Who Gets the Hours We Save?

The crew of the Nostromo should have been on their way home.

In Alien, a ship can haul a refinery between stars while its crew sleeps. The technology is extraordinary. The employment arrangements are familiar: arguments about bonuses, orders from the company, and a contract that makes refusing an assignment expensive. Parker and Brett would recognize a bad day at work in any century.

I usually borrow my future from Star Trek. For October, I've been thinking about the people who got the spaceship without getting the freedom.

We should pay attention to that possibility. Machines can absorb an astonishing amount of work while the humans around them remain thoroughly on the clock.

The hours are going somewhere

On October 29, workplace analytics company ActivTrak reported that AI users in its sample had workday spans of 8.4 to 9.1 hours, compared with 7.6 for non-users. The analysis covered 9,140 knowledge workers across 467 enterprises, using activity data collected from April through September.

Those are workday spans, which can include gaps, rather than a count of uninterrupted labor. This is observational research from a company selling workplace analytics. Different roles, workloads and employers could explain some of the difference. It doesn't establish that using AI added ninety minutes to anyone's day.

It does make the automatic promise of an earlier finish look rather premature.

Anthropic's October 16 study of its own workplace adds another complication. By August, employee-directed agents were logging about six times the hours attributed to their human handlers. That's agent runtime compared with a forty-hour-per-person estimate, not six human jobs replaced. The company had also tripled in size over the preceding year.

Alongside increased output, the researchers found that the median employee spent 20% more time in meetings than a year earlier. Growth and changes in the mix of roles help explain that figure; even technical individual contributors saw a 16% increase. The report describes people building systems to organize and supervise the growing volume of machine work.

Last month I wrote about the responsibility that remains when agents do the production. October gives us a reason to ask what that responsibility is going to cost in human time, and who decides how much is enough.

More output is one possible destination

Suppose a team gets a weekly reporting job down from eight hours to four, including the checking and corrections. There are several sensible uses for the other four hours. Serve another customer. Improve something that has been neglected. Teach a beginner. Let the person go home.

The last option usually requires the most explanation.

We have made an odd convention out of this. A successful automation can be celebrated as saving time, followed immediately by a new target that consumes every minute of it. By the next performance review, the higher output is the baseline. The employee has helped create a gain without ever getting a say in how to use it.

There are good reasons to produce more. A struggling business may need the extra sales to survive. A clinic that can spend less time on paperwork may be able to see patients who have been waiting. I want those gains. I also want us to be honest about the choice being made.

An increase in output doesn't tell us whether anyone gained control over their afternoon. If we care about a post-labor life, that deserves its own measure.

A shorter week has to include the paycheck

For somebody paid by the hour, fewer paid shifts can mean a worse life. For somebody already juggling two jobs, the immediate wish may be more dependable income. Calling either situation liberation because the person spends less time working would be insulting.

The useful goal is more choice over time, with enough security to exercise it. That makes pay, benefits, staffing and workload part of the discussion from the start.

There is evidence worth building on. A 2025 study in Nature Human Behaviour examined six-month four-day-week trials with no reduction in pay, covering 2,896 employees across 141 organizations. Burnout, job satisfaction, mental health and physical health improved, a pattern the researchers did not observe in twelve comparison companies.

That was a study of reorganized work and shorter hours. It doesn't show that an AI subscription will finance Friday off. It gives us a reason to test a concrete arrangement rather than assume that a full workweek is the only respectable destination for every productivity gain.

The people doing the work should help choose the arrangement. Some may want shorter days. Others may prefer a protected day for learning, a pay increase, or a predictable schedule that lets them collect their children. An executive's preferred version of freedom may be a poor fit for everyone else.

Actionable Steps

Before starting the next automation project, agree on what happens if it succeeds. Pick one workflow, establish the outcome and quality it has to maintain, and include all the human time needed to keep it running.

Then reserve a share of any demonstrated saving for something the workers helped choose. Try it for ninety days. Keep pay intact in a shorter-hours trial, and check whether the work has quietly moved into evenings or onto colleagues. Ask about actual working hours and control over schedules, as well as output. A Friday off with Thursday night spent catching up is a disappointing experiment.

Include the people outside the successful pilot. If the senior team gets flexibility while beginners lose their way into the profession, the organization still has a transition problem. Some of the saving can pay for supervised learning and a first chance to do useful work. July's argument about apprenticeships belongs in this budget too.

I have to apply the same standard to my own venture experiment. A fleet of agents can give me an almost unlimited supply of things to direct, inspect and improve. I could fill every waking hour and have plenty left over. That possibility is getting easier to demonstrate. Making room for music, friends, and an afternoon when nothing needs my approval is the part I actually want to earn.

Parker wanted to discuss the bonus situation before the trouble started. He had the right instinct. We should discuss what we get from the machinery while we still have a say in the arrangement.

I'd like the ship to get us home sooner.

LLAP. 🖖