TLDR: A comprehensive review of academic and industry approaches to integrating AI into Computer-Aided Engineering (CAE) software reveals a significant gap in user experience (UX) validation. While industry actively implements AI features like Large Language Models (LLMs) and adaptive interfaces for UX improvement, academic research often focuses on technical capabilities without rigorous empirical UX evaluation. The paper highlights key areas of divergence and proposes avenues for future research to better align academic innovation with industry demands for user-centric CAE tools.
Computer-Aided Engineering (CAE) software is vital for engineers to simulate and optimize complex models, from aerospace designs to automotive manufacturing. These tools allow for virtual testing, which significantly reduces costs and speeds up development. However, despite their power, many CAE tools present considerable challenges in user experience (UX). Engineers often face difficulties with complex tasks like preparing geometry, generating meshes, setting up physics, and interpreting results. Incorrect settings can lead to hours of wasted computation time, limiting the adoption and effective use of these powerful simulation capabilities.
Artificial intelligence (AI) is emerging as a transformative force in simulation workflows. For instance, large language models (LLMs) like AnsysGPT offer continuous technical guidance, while Siemens’ Industrial Copilot aims to make interfaces more intuitive and reduce cognitive load. Beyond assistance, AI can accelerate simulations by using neural networks trained on past results, allowing non-specialists to evaluate designs in minutes instead of hours, as seen with Ansys SimAI. This democratization of simulation capability represents a major advancement in user experience.
However, despite AI’s clear potential to improve CAE’s UX, there’s a lack of a clear, unified understanding of how these advancements are being implemented and validated in both academic research and industry practice. It’s often unclear if academic explorations align with industry needs, which AI-driven UX enhancements are gaining traction, and what specific research gaps are hindering the widespread, effective application of AI in CAE.
Bridging the Divide: Academia vs. Industry
A recent multivocal literature review (MLR) titled AI for Better UX in Computer-Aided Engineering: Is Academia Catching Up with Industry Demands? A Multivocal Literature Review aimed to address this knowledge gap. The study systematically analyzed AI advancements impacting CAE software UX, identified disparities between academic research and industry implementation, and mapped underexplored areas for future investigation. The MLR methodology combined a systematic literature review (SLR) of academic research with a grey literature review (GLR) of industry practices.
A key finding from the academic review was a notable absence of published empirical UX evaluations for proposed AI methods, despite frequent claims of user benefit. In contrast, industry practices, as identified through the grey literature from market leaders like Siemens, Ansys, Altair, and Autodesk, show a different emphasis. Companies actively market AI features, explicitly detailing how capabilities such as LLMs and automation are designed to improve UX. However, mirroring the academic gap, the reviewed industry literature generally lacks formal empirical validation results, such as usability metrics, potentially due to competitive sensitivities.
Also Read:
- GenAI’s Role in Automotive Software: From Concept to Code
- Large Language Models: A New Frontier for User Story Creation and Quality Assurance
Key Areas of AI Integration and Disparities
The review categorized AI applications in CAE into several key areas:
-
Workflow Automation and Efficiency: Both academia and industry are exploring AI for automating tedious tasks. Academic research often focuses on specific algorithms for tasks like mesh generation, while commercial efforts integrate AI more broadly into end-to-end workflows, such as CAD-to-CAE automation. A gap exists in academic research regarding holistic workflow integration.
-
Simulation Acceleration and Optimization: There’s significant overlap here, with both sectors using AI (especially physics-informed deep learning) for faster simulations and predictions. Academic studies tend to focus on methodological proof-of-concept for specific phenomena, while industry integrates these methods into comprehensive platforms for broader design behavior prediction.
-
Generative Design and Design Space Exploration / User Experience and Guidance: This is where a clear disparity emerges. The academic literature reviewed showed minimal focus on AI-driven user experience or empirical UX evaluations. Conversely, industry shows strong activity in using generative AI and LLMs for design creation, exploration, and direct user guidance (e.g., support, adaptive UIs, recommendations). The integration and UX assessment of such systems within CAE tools appear largely underexplored in academic research.
-
Data-Driven Analysis and Knowledge Management: Academic work emphasizes foundational aspects like synthetic data generation and formal knowledge representation. Industry examples focus on applying machine learning within tools to infer specific properties or predict performance based on existing data. There’s a gap in translating foundational academic work into practical, user-facing prediction tools in commercial software.
-
System-Level Integration: Both academia and industry are interested in AI agents and digital twins. Academic research often develops theoretical foundations or specific interaction models, while industry focuses on implementing these concepts within integrated, commercially-oriented platforms. The challenge lies in bridging foundational academic work with the practical implementation and UX considerations of complex industrial systems.
-
Manufacturing and Quality Control: While manufacturers acknowledge the link between CAE and manufacturing, academic research provided specific examples of AI predicting quality from design data. Industry literature showed less direct integration of AI for quality control within the core CAE UX workflow itself, suggesting a potential gap in tightly coupling AI-driven manufacturability and quality control predictions directly into the CAE interface for immediate user feedback.
-
Core AI Methodology Research: This is primarily the domain of academia, which focuses on advancing AI methodologies using CAE problems as test cases. Industry, on the other hand, concentrates on applying established AI techniques to deliver user value rather than publishing novel AI methods.
The review concludes that while there is considerable interest in using AI to improve CAE software UX, significant disparities exist between academic and industry approaches. The lack of empirical UX evaluation in academic studies and the limited formal publication of industry UX research are notable challenges. Future work should focus on developing novel methods and prototype solutions that tangibly improve the user experience of AI capabilities within CAE software, directly addressing these identified gaps and leveraging opportunities from broader AI/UX research.
