Pirelli Annual Report 2022
Interviews with Pirelli’s employees
Polo R&D Director*
Artificial intelligence has revolutionised everyone’s lives, in everyday life as well as at the corporate level. How then has AI transformed your work within Pirelli’s teams?
From the start, we worked with the team having a clear objective: to understand the potential applications of artificial intelligence in the technical field of materials, of the product and the production plant. This has allowed us to study the evolution rather than the revolution that this technology can bring to the working world, understanding its potential but also its limitations. The advantages and speed that are gained by using data science models are indeed extremely significant with an immediate extension of know-how and the possibility of both engineering and shop-floor benefits. However, it is crucial to understand how to set up the development, structuring and engineering of data that are the basis for effective models suitable for machine learning and artificial intelligence.
Machine learning develops and uses algorithms to analyse structured data, learns from these and based on this learning, makes increasingly precise decisions in order to reduce the error rate while maximising efficiency. Is this really true, that the machine learns and man does not?
We both learn, indeed man learns faster precisely because of the help these technologies can offer in quantifying complex, multi-variable phenomena, where a traditional approach would be extremely time-consuming and difficult, if not impossible. It becomes much more problematic, however, when the data are small and one goes outside the linear range. Indeed, the human intellect has far superior capabilities in connecting information regarding different patterns and experiences: it consciously uses these technologies with a clear roadmap of goals to follow.
AI: machine = man: x. What is the unknown and/or added value that we can find in the work carried out and developed inside and outside the production plant?
This is a very useful analogy as it provides a simple key. The added value is the continuous evolution of the machine, which takes place infinitely faster than the evolution of traditional machinery. Instead, the unknown lies in the validation and focus to be placed on the models and the surrounding environment, such as aspects concerning collaboration between the business platform and the data science group. The risk to be avoided is going to an extreme in either direction that could be exemplified in a denial of the value of data on the one hand, and an almost magical expectation of numerical models on the other.
AI is making it possible to share more knowledge, to make know-how broader and more global. How much do you feel this in a company like Pirelli?
Undoubtedly, this is one of the enormous added values that these technologies allow. Not so much in the sharing of know-how, an aspect that is very much taken care of within Pirelli, but in the ability to do so with precise and appropriately updated quantifications. It becomes easier to document, measure and predict conditions that have already occurred elsewhere or at other times within the company. The perception of this is certainly strong and significant, particularly in a factory environment where there is a great deal of data from many process machines. Having the ability to work with structured models that also horizontally link the process chain constitutes an extremely significant source of know-how from which improvements and efficiencies can be designed in an orderly and quantitative manner.
Artificial intelligence and human beings are often talked about as antitheses, but what if we talked about integration, in Pirelli?
Pirelli represents excellence in integration and is therefore quick and effective in developing these tools. In a company where the core business is not AI/ML products, integration is the conditio sine qua non to flank core science and technology with data science. It is necessary to learn a second language, that of digital and data, across all business functions, so that it becomes a compliant and compatible approach with the basics of data science. Integration comes through knowledge and this also applies to the digital transformation we have been experiencing for a few years now.
* Polo Industriale Pirelli – Settimo Torinese – Italy