Is there anyone left who hasn’t heard of Artificial Intelligence (AI)? They say it’s the next big thing after smartphone. And the biggest companies in the world like Google, Facebook or Microsoft are already, mainly, AI companies. It could be said that Tesla, to cite an example, is not just a car manufacturer, but a manufacturer of robots that make other four-wheeled robots that drive themselves. However, although many know this new intelligence, few understand how smart is this or if it really is.
And the word is the most powerful weapon. Describing a set of mathematical and statistical patterns with the term “intelligence” not only allowed consumers and investors to understand complex ideas, but also made them more attractive. From its origins, AI has been dressed in hype, controversy, and apocalyptic preacher messages at subway stops. Already in the eighties it was said that soon a computer would surpass human intelligence. Which would allow us to dedicate ourselves solely and exclusively to leisure and creativity.
The constant disappointments to meet such expectations paralyzed the investment of the Universities and companies. Until phones became “smart” thanks to the iPhone. Then came televisions, clocks, thermostats… They even wanted us to believe that a fridge with a screen was a smart fridge: Was it going to fill itself with food? None of these devices was smarter than a calculator or a watch. They are nothing more than small computers without a keyboard or mouse programmed to perform very basic and specific tasks. But the marketing departments had already achieved their goal, and everything became smart. An Internet-connected TV was no longer an Internet-connected TV, but a Smart TV.
But some companies began to adopt machine learning algorithms, substantially improving some of their software when it was eating the world. Machines could now be programmed to learn, not just do. And problems of enormous complexity for human understanding became trivial mathematical problems for a machine. It can perform more calculations in a minute than any human being in a lifetime.
Such has been the progress of these algorithms and the strength that marketing departments have had on the conception of public opinion about them that today we find people who believe that they are not only intelligent, but that some machines are also aware. We’ve come to this crazy point. And this is just the beginning.
Acting smart is not being smart
Researchers and companies have led us to believe that their algorithms “understand” what they are told or that they “think” or “analyze.” It is a humanization, a simplification and a lie. What we call AI is a far cry from what was imagined in the 1950s or 1960s. No system is yet programmed to reason. AI uses massive amounts of data to turn any complex task into a prediction problem based on human work itself.
This has made tasks that were unapproachable for us affordable for the machine. But, even if it programs itself, it is unable to see beyond. We can train a machine with a million images of dogs so that it learns to identify them, but it will not know how to abstract the idea of a dog. That is, you will know how to locate dogs without knowing what a dog is. And so with any task, you can be an expert on dogs and not know what a cat is.
AI is math. And the main object of mathematics is to simplify the world so that it can be understood by our minds. But it is only part of our intelligence, and the only thing that has been shown is that what is very complicated for us is easy for a machine, and vice versa.
The machines were programmed to make us believe that they were intelligent or creative like us. We think that Dalle-2 is a robot that paints and creates, but in reality it only selects and arranges images already created by man into a new one. But he’s very good at it. The same goes for chatbots, which are programmed to respond to us. They mathematically select the most appropriate text from all the text available on the Internet. That is why he is able to defend that he is conscious, because someone wrote it sometime somewhere. The Google engineer believed that the AI was conscious because, in fact, it is programmed to do so. With the same success, you can expose an allegation that defends just the opposite, which is not conscious. An AI is just as aware or intelligent as a calculator.
AI will be as smart as we think it is
Tricking a person into believing a program is intelligent is not the same as creating an intelligent program. But obviously we are the ones who judge what is intelligent and what is not. And what we believe does not usually coincide with what is true. Is a dog self-aware in the way that we humans are? No, but every owner believes that the dog understands and loves him unconditionally as people understand and love each other. Although we know that scientifically that is not possible. We humanize are attitudes, as we will do with machines if, under our understanding, they act like a person.
It is not necessary to elaborate any hypothesis to imagine this possibility, because it already happens. There are numerous cases of people who fall in love on the Internet with people who do not really exist or whose signs of love have some lucrative objective behind. Or that they think that the political activist on Twitter is really a person, when in reality he is a robot. There are people who believe in ghosts, that the date they were born determines their personality or that the Earth is flat. Man is always prone to believe anything that reinforces his beliefs or consoles his ailments.
It is worrying to think about the incidence that these robots will have in the future, clearly human. If we now fall for crude online scams and the propaganda spread by governments, the media, political associations or companies, it is worth thinking about what could happen when the information is spread and personalized by robots with the highest qualifications. Instead of a simple advertisement, a robot will be able to impersonate a person. This will talk to you with the aim of becoming your friend, and then he will recommend you invest in a certain cryptocurrency or resort to some pseudoscience to treat an illness. It all depends on who you work for.
The question is not whether AI is smart or not. The question is whether we are smart enough to coexist with machines that behave like us. Programmed to pretend they hear us. Designed to make us believe that they understand us. Made to help us or to deceive us.