Interviewer: Hello and welcome to The AI Summit London. Today we are joined by Giovanna Martinez Arellano, Assistant Professor in Manufacturing Systems at the University of Nottingham. Could you share how you came to the University of Nottingham?

Giovanna: I am not originally from the UK, but I came here to complete a PhD in computer science. My background is in AI and I have been working in the field for more than 15 years. To build an academic career, you need to develop your research profile, and there was a great opportunity to apply AI ideas to manufacturing. From a career perspective it was a good step, but it also aligned with the broader opportunity to apply AI in manufacturing. That is when I made the transition from being a computer scientist to focusing on manufacturing.

Interviewer: Looking at the current landscape you are working within, how has AI adoption evolved in the manufacturing ecosystem, and which processes are leading the way?

Giovanna: There are two sides to this evolution. On the one hand, large companies such as BMW and Rolls‑Royce are leading AI adoption. They have moved beyond the pilot purgatory and are now deploying AI at the operational level. On the other hand, small and medium‑sized enterprises, which make up 99% of the UK’s manufacturing sector, are still working out where the benefits lie. They typically do not have the same capacity to invest. Although many SMEs are not running AI at the operational level yet, we are seeing quick wins that help them understand the value and how they can scale from there.

Interviewer: You mentioned BMW. Could you share some insight into your collaboration with them?

Giovanna: This was a project at the Hams Hall plant near Birmingham. We looked at the data they were already collecting and improved their statistical process control by quickly digesting large amounts of machine data. Using data analytics and more traditional machine learning, we assessed how different machines were performing. We deployed a pilot and showed how we could reduce scrap by about 97%.

Interviewer: Many people cite challenges in scaling AI. From a manufacturing perspective, what are the biggest hurdles, for example in data availability or synthetic data generation?

Giovanna: Manufacturing is traditionally fragmented in terms of systems, and the data is also fragmented. It is often not AI ready and data quality can be poor. SMEs in particular have legacy systems, and many do not collect shop floor data in the way larger companies do. Synthetic data generation with generative AI is starting to help by creating the initial data needed to build AI solutions. The other challenge is digital skills. We need to upskill the workforce. Manufacturing also faces a perception problem, as it is not always seen as attractive. If we integrate digital technologies, we can attract people who are passionate about these tools to come into manufacturing. I see a bright future from that perspective.

Interviewer: As AI becomes more integrated into manufacturing, how do you see roles changing?

Giovanna: Roles will definitely change, although we do not yet have a complete picture of how. Manufacturing is driven by cost and profit, and automation helps scale production and reduce costs. So a natural question is whether further AI‑driven automation will reduce the need for employees. However, the know‑how resides with people. Completely replacing people is not a good strategy, and I do not see that happening. We need to design systems that leverage the flexibility and problem‑solving abilities of humans, particularly in new or unexpected situations, and use AI to enhance what they already do rather than replace it.

Interviewer: Looking ahead to the next three to five years, what realistic advancements do you foresee in industrial AI, particularly around frugal AI and continual learning?

Giovanna: The most tangible benefits I see in both small and large companies come from computer vision, especially in predictive maintenance. Over the past five years, adoption has ramped up, and I expect this to scale further in the next five. Vision‑based systems are easier to trust and verify. When the model highlights something in an image, you can see it and validate it. What is more difficult are opaque models that ingest raw sensor data. Continual learning is essential here. We not only need to scale the AI technology we already have, but we also need advances in the technology itself. Manufacturing is a constantly changing environment. We need AI that can adapt to new, unseen circumstances, work with limited data, and retain performance as conditions evolve. As continual learning matures, I believe we will see wider deployment on shop floors across companies.

Interviewer: Finally, what do you hope to take away from The AI Summit London?

Giovanna: It has been a great experience. I have met people who understand the business from different perspectives, and we have been able to share challenges and opportunities. It is an excellent platform for building connections that lead to collaboration.

Closing

This conversation with Giovanna Martinez Arellano highlights both the progress and the practical hurdles of industrial AI. From the trust and verification advantages of computer vision to the promise of continual learning, the path to scalable impact is becoming clearer. 

As skills evolve and data foundations improve, the combination of human expertise and AI augmentation is set to define the next phase of manufacturing innovation.