Using AI More Doesn't Make People Less Afraid of It

Gallup's panel data cut against the hope that familiarity eases fear. In Christos Makridis's study of U.S. workers from 2023 to early 2026, those using AI daily or several times a week were more than twice as likely as less frequent users to fear their job would be eliminated within five years, in most waves. The gap peaked in 2024 at 6.7 percentage points. 1

The fear costs something. Tracking the same people, Gallup finds that when workers saw their job at risk, they scored 0.20 standard deviations lower on engagement and were 4.1 points more likely to be actively looking for another job. 1

Among frequent users, the fear is weaker where workers feel cared for and respected. Gallup's headline credits managers, but those items measure workplace climate, not what managers do. 1

Use does not guarantee better work either. A 2023 Harvard Business School experiment found a 2023 language model raised speed inside its capability and cut accuracy outside it. 6 The adoption job is the frame around the tool.

1. 1. Heavy AI users fear displacement more, not less. In most Gallup waves, workers using AI daily or several times a week reported acute displacement concern at more than double the rate of less frequent users, though only 3% to 4% of all workers call their job very likely to be eliminated. 1
2. 2. Fear is lower where the workplace feels caring and respectful, but Gallup measured climate, not managers. Among frequent users, the link between use and fear is 11.1 points smaller for workers who feel cared for and 6.8 points smaller for those who feel respected. 1
3. 3. AI helps inside its capability and hurts outside it. In a 2023 Harvard Business School experiment with 758 knowledge workers using a 2023 language model, those with AI completed 12.2% more tasks, 25.1% more quickly, inside the frontier, and were 19% less likely to produce a correct solution outside it. 6
4. 4. Manager backing goes with more value from AI, and most employees lack it. Gallup reported in November 2025 that only 28% of employees in organizations implementing AI strongly agreed their manager actively supports its use. Those who did were 6.5 times as likely to strongly agree the tools are useful. 2
5. 5. The most common uses are not the best-rated ones. Fifty-two percent of U.S. workers use AI, most often for writing and editing (51%) or search (49%), yet 77% of coding and automation users report a positive productivity effect. 5

Brief 018 asked what a role is for once AI absorbs its routine work, Brief 025 how junior staff learn judgment, and Brief 026 why workflows need redesign. This Brief stays with the employee who already has the tool, and with what a manager supplies that a login does not.

Most adoption plans treat exposure as the cure: buy the licenses, run the demos, and let use settle nerves and raise output. Gallup's panel says use alone does not settle the fear. The Harvard experiment says use alone does not guarantee better work. Both point back to how managers frame the job around the tool. Neither study watched managers at work, so the step to manager framing is this Brief's inference. 16

Frequent users fear displacement more, and a vague role may be why

Makridis pools nearly 30,000 worker observations from four Gallup Panel waves; about 30% of respondents appear in two or more. By the first quarter of 2026, roughly 19% of all workers said their job was somewhat or very likely to be eliminated by AI. The same fear tracks higher burnout and lower job satisfaction. 1

In Gallup's reading, the tool that speeds up a task also raises the question of whether the person doing it is becoming redundant. Eva Selenko and colleagues argue that whether AI threatens a worker's sense of identity depends on whether it complements tasks, replaces them, or creates new ones. Without clear guidance from the organization, Gallup concludes, frequent use only heightens uncertainty about job security. 17

Climate buffers the fear, but Gallup measured climate, not managers

Gallup titled its 9 September 2026 report "Using AI More Does Not Reassure Workers, Managers Do." The data say less. Its five workplace items cover the organization's care for wellbeing, respect, culture, feedback in the past week, and promises kept to customers. The two that mattered most, care and respect, describe climate. Neither records what a manager did. 1

Within those limits the finding holds. The buffer is concentrated among frequent users; for infrequent users, what Gallup calls strong and weak management makes little difference. Comparing each worker with themselves removes stable traits such as personality and baseline optimism. 1

Gallup's adoption surveys do ask about managers, but about backing the tool. Employees with active manager support were 8.8 times as likely to say AI gives them more chances to do what they do best every day (November 2025) and, in a February 2026 study of 23,717 employees, 9.3 times as likely to say AI has transformed how work gets done. In July 2026 their engagement rate was 48%, against 30% without that support and 31% nationally. 234

Backing goes with more use, and more use goes with more fear, so a manager who pushes use without explaining it could be adding to the fear. No source here measures what this Brief recommends: a manager explaining what the tool means for a person's role. 123

The task, not the amount of use, decides whether AI helps

In a 2023 experiment with Boston Consulting Group, Fabrizio Dell'Acqua and colleagues randomly assigned 758 knowledge workers to work with or without a 2023 language model. On 18 realistic consulting tasks inside what they call the jagged technological frontier, AI users completed 12.2% more tasks, 25.1% more quickly, with better quality. On a managerial task chosen to sit outside it, they were 19% less likely to produce a correct solution. 6

The lesson for managers is a boundary, not a dose: someone has to say where the tool is strong and where its output must be checked. The experiment shows the cost of crossing that line. It did not test whether guidance prevents it. 6

Vo Phuoc Tai's desk review of 48 studies, screened from 489 records, notes that AI "may reduce performance when employees over-rely on inaccurate outputs, use AI for unsuitable tasks, or lack sufficient AI literacy." A low-profile review that draws on the Dell'Acqua team's work, it adds little independent weight. 8

Gallup's July 2026 reports point the same way. Coding and automation, each used by only 16% of AI users, draw the strongest productivity ratings. Only 17% of frequent users in AI-adopting organizations give AI the top productivity rating, against 50% of those who also have a clear AI plan and manager support and are themselves engaged. Engagement may lift that figure on its own. 45

Autonoma forecast: Over the next 12 months, Gallup's panel will keep showing more displacement fear among frequent AI users than occasional ones, and active manager support will stay well below half. Organizations that add seats without preparing managers to set task boundaries will see anxiety persist among heavy users and productivity ratings stay flat. 1246

• Gallup's next panel waves show whether the frequent-user fear gap, 6.7 points at its 2024 peak, widens or closes. 1
• A study tests what managers say and do against displacement fear. 1
• Active manager support climbs from the 28% in Gallup's November 2025 report toward half. 2
• U.S. employee engagement moves off the 31% it has held since 2024. 4
• A test with current models shows whether the 2023 frontier result holds. 6
• Your engagement data, split by AI use, show whether the heaviest users are slipping.
• Your manager training covers what the tool changes in each role and where its output must be checked.

For CLOs and L&D: Move budget from tool tutorials to manager preparation: what the tool changes in a role, where its output needs checking, and what still counts in a review. 1

For CHROs: Treat displacement fear as a retention risk, and read falling engagement among heavy AI users as an early sign of it. 1

For business and function leaders: Tell each team where AI output can be used as is and where a person must check it. 6

For CFOs and boards: Seat licenses are not a productivity measure. Fund the plan and the manager support that go with Gallup's best productivity ratings. 4

For CIOs: Pair each deployment with a named manager and a use case that fits the team's workflow. Gallup ties frequent use to both. 3

Weight: Strong. The strongest objection is that the fear is accurate. Gallup says workers "are responding to a real concern": the more of an occupation's tasks generative AI can do, the more likely its workers are to fear displacement. A caring workplace may ease the feeling without changing the outcome. 1

Five of eight sources are Gallup's. The within-person estimates rest on the smaller group observed across waves and are less precise, and the surveys capture what workers expect, not jobs actually lost. The Harvard experiment used a 2023 language model, not current models in live work. 16

Most of that is fair, and this Brief does not claim managers can talk away a real risk. Gallup's own line is that better management "may not erase workers' fears" but "can substantially ease those fears." That is enough to act on, since the fear already costs engagement. 1

Picture the adoption dashboard: logins up, prompts up, the rollout marked green. In the same quarter, engagement among the heaviest users slips. Gallup suggests reading that slip as an early sign of AI anxiety.

Without a manager's word on how the tool makes them more valuable, Gallup warns, heavy users may start to feel expendable. No login count shows whether that conversation happened.

Decision rule: if an adoption plan counts logins but not manager conversations, hold the next wave of seats until managers can explain what the tool changes in each role.