Shisa kanko: what a Japanese train driver can teach us about artificial intelligence
Published on 24 October 2026 at 05:32 GMT
By Joan Ubide
On Japanese railway platforms, something happens that, to foreign eyes, seems almost theatrical. An employee in an immaculate uniform extends an arm, points a finger at a signal, a clock or a door, and announces out loud what they see. The driver does the same from the cab. It is neither an empty ritual nor a cultural eccentricity. It has a name: shisa kanko, "pointing and calling".
I must admit that when I first came across this technique, its simplicity struck me as almost baffling. And that is precisely why it fascinated me. The method originated in Japan in the early 20th century, and the pointing gesture was added a few decades later. What convinced me it was not mere folklore was the data: according to laboratory work at Japan's Railway Technical Research Institute, the technique reduces human error by nearly 85% in simple, repetitive tasks, falling from 2.38 errors per 100 actions to 0.38 when pointing and calling are added.
A finger and a voice. Nothing more. No machine, no algorithm. And yet an improvement that many technological systems would envy.
The fragility and greatness of human attention
Shisa kanko starts from an uncomfortable truth: people make mistakes. Not out of incompetence, but because our attention wears thin, becomes automatic and, in routine moments, wanders off without asking permission. We have all left home unsure whether we locked the door, even though we have done it a thousand times.
But the technique also starts from another, far more hopeful truth: people are capable of designing brilliant solutions to their own limitations. Someone observed the problem, understood how our minds work and created a method that makes sight, gesture, voice and hearing work together. That is not a machine correcting a human. It is a human understanding and improving themselves.
It is not an isolated case. In 2009, a study published in The New England Journal of Medicine and promoted by the World Health Organization showed that a simple surgical safety checklist, applied in eight hospitals in eight cities around the world, reduced major complications from 11% to 7% and mortality from 1.5% to 0.8%. Once again, the same pattern: a humble tool, born of human judgement, saving lives.
A new colleague arrives
And into this scene steps artificial intelligence. I work every day integrating these tools into businesses, and I have no hesitation in recognising it: their capability is extraordinary. They can read in seconds what would take me days, detect invisible patterns in mountains of data, draft, translate, summarise, compare and suggest. They do not tire at five in the afternoon or lose focus on a Friday.
In a way, artificial intelligence might seem like the logical evolution of shisa kanko: a system that keeps watch for us, that never gets distracted, that checks what we would overlook.
And this is where I want to pause, because I believe we are at a crossroads with two very different paths.
The two poles of our relationship with technology
At one extreme lies blind delegation. It is the temptation to hand over to the machine not only the work, but also our attention. To accept what it says without checking, to copy and paste without reading, to let the system decide while we nod along. It is comfortable, and that is why it is dangerous. Artificial intelligence models can be wrong with absolute confidence; they can invent a figure, a quote or a reference and present it with the same conviction as a truth. Anyone who has stopped pointing and naming is no longer in a position to notice.
At the other extreme lies outright rejection. It is the mistrust that leads people to ignore these tools out of fear, pride or nostalgia. Those who choose this path give up genuine help and, over time, fall behind those who have learnt to use it.
Neither extreme convinces me. The first turns us into spectators of our own work. The second condemns us to work with one hand tied behind our back.
Pointing at the machine
What I propose is to apply the Japanese train driver's lesson to artificial intelligence. It is not about no longer looking because we now have an assistant looking for us. It is about pointing at and naming what the machine produces as well.
When artificial intelligence gives me an analysis, I point at it: where does this figure come from? When it suggests a decision, I name it out loud: does it make sense for this company, for this person, for this moment? When it offers me an answer that looks perfect, I stop for a second, just as the platform attendant does before signalling the train to depart.
That second of conscious attention is what keeps us in control. The machine brings speed, memory and computing power. The person brings judgement, context, ethics and responsibility. Neither is surplus to requirements.
Being human is not what is left over, it is what guides
It is often said that artificial intelligence will leave us only the tasks it cannot do, as if being human were a leftover. I see it the other way round. The human element is what gives direction. It was a person who understood that pointing a finger reduced errors. It was people who decided that a checklist deserved a place in operating theatres. And it will be people who decide what we use artificial intelligence for, within what limits and in whose service.
The technique has already crossed borders: in Canada, Metrolinx has incorporated pointing and calling into GO Transit's daily operations as a way to enhance safety. I like to think that this same humility, the humility of recognising that we can fail and putting a remedy in place, is what we need to bring to our relationship with artificial intelligence.
The future is pointed out with a finger
We are approaching a new era, and I do not believe its success will depend on how powerful machines become. It will depend on how attentive we remain. The best artificial intelligence in the hands of someone distracted is a risk. A reasonable artificial intelligence in the hands of someone attentive, critical and responsible is a formidable tool.
That is why, whenever I work with these technologies, I try to remember the image of the driver in the cab: arm extended, eyes fixed, voice clear. They do not distrust the train; they simply never stop checking. That, to me, is the attitude we need: neither fear nor blind faith, but shared attention.
Artificial intelligence can help us see further. But it is still up to us to point the way.
Written by a human author, edited with AI assistance.
Sources and more information:
Pointing and calling – Wikipedia
https://en.wikipedia.org/wiki/Pointing_and_callingOrigin - of the method in Japan in the early 1900s; 1994 Railway Technical Research Institute study showing an 85% error reduction; adoption by Metrolinx in GO Transit's operations.
How the ritual of pointing and calling, known as shisa kanko, embeds us in the world – Christopher Roosen
https://www.christopherroosen.com/blog/2020/4/20/how-the-ritual-of-pointing-and-calling-shisa-kanko-embeds-us-in-the-worldError - rate falling from 2.38 to 0.38 per 100 actions with pointing and calling.
On point – AOPA Flight Training Magazine
https://www.aopa.org/news-and-media/all-news/2021/january/flight-training-magazine/on-pointJapanese - names of the technique and confirmation of the 85% error reduction.
A surgical safety checklist to reduce morbidity and mortality in a global population – The New England Journal of Medicine
https://www.nejm.org/doi/full/10.1056/NEJMsa0810119Haynes - AB, Weiser TG, Berry WR, et al. N Engl J Med. 2009;360(5):491-9. DOI: 10.1056/NEJMsa0810119. Checklist implemented in eight hospitals in eight cities; deaths fell from 1.5% to 0.8% and inpatient complications from 11.0% to 7.0%.
PSNet (Agency for Healthcare Research and Quality)
https://psnet.ahrq.gov/issue/surgical-safety-checklist-reduce-morbidity-and-mortality-global-populationSummary - of the study and confirmation of its full reference and World Health Organization basis.
Before skin incision – QxMD
https://calculate-web-next.k8s.qxmd.com/calculate/calculator_228/before-skin-incisionConfirmation - of the study's mortality and complication figures.
The Harvard Gazette
https://news.harvard.edu/gazette/story/2009/01/surgical-safety-checklist-drops-deaths-and-complications-by-more-than-one-third/Total - sample of 7,688 patients and the study's authorship.
Youtube credits
