Is artificial intelligence revolutionizing manufacturing? In this interview, Barbara Caputo, Professor at the Polytechnic University of Turin, Italy, guides us through the promise of enhanced process efficiency and innovative digital solutions, while also addressing the challenges companies may face when integrating AI into their workflows.
Barbara Caputo
- Full Professor at the Polytechnic University of Turin, Italy, where she directs the University’s Hub on Artificial Intelligence.
- Co-founder of the Laboratory for Learning and Intelligent Systems Society, ELLIS Fellow, and leader of the PhD program on AI and Industry 4.0 from 2021 to 2022.
- Advisor to the Rector for AI at the Polytechnic.
- Contributed as an expert to the drafting of the Italian National Strategy for AI.
- Included in the 2018 “Inspiring Fifty” and “100 Experts for Italy” by the Bracco Foundation.
- Independent director on the Boards of Ites Reale, Ites Reale España, Teoresi SpA and Infocamere.
- Co-founder and Chairwoman of FocoosAI, a spin-off of Polito.
- Winner of the 2023 PNI, Italy’s National Prize for Innovation.
Artificial Intelligence is a broad concept. How would you define it relating to the industry?
Today, in industry, applied AI is mainly the automation of digital data processing. Everything becomes data: our conversations, images and gestures are transformed into information by increasingly affordable sensors. Buying a camera used to be a major investment; now it is a given that every device has one. This data represents a source of infinite value because it does not wear out and can be reused indefinitely.
By automating the process of value extraction, AI allows us to analyse in real time a mass of data and information that our human, analogue capacity cannot handle with the same speed. From an industrial point of view, there are more and more applications that demonstrate the value of using and transforming data; that is why artificial intelligence is and will remain pivotal.
What are the main applications of AI in the manufacturing industry today?
All industries are already benefiting greatly from what we call generative AI. Generative AI simplifies the drafting of documents such as manuals and leaflets, allowing an initial version to be created on which a person can then verify information and add unique details.
In the manufacturing sector, I see three distinct applications:
analysis and tracking to quickly and accurately detect any imperfection or defect in the product
digital prototyping, or digital twins, to test products without the need to physically manufacture them, reducing stock and unsold inventory
predictive maintenance, through which production chains can be optimised, stepping in with small alterations to avoid costly downtime and reduce declines in productivity.
What challenges do companies face in integrating AI into their production processes?
On the one hand, there is an uncertainty about “who to turn to”. Companies feel that if they do not act quickly enough, they risk falling behind, standing by while competitors take the plunge and gain market share. Information on the topic is noisy and varied, making it difficult to know what to do.
Another challenge is varying levels of digitisation. In many industries – especially in small and medium-sized enterprises – digitisation took place at different times: in the 1980s, one technology was introduced, in the 1990s another, and so on. The result is outdated operating systems that often lack interoperability; for a company whose purpose is to sell products and ensure that revenues exceed costs, updating and standardising IT systems is a huge challenge. But without a solid digital foundation, generative AI and prototyping remain fragmentary. It is like going from a rough paper pattern to a seamstress who creates a tailor-made suit: AI, in its current state, needs that “little bit extra” to be tailored to the specific needs of each company. In times of expansion, and for businesses with adequate resources, this investment may be worth the effort; but for smaller companies that perhaps are not yet growing, it represents a considerable challenge.
What skills do companies need to develop to take advantage of AI?
If you want to become an AI company, you need a high-level technical team – people with skills in computer engineering or STEM subjects, capable of not only implementing, but creating AI solutions, because changes happen monthly in this field. If, on the other hand, a manufacturing company intends to introduce AI as part of its digitisation, it is essential to have in-house IT experts or rely on specialised consultants. Without a solid digital foundation, however, you risk paying expensive consultancy fees without concrete results.
How do you see the relationship between Artificial Intelligence and sustainability?
From an environmental point of view, I have to admit that currently AI is not really in the good books. The models running via data centres are extremely energy-intensive: in addition, cooling, which is often done using rather primitive liquid-based systems – mostly water – has a significant impact. If AI is to become a pervasive, win-win technology, efficiency of the systems must be improved so that the benefits truly outweigh the environmental costs.
What developments do you predict for AI in industry over the next 5-10 years?
The manufacturing sector is focusing on AI and better management of automation and robotics. The open-source model with which AI has been developed so far has made this technology very accessible. The result will be increased production capacity for companies that adopt it and, consequently, an increase in product and service offerings and associated cost savings.
However, the open-source approach, which has enabled the rapid growth of AI over the last 3-4 years, may also be threatened by ongoing geopolitical tensions. The shift to a new model will change the evolutionary line of AI and how technology is accessed. Talking about 5-10 years’ time for AI today is almost like asking what the world will be like in 100 years, because we are moving at an incredible rate! However, we are confident that AI will continue to radically transform industrial automation, making it smarter and more integrated; the challenge remains balancing innovation, cost and complexity.
What do you recommend for companies that want to harness the potential of AI?
My advice is simple: start at the bottom. If you are behind in digitisation, upgrade and standardise your systems before thinking about AI. Then, equip yourself with in-house experts who really know what it means to create and implement AI solutions. Approach the world of academia and consultants to evaluate and measure every investment, without giving in to the emotional pressure around you.
From an industrial point of view, there are more and more applications that demonstrate
the value of using and transforming data; that is why artificial intelligence is and will remain pivotal.
However, we are confident that AI will continue to radically transform industrial automation, making it smarter and more integrated; the challenge remains balancing innovation, cost and complexity.
CONTENT CREATED USING AI
This interview was initially extracted and processed with the support of AI, based on the topic covered. The result has been refined by an editor to ensure fidelity, consistency and that indispensable human touch. We chose this hybrid approach to experiment with the power of AI, combining it with human experience and awareness.
