Imagine you’re standing on the production floor, watching a robotic arm that has been working flawlessly for months. Suddenly it stops mid-cycle and retreats to its base position. The part it was operating falls to the floor.
– Stupid machine,” chuckles the exasperated operator.
But the truth is different: this machine is not stupid at all. It just thinks in a completely different way than you do.
Today, after years of working in the field of production automation, I can confidently say that machines do exactly what we require of them – not a gram more, not a gram less. And we treat them as if they have intuition. And this is where the problem begins.
The answer is simple – they don’t think. They act. The control algorithm program is their only instruction for survival. Every machine – from a simple palletizer to a smart home robot – executes its cycle of operation in a loop:
This cycle repeats thousands of times a minute.
It can be compared to our condition in the morning. The alarm clock rings. I turn it off. I check how I’m doing. I determine what day of the week it is, start the “Monday morning” program and go to the bathroom, starting the coffee maker on the way. I automatically follow the steps – diagnosis, stimulus, scheme and action.
Join the newsletter and gain access to 40% of the book
“15 Steps to Buying an IT System“.
As you can see, the machine doesn’t know the context. It’s like a web page showing a 404 error. The server doesn’t know that you’re frustrated or that you’re in a hurry. It only knows that “the page was not found,” so it runs a script: “show default message”. Yes, the machine in unforeseen situations runs a so-called default program – a kind of fuse. A bit like an application on a phone that simply shuts down in case of an error. It’s not stupidity – it’s programmed caution.
Take a look at the cleaning robot. It carefully maps the living room, learns where all the furniture is, develops an efficient cleaning plan. One day you rearrange the furniture. The poor robot enters the room, expecting the coffee table to be in the corner, but nothing of that – it’s now in the middle of the living room. He discovers that his map is wrong. What does he do?
Option 1: He is trying to adjust and figure out the new arrangement.
Option 2: Performs the default “I’m confused” protocol – returns to the charging station and waits for human intervention
Most such machines choose the second option. Not because these machines are stupid, but because they are programmed for safety.
When a production robot stops unexpectedly, it usually does exactly what it’s supposed to do. Why? For a simple reason: we can’t program every possible scenario. The robot can recognize unexpected temperature changes, parts that don’t meet specifications or voltage fluctuations. Therefore, in case of unexpected events or data, it returns to its “base program.”
It’s like a ground fault circuit breaker in your house – it doesn’t care that you’re just cooking dinner for the in-laws. It simply determines that something is wrong and cuts off the electricity.
– But what about artificial intelligence? – you may ask. – Aren’t AI machines supposed to be smarter?
Yes and no.
AI gives machines more scenarios. But even the smartest AI system in production has limitations:
Even a smart vacuum cleaner won’t escape from a burning room – at most it will register the increased temperature (if it has a sensor) and stop.
An AI-driven quality control system can identify 99.7% of defects correctly, but when it encounters those 0.3% edge cases, it still comes back to: “When in doubt, refer for human review.”
The next time a machine does something that seems illogical, remember: it’s probably trying to protect itself, your product or your people. This “stupid” behavior may be the smartest thing it could have done, given its limited understanding of the world.
Machines don’t have bad days, they don’t get distracted or take shortcuts. They simply follow their software – even when that software leads to seemingly irrational behavior.
Honestly? In a world where a malfunctioning machine can cause injury or huge production losses, I prefer a machine that stops too often to one that keeps running when it shouldn’t.
Watch the next episode of Digitalize.co.uk if you want to better understand how machines work, how AI works and why digital transformation is not just a matter of code.
