“Most current applications of Large Language Models (LLMs) use static inference, meaning the model's state does not adapt or learn from the task at hand.”