Because we need more women who deal with artificial intelligence

Until a few years ago, talking about artificial intelligence in the world of work caused a single reaction: will we all become unemployed? A reaction justified by the advent of algorithms capable of leaf through legal documents, o di analyze x-rays, with great accuracy. Over time, this disturbing narrative has changed radically: it has been explained to us that artificial intelligence will not replace the human being, but will complement him. For example: if an algorithm analyzes x-rays, the doctor has more time to treat the patient; if a software analyzes the legal documents, the lawyer can devote himself to the tasks with greater added value.

But which of these two visions is more realistic? “We have been wondering about this for some time and we wanted to investigate” – told us Fabio Ferrari, founder and president of Ammagamma, a Modena-based company that offers artificial intelligence solutions to optimize industrial processes (and not only) – “We went to talk to people who suddenly found themselves working together with some algorithm and two aspects are emerging. First of all, it is mainly the monotonous and boring operations, which therefore free up time for the more interesting ones; then, if even this automation will inevitably lead to some reduction in the necessary workforce at the same time new jobs are emerging”.

The benefits of relying on AI
It would also be wrong to think that only giants can benefit from solutions based on deep learning. Among the examples provided by Ammagamma, there is the system capable of predicting the market demand for a given asset with one 30-50% higher accuracy, a software used by De Matteis Agroalimentare to anticipate the price of wheat; moreover, a gas maintenance system, adopted in Italy by the Hera group, is able to foresee the greatest possible number of failures in the networks, making the planning of interventions more effective and timely.

Even in these cases, rather than taking away the job, the artificial intelligence automates a part of the tasks and modifies others: “And that’s why it is it is essential that there is training of employees through specialized academies, in order to prepare them for this new world and also to adapt their mindset. Rather than inserting data into computers, the human being’s task becomes that of analyzing the results provided by artificial intelligence ”.

For this Ammagamma, with the collaboration of a higher institute of Modena, launched Lucy last November, the first Italian experimental school of artificial intelligence for lower secondary schools. Among the objectives, there is one of fundamental importance, that is to reduce the homogeneity of the people who design the artificial intelligence algorithms: “In our company there are 50 and 50 men and women in all departments except that of mathematicians and IT, where women do not reach 15% – added Ferrari – E this is a problem we are trying to address even working directly in schools, because it is precisely there that, after having obtained excellent results, the abandonment of scientific subjects by women is triggered “.

Parity is important, even for machines
There is a reason of fundamental importance to ensure that among the figures who develop artificial intelligence algorithms there is equality between men and women (and more generally a good dose of multiculturalism), and that is to avoid the “Algorithmic discrimination” that we have seen at work in image recognition, in personnel selection software, in automated loan disbursement systems and more: “The problem is always in the selection of data – explained the founder of Ammagamma – Depending on the type of data, my mathematical model interprets the world with certain eyes, and if I give it partial data I will get a partial answer. “A partiality that becomes inevitable when the overwhelming Most data scientists are male, young, white, who inevitably see the world based on similar experiences and whose vision is reflected in the algorithms they have developed.

Could this limitation, the main one regarding artificial intelligence, also affect the tools for the optimization of industrial processes created by Ammagamma? “It is possible, but there is also another side of the coin: through algorithms it is possible to show objectively how in certain companies or organizations preferential lanes or other develop, allowing us to pay attention to behaviors that perhaps out of habit are not even noticed anymore – is Ferrari’s answer – But it is also to overcome all the potential obstacles that we wanted to leading philosopher of science such as Sabina Leonelli, professor at the University of Exeter ed exponent of the Alan Turing Institute. We need an expert who puts us in difficulty, posing uncomfortable questions that make us reflect on the less obvious aspects of our work ”.