蜜桃传媒

The mission of the BEAMS Initiative is to engage the AI and modeling communities in devising tests for how well artificial intelligence and machine learning tools support the modeling process, so as to foster the development of more responsible and ethical tools.

AI-enabled tools in the simulation and modeling community are actively developing and many have already been released to wide swaths of practitioners. There are currently no generalizable guidelines for evaluating these tools’ ability to advance the state of the art in modeling and simulation, either in terms of correctness (i.e., do the tools produce coherent thinking?) or usefulness (i.e., regardless of whether or not the tool functions, is this a useful tool for solving problems?). The BEAMS Initiative brings together people who build these tools (e.g., researchers, software engineers and data scientists) as well as people who use these tools (e.g., educators, professionals and modelers) to establish benchmarks that allow us to confirm not only technical correctness but also to chart a pathway towards usefulness.

The BEAMS Initiative includes two working groups: a steering group and a technical group. The steering group advises and oversees the direction of the BEAMS Initiative, including developing the theory underlying the benchmarks along with prioritizing the benchmarks to be implemented. The technical group focuses on implementing the vision laid out by the steering group in the form of unit tests for the sd-ai project at .

Meetings of all collaborators in the BEAMS Initiative are held on an annual basis, while meetings of the technical and steering groups are held monthly (see ). All working group meetings are open to all collaborators in the BEAMS Initiative.