UpTrajectory Review
Rebecca Hinds, writing in Harvard Business Review, argues that artificial intelligence shifts rather than eliminates labor. Small businesses adopt AI tools to save hours each week, yet those tools demand a new kind of human work: monitoring outputs, correcting errors, and managing the systems themselves. Hinds calls this 'hidden labor' and urges leaders to count it honestly. The piece arrives amid a wave of small-business AI adoption driven by affordable, easy-to-deploy tools. Vendors promise efficiency. Operators discover, often too late, that the tools require babysitting.
For a small-business operator, this matters because payroll is finite and time is scarcer still. A five-person shop cannot absorb hidden labor the way a corporation can. If an AI tool saves four hours of drafting but demands three hours of review, the net gain is one hour, not four. Operators who skip this math set budgets and staffing plans on false premises. They also risk burning out the one or two employees who become the de facto AI supervisors, a role no one applied for and no one trained them to do.
The genuinely new point here is the framing: oversight is not a temporary tax during AI onboarding but a permanent operating cost. We agree. Many operators treat review time as a setup-phase nuisance that fades once the tool 'learns' the business. Hinds pushes back. Models drift. Inputs change. Edge cases accumulate. The labor persists. We are skeptical only of the implication that better tools will eliminate the problem. They will not. They will shift the labor from correction to judgment, and judgment takes longer.
The second-order effects cut unevenly. Larger firms can hire dedicated AI managers; small firms fold the work into existing roles, which means the office manager or bookkeeper absorbs it. That has a cost: slower closes, missed invoices, or a neglected customer. It also has a morale cost. Employees who feel like machine janitors disengage. Downstream, this hidden labor distorts the ROI math that justifies AI spending. Operators who measure only hours saved will overinvest in tools and underinvest in the people who make them work.
What to watch next: whether vendors begin pricing in oversight support, and whether operators start tracking review time as a line item. A reader could act now. Pick one AI tool in use. For two weeks, log every minute spent checking, correcting, or reworking its output. Compare that total to the hours the tool claims to save. If the net is thin or negative, renegotiate the contract, retrain the team, or drop the tool. Hidden labor stops being hidden once someone counts it.
“AI can automate hours of work each week, but it also requires substantial human oversight.” — Harvard Business Review
Takeaway: Track the hours your team spends reviewing AI output; if review time eats the savings, the tool costs more than it returns.
Excerpt from the original — Harvard Business Review
<p>AI can automate hours of work each week, but it also requires substantial human oversight. Leaders need to account for that hidden labor.</p>