TODAY IN NUMBERS
TECHNOLOGY · SATIRE · SELECTION BIAS

Japan's Government AI Saved Staff 110 Minutes a Week, Including the Third Who Say They Hardly Ever Open It

A Digital Agency survey of 6,439 ministry employees found most saving an hour a week or more. It also found 31.2 per cent rarely using the platform, or never using it at all.

Employees at Japan's government ministries and agencies have cut their working hours by an average of 110 minutes a week through the state's generative artificial intelligence platform, according to a Digital Agency survey published on Friday.

The online survey was conducted in August and September among workers at the ministries and agencies, and collected valid responses from 6,439 people, the Japan Times reported.

More than half of respondents said they had reduced their working hours by at least an hour a week. In detail, 35.1% said they had cut one to three hours, 10.9% three to five hours, 5.2% five to ten hours and 3.2% ten hours or more. A further 34.6% said their hours had remained largely the same.

Asked what the platform was useful for, many said they were able to "quickly create a rough draft", "find necessary information from vast volumes of documents" or "catch up with technical knowledge faster".

Use of it is uneven. Of those who responded, 10.4% use the platform almost every day, 11% three or four times a week and 26.4% once or twice a week. A total of 31.2% said they rarely use it or have never used it.

The platform, Gennai, was developed and is operated by the Digital Agency and is available to about 180,000 government employees.

It now includes an AI agent function capable of autonomously performing two or more tasks simultaneously.

The headline figure is an average, and it is worth asking across whom it has been averaged. Of the 6,439 people surveyed, 31.2% rarely open Gennai or have never opened it, and a further 34.6% report no change in their hours at all. The 110 minutes are shared out across all of them.

This makes the result more impressive than it first appears, not less. The platform is saving time for civil servants who do not use it, which is the most efficient form of time saving available to any software, because it asks nothing of the user and takes none of their afternoon.

Our analysts have followed the logic to the end of the dataset. If the weekly saving holds across the full population of roughly 180,000 employees, and if the non-users continue not to use it, then a measurable share of the Japanese state's productivity gain is now being generated by officials who have not opened the application and have therefore spent no time on it whatsoever.

The Department of Unnecessary Research notes that the agent function can perform two or more tasks at once, while an unopened platform performs an unlimited number, at the same moment, for nothing.

There is one limit to the method, and it is a matter of arithmetic rather than policy. The saving is reported per week, and the survey ran across two months. Nobody has yet asked the Digital Agency what Gennai would save if nobody used it at all.

WEEKLY HOURS SAVED, BY SHARE OF STAFFQUESTIONABLE ANALYTICS
351-3 hours
113-5 hours
55-10 hours
310+ hours
31Rarely or never use it
Real percentages from the Digital Agency survey, rounded for display: 35.1%, 10.9%, 5.2% and 3.2% of respondents saved those amounts each week, and 31.2% said they rarely use or have never used the platform.
THE BORING FACT

A Digital Agency survey of 6,439 Japanese ministry and agency workers found an average reduction in working hours of 110 minutes a week through the Gennai generative AI platform, while 31.2% of respondents said they rarely use it or have never used it.

Source: The Japan Times
OUR PROFESSIONAL CONCLUSION

Because 31.2 per cent of respondents rarely or never open Gennai and another 34.6 per cent report no change, part of the average 110-minute weekly saving is being produced by officials who have spent no time on the platform at all.

REALITY CHECK · SELECTION BIAS

Unfortunately, there's a small problem.

Averaging a survey across every respondent, users and non-users alike, makes the figure more conservative rather than less; it is the standard way to report such results. The joke pretends a shared average means every group contributed to it.

THE BS DETECTOR

Where did reality end?

Tap the paragraph where this story stopped being true. Then see where everyone else bailed.

 

READER CREDULITY METER

Did you believe this story before the twist?

Be the first reader to answer.

YOU WRITE THE CONCLUSION

Think you can end this story better than our analysts?

One deadpan conclusion, 280 characters at most. Start from the facts above and take them one step too far. Our editor picks the best one each morning, and it runs in the next edition.

280 left
No real people's private lives, no tragedy, nothing that accuses anyone of anything. Our editor reads every entry and prints one; the rest are never published.