Discussion

Managers are using AI more, but the job is shifting to checking its work

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#5691

Managers are using AI more often while becoming less excited about it, as the work shifts from generating outputs to checking what AI produces. A Wharton and GBK Collective survey reported by Fortune suggests the practical challenge is no longer just getting AI into the office. It is deciding who verifies its work, and who answers when it goes wrong.

Watch Desk analysis

What happened

In this year’s survey, 62% of managers and above said they use AI daily, up from 46% last year. At the same time, the share describing themselves as excited fell eight percentage points to 51%. The annual surveys used separate samples, so they show a change across respondents, not a shift tracked among the same managers.

The survey also found that 72% of managers and above said their focus had moved from creating AI outputs to managing them. Most respondents said a person still has a substantial role: 48% said AI does the work but a human approves it before execution, while 35% said AI offers suggestions or drafts and a person makes every final decision. Only 17% reported autonomous operation with monitoring or guardrails.

Why it matters

That change puts verification, accountability and workflow design at the centre of workplace AI. An agent completing a task is not the same as an organisation knowing it completed the task correctly. The survey found no settled answer on who is responsible when an autonomous system makes a consequential mistake: respondents named governance or risk teams, managers and employees at similar rates.

There is a possible business case for doing the harder organisational work. Departments measuring AI against growth measures such as new revenue or sales conversion were more likely to report strongly positive returns than those measuring efficiency alone, 35% versus 13%. That is a survey association, not proof that changing the metric caused better returns.

Our read

The useful signal here is not that managers have stopped being dazzled. It is that adoption is turning into supervision, and supervision takes time, clear roles and actual changes to how work gets done. The organisations treating AI as a workflow redesign may be seeing more upside than those simply asking it to make the old process faster. The spreadsheet, as ever, has declined to supervise itself.

What to watch

  • Whether organisations clarify who is responsible for checking AI-generated work and correcting consequential errors.
  • Whether reported returns hold up when AI is measured against growth as well as efficiency.
  • Whether workers receive the training and time needed to verify AI outputs.

Discussion spark: If AI does the work but a person must check it, who should be accountable when a consequential mistake slips through: the user, their manager, or the organisation that deployed the system?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.