In March of this year, Ansys Chief Technical Officer, Dr. Prith Banerjee, took center stage as the keynote speaker at OzenCon, a simulation conference co-hosted by Ansys and Ozen Engineering. The buzz leading up to the event was not just about Banerjee's insights but also about the AI chatbot ChatGPT, developed by OpenAI, which showcased its impressive natural language processing skills.

ChatGPT: Transforming Simulation

During the presentation, Dr. Banerjee emphasized the potential impact of ChatGPT on the world of simulation, especially in technical fields like finite element analysis (FEA). Current simulation tools often involve complicated parameter setup, requiring expertise in areas such as aerospace engineering. However, ChatGPT aims to simplify this process by enabling users to interact with simulations in a more conversational way. This could be a significant advancement in simulation technology and could significantly lower the learning curve and expand the accessibility of simulation tools.

Design Intent Unveiled

Alex Tessier, Director of Simulation Optimization and Systems Research at Autodesk, sees immense potential in chatbot-driven design workflows. While CAD programs offer granular control over geometry, they cannot often capture the design intent. ChatGPT's conversational interaction style could bridge this gap, helping engineers explore design options more naturally.

Augmentation, Not Replacement

Tessier emphasized that AI-driven chatbots should be viewed as an augmentation of human capabilities, not a replacement. While they can streamline tasks like programming and design, engineers must still verify the outputs for quality and accuracy.

Reduced-Order Models and Uncertainty Quantification

Simulation job cost often hinges on solving time, and reduced-order models (ROM) and uncertainty quantification (UQ) are emerging as valuable tools to speed up this process. Ansys offers a suite of ROM technologies that can drastically reduce the complexity of simulations.

COMSOL has introduced a UQ Module that uses machine learning to build ROMs based on historical data, bypassing the need for full physics simulations. These advancements are reshaping how simulations are conducted.

ChatGPT and 3D Content Creation

NVIDIA's Omniverse has integrated ChatGPT-style input for scene creation. Instead of relying on predefined commands, users can use natural language prompts to build environments, making 3D content creation more accessible and intuitive.

While ChatGPT-style input offers freedom in phrasing commands, it also introduces challenges due to natural language's interpretive nature. Users must understand their domain to assess the quality of AI-generated output.