What can happen when, while you work, you are distracted by a young beautiful woman? What can happen when, because of your workload, you are making a mistake, such a mistake that you cannot do anymore what you were supposed to do?

As a youth, eating his lunch of bread and ewes’ milk cheese, saw a beautiful girl in the distance. Abandoning his meal in a nearby cave, he ran to meet her. When he returned a few months later, the mold had transformed his plain cheese into what is called today Roquefort.

Stephanie, who did most of the cooking, was overworked one day. She started to make a traditional apple pie but left the apples cooking in butter and sugar for too long. Smelling the burning, she tried to rescue the dish by putting the pastry base on top of the pan of apples, quickly finishing the cooking by putting the whole pan in the oven. After turning out the upside-down tart, she was surprised to find how much the hotel guests appreciated the dessert. Yes, it is Tarte Tatin, one of the most famous dessert.

We just wanted to begin with these two stories. Humans make errors. Sometimes for good, sometimes for bad, but they have such a freedom. But machines? Will they? What kind of results can we expect from those machines? Whatever they will do will be done following an exact procedure and will produce results as expected. This is what a robot is doing. No matter which calculator I use, 1+1 will always be equal to 2.

Those solutions are processing incoming data to extract those who are compatible with an expected pattern, collect the info, use the info… The whole system will wake up and produce what is needed as soon as the data triggering those events is available. All these solutions look at the world the same way and will produce similar outcomes. A little scary.

And what will happen if the data is not available, if there is a new item which is not known from the machine, which was not available to the machine learning solution who build the knowledge? Is this solution able to do something? Will this solution be able to recognize the item? How will this solution decide, recommend, take actions? They all need to learn first! This is again not how humans are learning. Question: when you go to a shop selling furniture and you see the latest chairs who just came out on the market with this new design, you, as a human, do you have problems to recognize this new chair as being a chair? Do you have any problems to use it? For sure no. You don’t have to learn again what a chair is, even if you never saw that chair before! Can the machine do the same knowing how knowledge is “build” by those machine learning solutions?

Predictive analytics is a probabilistic approach based on analyzing data from the past giving you a result with a chance that what is forecasted will be realized. And what this artificial solution is doing here is not what humans would do in such a situation. No, humans don’t do this. Neurons don’t do this. Something biological is not working like this.

In the past, people used their hands to manufacture goods. They were replaced by robots. Office work is performed by humans because you need to read and get an information on a piece of paper and do whatever you must. These are now replaced. The technology has eyes and is acting faster… You will find many examples. All these examples are based on recognition capabilities the machine can now do on behalf of humans. Interfaces are changing also, going from keyboard and display to voice recognition and speak. Do you really think that when you ask one of these smart assistants you have at home to turn on the light, put music on, program and alarm or tell you about the weather forecast that this does require any kind of intelligence? Are those machines able to make errors? And if one day they could, will they learn from that? Will they be able to understand that the error they did produced a different output as the one expected, but this output is good, even better and from then that they learn from this mistake and are happy to reproduce it? We are far from this as of today, most probably because we designed those systems with components who are not able to process information like this. Machines maybe have eyes but they don’t see the world like we see it.