AI as a technology for automatically discovering and exploiting valuable information coming from data is expected to be one of the key factors in the evolution of every productive sector, let it be manufacturing, services, utilities, and even agriculture.
In fact, the increased ability of monitoring the processes concretizing the above value chains, and thus of gathering fine grain readings on their activities, enables the development of an information layer paired with any physical activity. A thorough exploitation of processing in such a layer theoretically allows an unprecedented level of control and optimization of the corresponding physical activity, that goes beyond the sheer reactive and automation level to aim at the proactive and intelligence level.
Yet, to unleash such a potential, one needs data-to-information processing tools, such as those of AI and machine learning, that are refined to industrial standard, i.e., able to deliver constant, guaranteed, reliable, highest-level performance within feasible resource budget and sometimes strict implementation constraints.
Further to that, operating in high economical value environments, often entailing the interaction with human coworkers, puts an extra stress on the significance of errors, failures, and connected liability. This, along with the dependence of industrial-level AI development on massive capital investments, highlights a critical relationship with regulations, that must be addressed within a broad multi-disciplinary view.
Hence, “AI for industry” needs both applicative and methodological competences coming from multiple knowledge domains whose contribution will shape the productive sectors of the future developed economies.
“AI for industry” research areas can be summarized as follows: