If AI multiplies productivity, where does the saved time go?

A short essay published on the personal blog mlsu.io in late May 2026 asks a question that follows directly from the promises surrounding artificial intelligence: if the technology really is about to multiply white-collar productivity, where does the extra time go? The post, titled "Can we have the day off?", takes the industry's own claims at face value and then asks for a share of the gain in hours rather than in extra output.

The argument

The author starts from the familiar claim that the world's white-collar workforce, and much of the US workforce with it, is on the cusp of a productivity revolution, one that will change how people work, learn, socialise and interact with the world. If that is true, the reasoning goes, the same amount of output that used to take a full week should be achievable by midday on Monday. From there the request is modest: work Monday through Thursday, and take Friday off.

An "AI workers' day"

The post even offers a mechanism. Thursday would be spent writing good prompts, and on Friday the agents would churn through them while their human colleagues stay away. Under that arrangement, the author argues, an employer hardly loses a day. The idea is extended upward as well: boards of directors and executives could spend the free Friday on a full 18 holes of golf, because nobody needs to be in the office — the AI agents are there instead.

The note about childcare

The essay closes with an aside addressed to Elon Musk. The author writes that they are trying to increase the fertility rate, notes that childcare for three small children costs 6,000 dollars a month in California, and asks why five days in the office should be non-negotiable: why not four?

Why it is worth reading

This is opinion, not research: the post cites no data and proposes no policy. Its value is that it turns the productivity narrative into a simple test. A technology sold as a multiplier of output invites the question of who keeps the surplus, and a free day is a clearer indicator of that than any benchmark score. If the gains are real, the argument runs, they should show up somewhere other than a longer to-do list.