Exploring AI’s Taste for Color and Shape: Can Machines “Taste” What Humans Do?

The article explores how the human brain often blends sensory experiences, a phenomenon known as "cross-modal correspondences," where colours, shapes, and sounds can influence our perceptions of taste and other sensory qualities. For instance, red and pink are often associated with sweetness, while green or yellow hues are linked to sourness. 

This blending of senses is so ingrained in human perception that even artificial intelligence (AI) systems, like generative AI models such as ChatGPT, exhibit similar associations, reflecting the biases present in the data they are trained on.

Researchers, including Carlos Velasco from the University of Oslo, have found that AI can "taste" colours and shapes in much the same way humans do. AI's associations often mirror human perceptions, showing that this cross-talk between senses is widespread and fundamental to our sensory experiences. However, AI does this by reflecting patterns in existing literature, rather than experiencing the world in the same nuanced way humans do.

This blending of sensory modalities extends to the marketing world, where the design of food packaging and product shapes can influence consumer perceptions, like how a product's colour can alter the perceived sweetness of chocolate. 

Researchers are now exploring whether AI can uncover new cross-modal associations, potentially offering fresh insights for product design, though human creativity and intuition still play a crucial role in harnessing these findings effectively.

Ultimately, the article reflects on the surprising overlap between human sensory experiences and AI's ability to process these associations, suggesting that AI might be a tool for further exploring and exploiting these sensory links in fields such as marketing and design.