Artificial Intelligence

About the Header

Using advances made through an internal research initiative, SwRI is developing an electronic warfare LLM to autonomously generate mission data files (MDFs) that optimize stealth flight paths, identify and classify threats, and execute countermeasures. Traditionally, aircraft MDFs required the expertise of a human subject matter expert.

As artificial intelligence (AI) proliferates across the full range of human experience, Southwest Research Institute (SwRI^®) is leveraging its computer and data science expertise to navigate the opportunities and challenges associated with its use, particularly with Large Language Models (LLMs). SwRI is advancing responsible AI to make complex workflows faster and more intuitive. Through an Internal Research and Development initiative, SwRI developed a framework for applying LLMs to real-world systems, including a prototype that lets transportation system operators use natural language to perform multi-step tasks. Instead of navigating multiple screens or commands, staff can ask a question and get results instantly, boosting productivity while maintaining oversight and accountability in AI-driven tools.

Comparison of a human posed next to 3D animation from markerless motion capture technology

SwRI advanced a markerless motion capture technology to capture human movement to create 3D animations for gaming, training and entertainment applications. The Biomechanical Evaluation and Animation Motion Capture — BEAMoCap™ — tool converts video (left) into realistic 3D animations (right) without the conventional marker suits worn by actors.

Woman looking at her computer screen using SwRI's LLM initiative chatbot for ITS technologies

As part of SwRI’s LLM initiative, analysts integrated a chatbot into its Intelligent Transportation System technologies to evaluate how natural language can simplify complex workflows and improve operational productivity.

Man speaking to SwRI's Rhodium drug development software projected on a screen

SwRI developed the Generative Approaches for Molecular Encodings (GAMES) LLM to understand and generate Simplified Molecular Input Line Entry System (SMILES) strings. These strings represent the structure of molecules using a short series of text characters. These facilitate storage, retrieval and modeling, particularly for SwRI’s Rhodium^™drug development software, shown here.