haven't seen since COVID150 words of warm fog.
  • The climber who plays squash with somebody influentialTwo careful paragraphs about his “always responsive behavior” - technically true, if you overlook the nature of his responses.
  • Your closest work buddyLegendary meeting chaos you will never, ever put in writing.

Under the rituals sits the substrate, and the substrate is us. In an old survey of thousands of golfers (reported in Dan Ariely's The Honest Truth About Dishonesty), people said 23 percent of other golfers would nudge the ball into a better lie on an empty fairway. Themselves? 8 percent. Ask people about themselves and you get the oil painting; ask about others, anonymously, and you get something closer to the photograph.

02

Substantially more honest

Then AI walked into the building, and something odd showed up - first in my own consulting work, then everywhere I checked.

People are more honest with AI than they are in most meetings. Not just a little more honest. Substantially more.

People tell the machine they don't understand the strategy. At the town hall, the same people ask about parking. The reason isn't the technology's trustworthiness. It's the absence of social consequence. Say “I don’t understand” to a colleague and you haven't shared information - you've amended your record. The machine holds no opinion of you.

The lab measured this a decade before chatbots were fashionable. In a USC study - Lucas and colleagues, 2014 - people disclosed significantly more to a virtual interviewer the moment they believed no human was watching. Early workplace reports echo it: employees confess uncertainty and mistakes to chatbots before they tell their manager, sometimes instead.

When someone feels safer confessing uncertainty to a chatbot than to their own manager, it means the workplace hasn’t made honesty feel safe.

This is not a technology story. People want a space where they won't be evaluated while they are still thinking. That used to be what a good manager provided. Used to be. The role stood empty, and the chatbot moved in. It didn't take the manager's job. It took the vacancy.

03

The threshold

Every relationship you have runs on the shame threshold: the point up to which you're willing to show yourself. You talk until a small internal alarm whispers “okay, that’s far enough”, and there you start editing. Your therapist, your coach, your best friend, your spouse: each gets the truth up to a threshold, none at zero. They're human. They remember. They retell. They judge.

AI is the first entity you will ever talk to that has no threshold to trip. Not because it's smarter, but because you know it isn't human: no social circle, no family dinner where your story becomes an anecdote, no ego to bruise.

You will go deeper with it than with any human helper, ever.

That doesn't replace the therapist or the coach. It prepares the road to them. Once you've set the weight down somewhere, showing it to a human stops being unthinkable. The threshold doesn't vanish. It slides lower, because the secret has already survived being said once.

04

The other direction

The direction that caught me off guard runs the other way: leaders being more honest through the machine. In February a coach - one of the most emotionally attuned professionals I know, a few months into running his own agents - put his beer down and said:

“I love leading agents. I don’t have to fake my feelings.”

The relief he named wasn't speed or scale. It was dropping the mask. Every leader packages honesty for humans: you time the hard message, dose it, wrap it. You manage their feelings by managing yours. Ask any executive what wears them down most. Never the decisions. The choreography.

 Leading humansLeading agents
The hard messageTimed, dosed, wrappedNo sandwich, no dosing
DisappointedYou present “curious”You say it's wrong
Who is fakingYou, to protect the roomThey are - the warmth in the reply is pattern

That easy honesty, after decades of faking it, is addictive - by spring every agent-track leader I talked to (see leading AI agents) was circling the same confession. The open question is mixed teams: when half the standup is human, does the agent-honed bluntness leak onto the people?

05

Pretending to be sure

Most leaders I work with are exhausted, and not from the work.

They’re exhausted from pretending to be sure.

The higher you climb, the less you're allowed to say “I don’t know.” Decades of vulnerability workshops and laminated elevator posters didn't change that, because the incentive stayed put: admit doubt, watch your record get amended. Meanwhile the AI labs made calibration - the technical kind - a quality bar: a model that performs confidence it hasn't earned is a worse model. High time to learn from our robots. We can keep the posters.

Then the reckoning with my own profession. I built an AI that scores coaching conversations, and at the diagnostic level it beats a tired human assessor at 4 p.m. But when a client actually changes, the mechanism is almost never information. It's a human whose opinion matters, hearing the unsayable without flinching. Being heard by the machine costs nothing. Being heard by a person costs something. That isn't a skill. It's a stake.

The machine takes the diagnosis; the meaning, the safety and the change stay human.

06

What to do on Monday

  1. Say “I don’t know” once this week.Out loud, in a meeting where it costs you something. No softening clause. Count how many people say theirs within ten minutes - the going rate is two.
  2. Run your next decision through the no-consequence mirror.Before the meeting, tell the AI what you actually think, then ask it what you're not saying and why - on an approved tool, without colleagues' names or deal terms. The distance between that conversation and the one you have in the room is your personal honesty gap.
  3. Fix one feedback question.Stop asking people about themselves (“how engaged are you?”) and ask them about others (“what would your team say about this?”). In the golf survey the self-other gap was roughly threefold, and testing it on your team costs you one rewritten sentence.
  4. Find last cycle's drifted 4.Your team has one - a 4 that became a 3 for grid reasons - and you know exactly who. If ratings carry pay, HR hears it first. After that, the correction is a conversation, not a form.

The most honest conversation in your organization is already happening today. It just isn’t happening with you. Yet.

Next chapter: the mirror problem.

07

In the book

This page draws on Chapter 11 (The Honesty Gap) of Horizontal Thinking: Leadership after the end of the specialist era by Dorka Nagy-Józsa, PCC, published by Y2Y Ltd. on January 1, 2027. The book has fifteen chapters and two letters from the future, and every chapter ends with something you can actually do on Monday.

Dorka Nagy-Józsa is CEO of Y2Y Ltd., an ICF PCC coach and an international keynote speaker on AI and leadership.

09

Questions

What is the honesty gap?

The honesty gap is the space between what everyone in an organization sees and what the system is willing to write down. In her book Horizontal Thinking, Dorka Nagy-Józsa argues that organizations don't run on lies but on softened truth: truth rounded toward safe at every interface - in performance calibration, feedback and status reports - one defensible adjustment at a time. And rounding compounds.

Why are people more honest with AI than with their manager?

Not because the technology is more trustworthy, but because there is no social consequence. Every honest sentence said to a human gets filed in how they see you; the machine holds no opinion of you. A USC study by Lucas and colleagues (2014) found that people disclosed significantly more to a virtual interviewer the moment they believed no human was watching.

What does “The chatbot didn’t take that job. It took the vacancy.” mean?

People want a space where they won't be evaluated while they are still thinking. That used to be what a good manager provided. In many workplaces that role stands empty, so people take their uncertainty and mistakes to a chatbot instead. The chatbot didn't replace the manager; it filled the place the manager had left vacant.

Why do some leaders prefer leading AI agents to leading people?

With humans, leaders package their honesty: they time, dose and wrap the hard message, and manage other people's feelings by managing their own. With agents they can say “This is wrong. Redo it.” with no face to manage on the other side. As one coach put it: “I love leading agents. I don’t have to fake my feelings.” The book calls that easy honesty addictive, and asks what happens in mixed teams.

Where does the idea of the honesty gap come from?

From Chapter 11 (The Honesty Gap) of Horizontal Thinking: Leadership after the end of the specialist era by Dorka Nagy-Józsa, PCC, published by Y2Y Ltd. on January 1, 2027.

Read the whole argument.

Horizontal Thinking is published on January 1, 2027. The Kindle edition is open for pre-order now; the paperback goes on sale on Amazon the same day.