Recently, Altair’s Jeff Glueck, senior vice president of global system integrator alliance relationships, sat down with Stavros Stefanis, a principal at Deloitte Consulting LLP, to discuss the future of generative AI (genAI), the future of engineering workflows, and more.

Jeff Glueck: I am joined today by Stavros Stefanis. He brings 27 years of consulting experience with deep knowledge in product design development and supply chain operations. Earlier this year, Deloitte published its first in a series of reports on the state of genAI in the enterprise. Stavros, can you give us a little background on the report?

Stavros Stefanis: Deloitte does a number of surveys quarterly; the state of generative AI in the enterprise was the first of a series of these quarterly surveys. The idea behind it is to capture the pulses of the market and the pulses across the industry of where generative AI is heading and its adoption.

JG: I understand that the survey results suggest that many AI-fueled organizations are on the verge of scaling up their efforts and embracing genAI in a more substantial way.

SS: Absolutely. Everyone, across industries, seems to be planning to use genAI. Confidence is increasing and the intent to build more knowledge around genAI is there. There is also a sense of pressure as well, [a sense] that you have to do something because your competitors are going to use genAI to gain an edge over you. The biggest areas I would say we see usage of genAI is around efficiency, productivity, and cost reduction.

The other thing to note is there is no “common” [genAI] platform in use so far. Everyone's trying to use different genAI solutions, relying on different capabilities.

JG: I find it very interesting, reading through the report, that organizations are focusing primarily on efficiency, productivity, and cost reduction as you've indicated. Yet I think we know that the big breakthroughs will likely come in the areas of innovation and growth. Let's talk about that.

SS: GenAI is applied across the supply chain, but let's focus first on the product development side. As you know, product design relies on complex data; many times this data is scattered across the organization. [A common entry point to genAI] is to get insights and bridge data sources that are disparate and allow engineers to connect different data-intensive areas to gain insights and improve development practices. As you evolve from that – from the more reporting/analytics types of use cases – you start getting into areas where genAI could be used to create automated bills of materials.

You could use genAI to write requirements and enhance requirements. Requirements are very semantically heavy, and they are a very, very good fit for the use of genAI to improve requirements management processes and the testing practices that go with requirements.

We also see more and more genAI use in the areas of ideation, leveraging existing design libraries, and giving users ideas on how to make the next generation of designs.

And the last area – but not the least – is the space of simulation management, where you have simulations that, before, you would have to create very complex operational domains to test them. Now, genAI can create those domains and scenarios a lot easier and give your organization a more comprehensive testing capability. In essence, genAI is used for more reporting and analytics now, but it’s very quickly going into the core areas of design, testing, requirements management, software simulation, etc.

JG: It seems like that genAI would go really well with the ulterior product set focused on digital twin as well. Do you see that as where the future of genAI is going to have the biggest impact on product development?

SS: Absolutely. In fact, one of the newer areas we're working in is battery management. The modeling capabilities in that space are evolving. GenAI and digital twins give us the ability to really understand battery performance not only during the design of the battery, but also linking it to the operations of that battery and how it performs in the field. That's a great case where you could use genAI to bridge all that data together to create the right inputs and insights that you need during the development cycle.

JG: There are going to be some organizations that rapidly embrace genAI, probably by starting in the efficiency/cost reduction phase, but then move quickly into innovation. What do you think will be the key characteristics of organizations that succeed in that transition?

SS: An important one is getting the right skills. A lot of companies haven't specialized in creating operational excellence using data science, so data science is a critical element to have. 

A second important skill is the understanding of the underlying technology, because genAI requires a lot of capability from a hardware perspective. So hardware knowledge of cloud engineering is also very important. You see a lot of companies are building more chip design expertise in-house because they want to collaborate with the semiconductor companies to create the next generation of chips that are used to power this very important technology.

And a third point that is starting to happen is that the idea of a super specialized engineer that knows and focuses on just one area is becoming history. Engineers now are asked to have experience in many domains. They are asked to know their customer and be in contact with the customer’s needs and wants – and to be able to incorporate those needs and wants into products.

The majority of products are either directly software enabled or indirectly software assisted. So the notion of building solutions that go through the life cycle of the product, even beyond the one-time sale, is also extremely important.

JG: You talked about the need for organizations to have skills around data science, the need to have product development teams that understand cloud engineering and hardware, and the need to be capable of understanding a variety of skill sets. I'm sure as organizations try to enhance capabilities for their product development, they're going to need to work with technology organizations that can bring a variety of tools to enhance their capabilities.

SS: Absolutely – I would say part of it is bringing the right data science and engineering talent to the table to help these companies grow their skills, because they're going through a skill transition. When you think what Deloitte brings to the table, it can give people the ability to link different functions and have a more holistic view of how products need to be developed and the methodology and rigor behind it. It's a very, very powerful play.

So organizations should look at partners who have a holistic view of the technology landscape, who can coach a client and say, You know what? Instead of running your simulation environment in a very fragmented mode, you need to see it in a more integrated way. And you need partners to help you optimize your infrastructure in parallel as you develop these new capabilities: there is high-performance computing (HPC), there is cloud enablement, there are different areas of opportunity that you need to see holistically.

JG: Could you speak more about the integration of simulation and genAI?

SS: Sure. Simulation is very important because not only it is an advanced way to check behavior, it is also the most cost-efficient way. Why is it cost-efficient? Because it prevents or reduces the need for physical prototypes, which cost companies a lot of money. It's so powerful because not only you can simulate specific domains, you can simulate the product as a whole and see how these domains come together, whether it's mechanical, electrical software and so on. You can also simulate the operating environment that the product will be working on in the field. You can simulate land, you can simulate the roads, you can simulate cities, etc. It takes it to the level where you can really simulate your product in the field as if it were launched, which gives you amazing capabilities and insights.

With genAI, what you can do is really optimize the scenarios that you run on those simulations and collect inputs from different sources, whether it's engineering sources, customer sources, and so on. And through large language models, you can train your designs to become better at meeting simulation results, improving simulation results, and meeting the targets you have.

JG: Stavros, in terms of genAI, put on your future cap. The pace of AI is moving at a scale that's really different than what we've seen in other technologies. Within the field, what do you think is going to happen a year from now? Not even five years from now, just a year. How are people going to use genAI in the product development space in ways they're not even conceiving today?

SS: The pace is so fast that it's hard to predict the future because the future is really happening right in front of us every day. Putting on my future cap, in a year from now – especially in the engineering space – adoption could increase so much that people could start using genAI the same way we use the web and searches in our day-to-day lives. It can make things a lot faster for us [engineers]. It will not be able to do all our most important work for us, but I think it will be used very actively as a starting point in getting work done and in making decisions.

The other prediction that I have is that different functions and domains that were once very siloed before are going to start working more closely together because genAI will give people the platform to really provide insights from one function to the other and create dialogue. It will create new types of meetings. It will make everything more future-focused and predictive-driven.

To listen to the full interview, visit https://altair.com/resource/the-state-of-generative-ai-in-the-enterprise-now-decides-next?lang=en.