AI use cases: The right tech for the right job

MARTINE HANNEVIK     Welcome to the Trust in Industrial AI video series, where we explore how to implement AI with both speed and confidence in safety critical industries. Today, we're exploring a very exciting topic, how to apply the latest AI technologies for key use cases within industries such as energy and maritime. I'm your host, Martine, and joining me today are two of our AI experts here at DNV, Abdillah and Luca, welcome to both of you.

LUCA GARRÈ     Thanks, Martine. It's great to be here.

ABDILLAH SUYUTHI     Thank you, Martine. Happy to be here. 

MARTINE HANNEVIK     So I think the first question will go to you, Luca. And that is, what AI technologies are best suited for different use cases within critical infrastructures.

LUCA GARRÈ     Absolutely, Martine. When talking about critical infrastructures, we're effectively talking about industries such as energy, transportation, food and agriculture and the likes. Each of these sectors experiences its own unique challenges and needs, and as such, AI solutions to be entertained. These sectors all to be contextualized to these challenges, needs and to the specific case ahead. In the maritime industry where I work as a data scientist, there are various trends among which, for instance, a trend towards predictive maintenance and that alone touches upon very many use cases that relates to AI and machine learning.  We have for instance, a need to predict when a next failure may possibly happen. In that situation that we may use for instance, models that operate on time series data. We may have a need for contextualizing data feeds coming from sensors and in that respect we may use traditional machine learning models to operate that contextualization. So we really have a large and broad spectrum of different use cases just in the maritime industry for instance, that could be effectively served by AI technologies. And then we have the latest breed of models arising from deep learning. Things for our models such as transformers like ChatGPT or convolutional neural networks that are proven to be very effective, for instance, at enhancing interactions between users and IT or content, digital content, for instance, or tasks such as computer vision, image processing and so on.

MARTINE HANNEVIK     Very exciting. So a lot of potential in the maritime industry.

LUCA GARRÈ     Indeed.

MARTINE HANNEVIK     And the, are you working on many specific exciting projects right now?

LUCA GARRÈ     Yeah, we obviously work on various projects across this broad spectrum.  
Off the top of my head, we have we're actively exploring the use of deep learning models to support our experts in visual processing, for instance, automatic object identification detection, anomaly detection and so on. So this is a broad field that we are actively exploring. For instance, together we are experts in our maritime services.

MARTINE HANNEVIK     Great. So you've mentioned a few various AI technologies and there's a lot of hype these days around the generative AI and large language models. And they're very effective for some use cases, but maybe not for others. Where would you use large language models and where would you use other technologies?

ABDILLAH SUYUTHI     This is an interesting topic. Yeah, let's see. The power of LLM is actually in their abilities to understand and generate human like text. And this ability enables LLM to comprehend unstructured data. So this is key differentiation. Now when we talk about application of AI for safety critical infrastructure, we usually think about traditional machine learning where it processes structured data. However, we know that there is huge untapped potential of unstructured data. Now you imagine. What we talk here is about design engineering report, for example, maintenance log, even design simplification and even e-mail communication and also incident reports and many more. So imagine the time when industries start to crunch all this unsanscripted data, there will be a lot of opportunities.

MARTINE HANNEVIK     That's exciting. So what you say then is you combine that with other technologies such as machine learning.

ABDILLAH SUYUTHI     You got it right. Well, actually combining LLM and semantic technologies. I can give you a real example on this. Let's say when managing the energy grid, the energy companies can have a more robust and reliable system for managing energy grid by combining LLM and semantic technologies. And in practice, this means that LLM can use a structured information framework that is provided by semantic technologies to represent its, its, its grid elements, the relationship and more accurate operational data. So we in DNV, we have been dealing with this for quite some time with the machine learning now we also focusing more and more on the generative AI. So by combining LLM and semantic technologies, LLM enable the operators to have more accurate insight and more useful recommendation by simply using natural language.

MARTINE HANNEVIK     So it's about combining the various technologies for the right use cases.

ABDILLAH SUYUTHI     Yes, just like an orchestra.

MARTINE HANNEVIK     That's great. So you've talked about a lot of interesting technologies and how to match them with use cases. But I'm sure a lot of our listeners are wondering, you know, how to get started. So what would be your advice here, Abdillah?

ABDILLAH SUYUTHI     OK. My advice would be start from your pain point, start from your problem and your goal and then you can engage with the expert to get you through to the processes to ensure that whatever AI solution that you choose is indeed appropriate for your problem at hands.

MARTINE HANNEVIK     Thank you. What about you, Luca, any final thoughts or advice?

LUCA GARRÈ     I just want to emphasize that AI is obviously a very powerful tool and as such, must be handled with care, must be implemented thoughtfully and responsibly. So there is a fine balance there. And we at DNV are committed to help our customers strike that fine balance when it comes to their own needs and use cases.

MARTINE HANNEVIK     Well, thank you so much, both of you for sharing your insights with me and our listeners and thank you for tuning in. If you have any questions or want to learn more about how we at DNV can support you with safe application of industrial AI, then please visit our website. Thank you.