As AI becomes more widely used across industries, it will continue to have a profound impact on the development and progress of human society and revolutionize every aspect of technology and human interaction. By 2024, Forrester predicts that enterprise AI initiatives will help increase productivity and creative problem-solving by 50 percent and will impact the work of engineers and educators, among others, by helping them save time and giving them more energy to focus on other projects that advance the cause of science and engineering.
AI and Simulation are Critical to Designing and Developing Engineering Systems
As AI moves into the mainstream across industries and applications, complex engineering systems that don’t use AI will seem out of place. Engineered systems combine components and subsystems from multiple domains to create intelligent systems that sense and respond to the world around them. For example, wind turbines use a combination of mechanical components (turbine blades and gearboxes), electrical components (generators), and control components (blade pitch). Complex AI systems have flourished largely because of the increased integration of simulation into the design and development of these systems.
Simulation is a widely validated methodology for performing the multi-domain modeling and simulation required to develop complex systems that can process sensor data to help develop perceptual and autonomous systems. However, as system complexity increases, some simulations can become too computationally intensive for system-level and embedded designs, especially in tests where models need to be run in real-time. In such cases, AI can also enhance the simulation by using a reduced-order model.
Reduced-order models (ROMs) can accelerate simulation while providing acceptable accuracy for system-level testing of control algorithms.ROM models can complement first-principles models to create variant implementations that can perform trade-off analyses between accuracy, performance, and complexity.
Increasingly, engineers are exploring ways to integrate AI-based ROM models into their systems. This can help accelerate desktop simulations influenced by third-party high-fidelity models, enable hardware-in-the-loop testing by reducing model complexity, or accelerate finite element analysis (FEA) simulations.
AI Practitioners Must Consider Performance When Deploying Models to Edge Devices Where Speed and Memory Are Critical
AI Model Selection
AI models can have millions of parameters and require large amounts of memory to run. Accuracy is a primary consideration in research, but there is a trade-off between memory and accuracy when deploying AI models to hardware AI practitioners must consider how their performance will differ when deploying models to devices where speed and memory are critical AI can be added to existing control systems as a smaller component without relying on end-to-end AI models, such as those commonly used in computer vision to detect objects. vision to detect objects.
A significant topic when discussing smaller AI models is incremental learning. Incremental learning is a machine learning approach that enables models to continuously learn by updating their knowledge in real-time as new data becomes available; it’s an efficient way to deploy at the edge.
The success of complex AI systems depends on integrating simulation into the design and development of engineered systems
GenAI Helps Engineering Professors Teach More Advanced Topics
Generative AI (GenAI) is a disruptive technology. In 2024 and beyond, engineering professors will be using this technology at scale to help students in the classroom. Much like the Internet or mobile phones, GenAI is starting a revolution that will improve the entire field of engineering education.
The main advantage of using GenAI in the classroom is that it can help save time when teaching basic skills, such as computer programming, to engineering students. Instead of having to spend as much time teaching low-level concepts, professors can now focus on teaching advanced topics such as the design and implementation of complex engineering systems. By using technologies like ChatGPT to run simulations and create interactive exercises and experiments, professors can save time and better engage students.
Professors can teach students the skills necessary to effectively master GenAI, such as prompt engineering. This helps students develop critical thinking skills that are learned and applied rather than relying exclusively on computers to solve problems. Therefore, students are best served by being independent learners in a variety of engineering disciplines, and engineering educators can further expand the curriculum while sharing expertise in more advanced concepts.
As AI matures, it will play an increasingly visible role in increasing the productivity and potential of engineers and educators. When building complex engineering systems, engineers are wise to adopt AI-assisted simulations and smaller AI models. In academia, generative AI is helping educators save energy and make students more independent. With AI, many industries and educational institutions can make smarter decisions, get actionable advice, and increase efficiency.