Computer Modelling with Single Prompts (POSTER)
Maciej Matyka
Regardless of how we look at AI large language models (LLMs) - as a massive collection of data from which we can cleverly extract information, as an assistant who can perform simple tasks for us and write simple codes, or perhaps as a machine that randomly selects words, in a sense guided by what it have had has seen in the past - we are undoubtedly witnessing a revolution. I will present selected aspects of the use of modern large language models, such as Gemini, Grok, ChatGPT, DeepSeek, and Claude. I will discuss the concept of a single prompt and its use to generate computer code for dozens of models across computational physics, statistical physics, computational fluid dynamics, and more. I will illustrate the presentation with practical examples of how language models generate code for research in computational physics with a focus and specific examples in porous media flows and computing tortuosity.