Afraid in the town hall, certain in the budget

Meta description: AI fear and hype coexist when leaders and employees operate with fragments of information and no shared view of change. More on his work: his consulting practice.

In the same quarter, sometimes the same week, I have watched a company produce two incompatible stories about AI. The town hall hears caution: jobs are safe, adoption will be careful, nothing will change without people at the center. The budget cycle hears certainty: aggressive targets, committed spend, an investor narrative about transformation. Leadership experiences this as a communications challenge — how to land both messages. I have come to see it as something else. The two stories are not a contradiction to be resolved. They are a structure to be read.

A company can speak with two voices about AI in the same week. In an investor presentation, management describes a coming transformation. At an employee meeting, staff ask whether their jobs will disappear. Both groups see only part of the situation. The disagreement can appear to be a communications problem, but it may start with a shared absence: neither side can explain precisely what is changing.

Where do the two stories live in the same week?

The stories are geographically stable. Fear lives where staff gather: town halls, team meetings, HR communications. Confidence lives where capital gathers: board packs, investor calls, strategy offsites. Each audience gets the story calibrated for it, and the calibration is so routine that nobody experiences it as a choice. What I watch for is the crossover moment — when someone from the town-hall audience reads the investor deck, or vice versa. The damage happens there, and it is usually attributed to a leak rather than to the structure.

Why does a single narrative make the split worse?

Different groups live with different consequences. Executives hear promises of scale, lower costs, and improved productivity. Employees see individual tasks changing and worry that their skills will no longer be needed. Neither perspective, on its own, describes the entire operating system.

A management team might announce that an AI tool will replace hours of repetitive work without saying who remains accountable for reviewing the output. An employee might reasonably infer that the company has decided to remove a role. The manager may believe the statement is encouraging; the employee may hear a threat. The resulting disagreement is not solved by polishing the original slogan.

The false explanation is messaging: the company needs one narrative, consistently told. I have watched communications teams attempt exactly that, and the unified narrative makes things worse, because it forces a story that is true for neither audience. The caution was not spin — parts of the organization have real reasons to be careful. The confidence was not spin either — parts of the market story are real. Flattening both into one message produces a statement nobody believes, and now the company has a credibility problem on top of a structural one.

Who is served by each story?

Every version of this pattern I have seen has beneficiaries. The fear story serves leaders who need stability during a difficult transition; it keeps attrition calm and unions quiet. The hype story serves leaders who need a growth narrative for the market; it supports valuation and recruitment. Each story is locally rational. That is what makes the split durable — it is not an error people fell into, it is an arrangement that works, right up until the moment the two audiences compare notes.

What pattern keeps both stories stable?

The stability mechanism is separation. As long as the audiences do not overlap, each story can be maintained in good faith by people who believe it. I have noticed the companies work to maintain the separation — internal comms and external comms are different teams, with different approvals, and nobody owns the space between them. The split survives because no role includes reconciling it. It is an org-chart gap wearing a communications costume.

The response I favor begins with making the actual conditions and boundaries visible. People need to know what the work is, how it will change, where human expertise remains necessary, and what evidence will determine whether the new process is helping. That information is more useful than either extravagant promises or vague reassurance.

A shared operating language can help groups compare what they believe is happening with what the process actually shows. That does not guarantee agreement, especially where jobs or pay are at stake. It does make disagreement more specific and therefore more useful.

One of the strongest barriers to effective adoption is the belief that mistakes are unacceptable. I see a different pattern in learning. A child may fall repeatedly before walking with confidence. Organizations should distinguish between an error discovered during supervised learning and an uncontrolled error released into production. The former can be valuable evidence; the latter can cause real harm.

What question does not force a false unity?

The instinct in these companies is to align the messages. I have sat through that exercise and watched it produce a third story, vaguer than either original, that fools nobody. There is a different question available — one that examines both stories as information rather than as communications — and in my experience it does not come from the communications function. It comes from whoever is willing to treat the contradiction as a finding.

The central question for an executive team is not how to make the whole company repeat a single optimistic narrative. It is whether employees and leaders possess enough shared, verifiable information to disagree about AI without talking past one another.

PR excerpt (80–120 words, Miklos’s own words): A company can speak with two voices about AI in the same week. In an investor presentation, management describes a coming transformation; at an employee meeting, staff ask whether their jobs will disappear. Both groups see only part of the situation. Optimism and fear become two subjective readings of incomplete information. The response is to make the actual conditions and boundaries visible — what the work is, how it will change, where human expertise remains necessary — which is more useful than either extravagant promises or vague reassurance.

Miklos Roth, author of Signal Over Noise