Study shows that we believe more in machines than in ourselves

The research concludes the opposite of the usual comments on social networks, which generally express distrust and fear of artificial intelligences

Artificial intelligences are already ruling our destiny

A new study shows that we rely more on machines than humans to make decisions. In fact, he claims that confidence in the machine increases as the problem becomes more complex.

It is a conclusion that seems contrary to the usual comments on social networks, which generally express distrust and fear of what is popularly called “the algorithm”. But the study published in Scientific Reports – which has been funded by the US military as part of a project to examine collaboration between humans and machines – seems clear enough. According to study co-author Eric Bogert – a PhD candidate in the Department of Information Technology at the University of Georgia School of Business in Athens – “we seem to tend to rely more on algorithms as tasks get more done. difficult, and that confidence is stronger than the inclination to depend on the advice of other people”.

The study

To reach that conclusion, Bogert and his research colleagues, Aaron Schecter and Richard Watson, conducted three experiments on 1,500 people who were asked to count the number of people in a series of photographs. The photos increased in number of individuals as the experiments progressed.

The Force does not exist. Trust the machine, Luke

During the experiments, the subjects had the option of choosing between the suggestions of a group of 5,000 people – according to the researchers, we tend to place more value on the recommendations of others the greater the number of sources – or those of a matching algorithm. artificial intelligence trained with 5,000 images.

The researchers say that the counting experiment is ideal for drawing conclusions because artificial intelligence algorithms are now regularly applied to functions that require calculations of multiple numerical data. The study data is revealing, because the general impression is that humans do not trust the designs of artificial intelligence. According to Schecter, the study shows that this is not true. For example, one of the supposed problems with artificial intelligence is when it is used to grant credits or approve loans. It is a decision that basically depends on the numbers, so people think that it is a suitable job for the algorithm.

The biases of the algorithm

However, Schecter also recognizes the problem with the algorithm right now: many are influenced by their learning process. Learning that can lead to erroneous conclusions if all the factors necessary to make a judgment are not considered.

Human referee or artificial intelligence connected to hundreds of cameras?

 Following the conclusions of the study, the next logical step is to ask if people would prefer an algorithm to choose if you lose your job or receive a raise instead of your human boss. Or if an artificial intelligence – connected to dozens of cameras and laser measurement systems – would be a better option to direct a Barça-Madrid than a referee, two linesmen and a VAR. Taking it macro, would you prefer an artificial intelligence system, connected to hundreds of thousands of economic and social markers in real time, to regulate the economy and run the state and your city? It seems that the answer, as long as there is total transparency in the operation of that algorithm, is no longer yes or no, but why not try it?

Change political and economic ideologies for artificial intelligence

There are entire planets of evidence showing that, until now, humans have failed miserably in making decisions and continually evaluating the consequences of those decisions.

There are still incompetent bosses – and employees – everywhere, arbitrators who screw up at key moments, and politicians (all) unable to apply the scientific method and objective assessment to our problems, making decisions based solely on ideology and ulterior motives that they have nothing to do with the needs of the population. Our passions and inclinations, the lack of natural partiality of human beings, the reptilian instinct that governs much of our reactions, our fallibility, in short, have resulted in a succession of economic and political nightmares, decade after decade, century after century.

 There is no question that humanity has improved dramatically since we left the African plains. As much as it cannot be questioned that we could have advanced much more, orders of magnitude more, if we had not fallen victims, time and again, of the whims and interests of supposed economic, political and cultural leaders who have led us to war in war and from crisis to crisis.

The disastrous management of the pandemic in the vast majority of countries, to take the closest example, demonstrates this manifest inability of the human being to efficiently manage resources and make rational decisions. If the inability of economic and political elites – to provide not just stability and prosperity but a balanced economic and legal framework – has been demonstrated time and again by thousands of crises, it may be time to switch to something new, like systems artificial intelligence.

Do you think finally artificial intelligence will take over human intelligence?