AI and the humanization of machines

Can machines think? This is one of the questions raised in 1950, by the famous English mathematician Alan Turing, to whom the deciphering of the German army codes and the great successes of British counterintelligence during World War II are due. Projecting the specifications set by Turing at the time to the present day, we would say that a machine has intelligence if it is able to process and recognize images, recognize and synthesize speech, process natural language and be able to synthesize knowledge. In a complementary direction, the American Norbert Wiener, a prominent Professor of cybernetics and systems science, defined the intelligence of a machine as the ability to acquire and process information to achieve predetermined goals.

In the decades that followed, machine intelligence was primarily associated with the ability to simulate human thought. For example, in 1997, the top Russian chess player Garry Kasparov lost a game of chess to IBM's Deep Blue computer. In 2014, a program that pretended to be a 13-year-old convinced a panel of judges that he was human by 30% after just a few minutes of conversation. In 2016, Google's AlphaGo computer defeated champion Lee Sedol in a Go match. In 2017, Deep Mind's AlphaZero computer, which is based on neural networks, uses reinforcement learning, and self-taught, defeated the Stockfish computer, which used traditional databases of chess moves from real matches, in a game of chess.

Overall, what is artificial intelligence (AI)? There is no doubt that in the era of the 4^th industrial revolution, this is a key question that concerns the entire public opinion. To answer simply, AI is the scientific field that has as its ultimate goal the construction of machines that possess intelligence. AI and machine learning, which is a subset of it, focus on the processing of big data using algorithms (i.e. problem-solving recipes, expressed in a programming language), to produce predictions (mainly classifications) exploited for decision support. AI tools, as artificial beings, have the so-called supercomputing capacity.

However, a series of critical questions arise. Can AI lead to the humanization of machines? In our view, machines can only be trained in our preferences and reflexes. However, if they create their own space of thought, then we will not have access to it, just as we do not have access to the world of our pet. It will simply be an intelligence of a different type. Is our democracy threatened by the algocracy of AI? The answer is affirmative, if one takes into account the practices of totalitarian regimes that monitor every trace of movement, both physical and digital, of their citizens. Finally, can machines impose themselves on the human being through AI? The answer is again affirmative, if humans rely on the decisions proposed to them, as a result of using AI, without conscious comprehension. In this case, AI will clearly limit human free will.

However, it must be made clear that machines are not consistent with the concept of anthropomorphism, do not embody the creativity of the human brain, and lack the potential of human inspiration. Simply put, they lack intellectual adaptation. Furthermore, there is still an open debate in science about how the human brain works and how human intelligence is produced. Consequently, the questions of how to infuse emotional intelligence in computers and how to mechanically reproduce the function of the synapses of the human brain remain under scientific investigation.

Another open question has to do with the anthropological implications of AI and how it affects human reflexes. AI has transformed everything. From the information market to the methodology of human socialization, to the extent that a new type of human, another human status has emerged, that of the complementary human to technology. And this means that the traditional social contract must be rewritten, taking into account the concept of techno-ethics, as a consequence of non-human factors that man himself has created. Simply put, man is no longer alone and coexists with machines. AI has brought a real revolution in image processing, in understanding speech as a series of voice instructions (see Apple's Siri and Amazon's Alexa), in high-quality machine translation of foreign languages ​​(see Google Translate), in the creation of text from large language models (LLM), etc.

At the same time, however, there is also the negative connotation of the underlying view, such as deep fake image and voice elements, cyberattacks, technological crime and in general the vulnerability of the AI ​​ecosystem, which intensifies as we transfer our activity to the internet. Undoubtedly, when an algorithm is a black box, it cannot be reliable. For example, can an algorithm that has been trained to drive a car in Texas also drive it in Luxembourg? And if self-driving causes an accident, who will be blamed? The user, the vehicle manufacturer or the engineer? And if the accident involves a grandfather and grandson, who will the algorithm choose to save? The above are just some of the aspects of techno-ethics, when legal issues arise in a criminal ground. For this reason, cooperation protocols between countries are necessary, in the event that an algorithm that has been developed in one country should be applied in others. An additional example from the legal field is the use of AI algorithms in the legal system (e.g. the US), where it must be decided whether a defendant should be detained or not. Does AI help in legal jurisdictional judgment? Or will the algorithm be biased, e.g. in favor of gender and nationality, and the machine itself reproduces social stereotypes?

So how do we position ourselves regarding the future and what is coming? A few decades ago, at the beginning of the internet, we believed that the internet would bring democratization, equality, unlimited access to knowledge and information, freedom of expression, etc. However, along the way, technology was allowed to enter our everyday lives unchecked, without controls and a framework of transparency. The economies of the US, Europe and Asia are arguing about their excellence in the field of AI. The argument is well-established that China has an advantage as, due to its political regime, but also its population, it collects extremely large amounts of data from its citizens, which can make a difference in the training (and therefore the effectiveness) of AI algorithms. In our opinion, the West will win in the battle of AI. Because it is a world compatible with the ideals of democracy and liberalism. A few decades ago, during World War II, for related reasons, the West allies won the battle against the totalitarian threat, in the race to build nuclear weapons.