can't parse a clock face. Researchers from Scotland's University of Edinburgh have found that AI models that can process text and images — otherwise known as multimodal large language models ...