Over the past fifty years, shape grammars helped develop a critical outlook in the development of computational design by emphasizing the complexity and interplay between algorithmic processes and artistic intuition. Having previously shown how computation could be contextualized without representations or electronic digital computers, shape grammars hold the potential to provide a critical framework for generative artificial intelligence (AI) development. To this end, this paper outlines shape grammars’ historical evolution at the intersection of visual creativity and computational thinking, with a focus on the concepts of “seeing” and “doing.” The identified gaps between human vision and perception, language, and the inner workings of artificial intelligence systems are further studied through three interwoven questions that relate to visual ambiguity, processes of meaning-making, and mereology. The presented inquiry aims at fostering a deeper understanding of human creativity and computational design, which are much-needed ingredients that can help shape the future of generative AI.