In 'Blind Ambitions', players step into the quirky world of vision language models (VLMs) that can read texts but struggle hilariously with low-level vision tasks. As a budding AI researcher, your goal is to train a VLM and help it master simple tasks involving shapes and patterns. Each level adds unique challenges where players must design visual puzzles, monitor the VLM's learning curve, and hilariously interact with its often wrong answers as they try to complete objectives layered with educational content about visual perception and geometry. Engage in humorous dialogues with the VLM, curate training datasets, and solve problems before the deadline to ensure your research paper gets published!