Will artificial intelligence be the end of us? News about its development now arrives every month. Fear grows in many, while others almost hail the advent of this new Saviour.
As CEO of the translation services agency Studio Moretto Group, I shall explain here what is really going on.
Our business as translators is among the first to be hit by the wave of artificial intelligence in the form of machine translation. We have had sufficient time to study how machine translation software works and its effects on humans and their work, both positive and negative.
Artificial intelligence is an innovation, as were machines during the Industrial Revolution, the Internet, mobile phones and facial recognition. The promise behind each of these innovations is the possibility of doing certain activities in less time and more cost-effectively: in other words, a life that is easier, safer and more efficient, enabling people finally to have more time to devote to the things they love.
Now, the question is not whether these promises have been fulfilled or not, although we all continue to work too hard, there is never enough money and we live in a world of conflict, inequality and poor values, which puts us and our children at risk. The point is, rather, to understand what makes us blindly believe all these glittering promises each time, without asking ourselves who they come from and the reasons behind those who promote them.
We are like a child in the park who believes anything and goes off into the woods because someone promises them that they will find some sweets. We, the grown-ups, also believe the promises of the market and welcome innovations with open arms, only to discover years later that they were not actually the solution to our ills. This is what has happened in linguistics: it is now clear that machine translation software does not understand what it is translating because, technically, it has no consciousness and that, rather, it consists of systems that process and re-propose to the user things that have already been written in the past, identifying in their memories sentences whose contexts are more or less similar to those to be translated. In order to function, artificial intelligence must be trained, i.e. it must be loaded with millions of previous texts to which it can refer, thus creating a sort of reference memory. This has generated a lot of controversy: what is the quality of the texts loaded into the memory? Who do they really belong to? Are they tendentious? Who programs the algorithms to manage them? What is the relationship with the Deep Web and the Dark Web?
The fact that machine translators are trained means that they use syntactically fluent forms even though, in reality, they understand nothing of what they write and, therefore, cannot guarantee or certify the result. In short, in machine translations, we cannot know how many errors there will be, nor where they will be, we only know that there may be what are known as ‘hallucinations’ in the text, regardless of the fluency of the human-like language.
Yet, machine translation is not to be discarded a priori: it is an absolutely effective solution to be able to read large quantities of texts otherwise untranslatable by humans, identify the parts that are relevant to us and entrust them to a conscious human translator to correct and certify. It can also usefully provide professional translators with a draft translation to work on, which can save time, although this is still not possible for many languages and text types because the machine, unthinking and trained with sometimes inaccurate and tendentious data, makes so many mistakes that it ends up being uneconomical.
Instead of using it consciously in a targeted manner, translations by artificial intelligence are used indiscriminately, causing misunderstandings and adding to the mistranslated texts in circulation, which, in turn, will be loaded into automated systems, thus producing other mistranslated texts in the future. In this chaos of recent years, the prices of translation services have plummeted due to a market dynamic that has sought to place humans on the same level as machines, so much so that many young people have decided not to pursue a career in languages, believing it to be without prospects. The marketing of these systems has glorified the advantages for the user and they have even been offered for free, with the purpose of disseminating them as quickly as possible. This colossal operation of selling below cost price will sooner or later necessarily be followed by a sales phase, when these services will be perceived as indispensable and there will be a scarce availability of human translators due to people abandoning the profession.
What we have noted for linguistics also looks to be the natural evolution for other jobs in which artificial intelligence is already being proposed as an amazingly effective solution: the creation of graphics, written content, music, technical designs, legal opinions, accounting services, medical diagnoses, etc. The idea is to make those from outside the profession believe that the non-thinking machine is equal or superior to the professional, that is, the person who understands what they do and who has always based the value of their work on the notion of competence. The marketing mechanism that leads humans to prefer a machine to another human is as simple as it is terrible: the service is offered for free and is user-friendly.
Let us now return to the original questions: who made this hyperbolic promise and for what purpose? Why do we launch technologies on the market by enticing consumers to use them far beyond their rightful potential? Who promises a better life and, in reality, makes people believe that professional skills are not required, causing them to abandon human jobs? We could end up having only machines to do these jobs and with the general acceptance of their poor results, thanks to the diminishing human perception of quality and beauty. The continuous injection of low quality and low culture seems aimed at lowering the commonly accepted standard in a new economic, social and labour system heavily based on artificial intelligence. Certainly, other professions will be created, but let us not forget that it is this progress that has also generated the job of home delivery rider today.
In a world that has advocated globalisation and rapid interconnection, the promises are many and the risks are directly proportional, if not exponential. We must all make efforts to understand what we are offered, without falling into hasty extreme positions on either side. Unlike artificial intelligence, we are intelligent and our consciousness of what we are is the weapon of salvation against a design that threatens to direct human thinking towards a murky society.