Large language models (LLMs) are making strides, but replicating the visuospatial prowess of specialized AI like AlphaFold in general-purpose applications is proving difficult. The AI community is actively debating the successes and shortcomings of current LLMs when faced with visuospatial tasks. Online discussions, particularly on platforms like Reddit’s Artificial Intelligence forum (https://old.reddit.com/r/artificial/comments/1muofp4/visuospatial_reasoning_successes_and_failures/), highlight user experiences and spark conversations about how to further integrate spatial reasoning capabilities into widely accessible AI interfaces.