Imagine an AI tasked with scooping flour, but there's no spoon in sight. Could it creatively use a bowl instead? This is the essence of creative problem solving, a crucial aspect of intelligence where one devises novel solutions using available resources. While large language and vision models (LLVMs) excel at many tasks, creative problem solving remains a challenge. This research paper delves into how we can bridge this gap, drawing inspiration from the field of Computational Creativity (CC). One key idea is to augment the "thinking" of these models by enriching their embedding spaces—the core representations of concepts within the model. The paper explores three strategies: exploratory (searching the model's possibilities), combinational (blending existing concepts), and transformational (reframing the problem). A simple experiment demonstrates the potential of transformational creativity. By providing prompts that highlight the shared function of different objects (like a bowl and a spoon for containing), LLVMs showed improved ability to choose creative substitutes. While preliminary, these findings suggest that by integrating principles from CC, we can enhance the resourcefulness of AI, paving the way for more adaptable and truly intelligent systems. The future of AI problem-solving may lie in its ability to think outside the box, just like a resourceful human.