Product Development Innovative Engineering Requires Humans and Machines

Related Vendors

Dell GmbH - Technologies

Wuxi InfiMotion Technology Co., Ltd.

Exciting times lie ahead for designers and engineers: Traditional and generative AI will provide relief in daily tasks while simultaneously fostering groundbreaking ideas. For this, companies now need to establish the right foundation.

Despite the general AI euphoria, artificial intelligence (AI) is still in its infancy, especially in engineering. It is undisputed that CAD software and design processes will fundamentally change through AI. Above all, routine tasks will become easier for engineers in the future. In this way, they can then focus more on the creative and strategic aspects of their work. However, it will take a little longer before the even more powerful generative AI or AI agents, as assumed by specialists, have penetrated all subfields. Then, however, the diverse possibilities of (generative) AI and newer technologies based on it will not only lead to efficiency improvements but also bring about a completely new way of designing and developing.

AI can now do more than just take over monotonous and redundant tasks.

Because artificial intelligence can now do more than just take over monotonous and redundant tasks. Experts also agree that, particularly in the form of generative AI, it will elevate the creative potential of engineers to a whole new level. A relatively new development is AI agents (Agentic AI), which are also causing disruption. They can not only act as assistants on users' commands but also interact socially in an autonomous manner, meaning they can receive, process, and handle information. Analyzing or generating data and passing the conclusions drawn from investigations on to other AI agents or human engineers are also among their capabilities.

AI is Making Its Way Into Everyday Design Tasks

Even today, there are initial applications of artificial intelligence in design that offer a glimpse into the future.

• One example is the automated dimensioning of 3D models according to DIN standards. AI can provide precise dimensions within seconds. This task alone has previously cost designers a lot of time: either because they had to set the dimensions manually or had to subsequently check tools that only sporadically worked flawlessly until now.
• AI can also be used in the preparation of simulation models, for example, to filter out irrelevant details and thus optimize working time. For instance, if an engineer designs an exhaust and wants to simulate the exhaust gas circulation within the structure, it is unnecessary to consider the strength or the exact nature of an external bracket. However, these values are stored in the model, and it would take a lot of time to remove them manually for the simulation. Artificial intelligence can do this reliably in a matter of seconds.
• Similarly helpful is AI in the visualization of 3D models. For virtual prototypes and CAD models, it is important in the design process that the representation is not only extremely detailed but also that the workstation can render the components in real-time. For this purpose, Nvidia's DLSS (Deep Learning Super Sampling) technology is used, for example. It utilizes AI to render images in low resolution and then upscale them via a neural network. The image quality does not suffer from this type of upscaling, but the graphics card requires less power compared to traditional methods. DLSS can also enhance image output in flow simulations and crash analyses and visualize movements, such as for 360-degree views, more smoothly.

The Boundaries of What is Possible in Design and Development are Shifting

Far-reaching applications are already emerging for the near future. Generative AI will, for example, be able to design components that are entirely novel in form and function. The creativity and analytical abilities of designers are based—just like those of artificial intelligence—on what has existed so far. However, humans limit themselves by always keeping in mind how components are manufactured. Even five-dimensional milling has its limits, which generative AI does not need to consider. Through sintering, modern 3D printing, and additive manufacturing methods, designers may in the future be able to develop and produce organic structures and components that exceed the performance of classical designs and defy current imagination.

Another fascinating future scenario is AI-supported optimization of existing designs. Here, AI could independently identify weaknesses, suggest alternative solutions, and improve the efficiency of final products. The designer would thus become the "curator" of various AI-generated options, representing a completely new type of creative process.

Product Development

How AI is Advancing Ship Development

"Trust me, Bro": The Eternal Question of Trust in AI

Despite the enormous potential, several hurdles still stand in the way of the widespread introduction of AI in design and engineering. A central challenge is the question of trust: Can engineers and companies trust the results of AI blindly? The answer is short and clear: No. Therefore, the verification and validation of AI-generated designs will initially mean additional effort. However, this effort will decrease as AI models mature. Furthermore, it is possible that companies or software providers may offer their own AI tools in the future to verify specific calculations and models—essentially an AI that checks the design AI based on specialized data foundations.

Can engineers and companies blindly trust the results of an AI? The answer is short and clear: No.

Legal questions also remain unresolved. In principle, it is possible to feed an AI model with all the knowledge from all patents and developments that are publicly accessible. This would make a lot of sense, as it is the only way a company can avoid reinventing the proverbial wheel. However, the question arises whether a component based on such a wealth of data might violate patent rights. This is a problem that remains unresolved and one that lawyers are currently finding difficult to assess—simply because the necessary legal precedents are lacking. Although the EU is working on corresponding regulations to create legal certainty, they have not yet come into effect.

Creating the Right Foundation for AI

To use AI profitably and effectively in design, companies must first establish the right data foundation. This is perhaps the greatest challenge. However, this does not only involve collecting data but also selecting and curating it: not all information is truly relevant for training a highly specialized AI. Above all, it is essential to incorporate the specific knowledge and experience of employees. At the same time, the data must also be regularly updated, ideally in real time. The good news, however, is that it is becoming increasingly easier for small and medium-sized enterprises to train their artificial intelligence.

Using the Right Hardware for the AI Era

In addition to these software-related challenges, integrating AI into design processes also places new demands on hardware. Companies need not only new workstations to accelerate processes for the AI era of engineering. Powerful work devices are, moreover, the key to fully leveraging the potential of AI-supported CAD software. Workstation manufacturers like Dell Technologies rely on a combination of high-clocked CPUs and high-performance GPUs with large graphics memory to achieve this. The reason lies in the architecture of modern CAD software, which still primarily benefits from processors with fast single cores. Over the past decades, functions that benefit from multi-core processors have only been introduced sporadically to the market and have yet to gain widespread adoption. GPUs have proven effective for AI-based features, as they are also essential for visualization and rendering. Neural Processing Units (NPUs) could become more important in the future, but so far there are no applications in the CAD field that would benefit more from a Neural Processing Unit than from a GPU.

The Daily Work of Engineers Will Change

All these technological advancements in the field of AI will have a tremendous impact on the daily work of designers and engineers in the future. Theoretically, highly specialized AI agents are already capable of processing information, making a decision, and passing it on to another AI agent or a human designer.

Example: If a brake pedal for a new electric vehicle is to be developed, it usually takes weeks from concept to simulation. This time could be shortened with today's technology by providing an AI agent with specific information about the maximum pedal force required, the length of the travel, the material, the mount, the target weight, and the bore for sensors. Once the CAD AI agent has completed its work and created the concept, it passes the results to the CAE AI agent, which subjects the model to a simulated durability test with 800,000 cycles and then returns this result to the human engineer for review. While this is still theoretical, the only remaining obstacle is that selecting the data and training these AIs is currently not possible without highly specialized experts.

AI Acts as a Source of Inspiration for the Engineer

Currently, there is still a shortage of such specialized personnel in engineering. In the long run, however, companies in design will definitely have to address the development of AI expertise. Only in this way can they remain competitive and likely innovative in the long term. AI-supported designs certainly have potential, even if they serve "only" as a foundation and inspiration for human engineers. But engineers need not worry about their jobs in the future: AI is and will remain a tool that supports them in solving complex problems and enables new approaches. It cannot replace them. Harnessing and leveraging the synergies between human and artificial intelligence is the key to success in the industry.

AI remains a tool that supports solving complex problems and enables new approaches. It cannot replace them.