Published on 1/8/2025 | 4 min read
The human brain has an incredible ability to blend sensory experiences—a subtle phenomenon that plays a key role in how we perceive the world around us. Associations between colors, shapes, and tastes can strongly influence our experiences. For instance, the color of a glass or the background music at a bar can alter the way we taste wine.
Recent studies suggest that artificial intelligence systems, particularly generative AIs like ChatGPT, are beginning to exhibit similar behaviors. These systems can seemingly "taste" colors and shapes, reflecting a process known as cross-modal correspondences—where sensory qualities such as taste, sight, and sound interact. Once thought to be a purely human trait, this fascinating capability is now being observed in AI.
Humans naturally associate certain colors with specific tastes. For example:
These associations are not just personal but shared across cultures. Studies have shown that people tend to have similar sensory responses to colors and shapes regardless of their background.
Experimental research also reveals that:
These associations may reflect an evolutionary survival mechanism, where round shapes are perceived as safer, while sharp ones signal potential danger.
Encouraged by these findings, researchers led by Carlos Velasco at BI Norwegian Business School explored whether generative AIs trained on vast amounts of human data would display similar sensory associations. In a recent study, ChatGPT was asked about the taste associations of various shapes and colors, and its responses were strikingly similar to those of human participants in past studies.
For example:
These similarities suggest that AI systems pick up on the same sensory biases humans have formed over time, likely reflecting the data they were trained on.
AI’s ability to replicate sensory associations offers valuable insights for marketers. By understanding cross-modal correspondences, AI could assist in designing effective product packaging and marketing strategies. For instance:
However, experts caution that while AI can identify these patterns, it lacks the creative nuance that humans bring to design. As Charles Spence from the University of Oxford notes, AI’s findings should serve as inspiration rather than a definitive solution. Human creativity remains essential for adding subtlety and uniqueness to sensory experiences.
In a lighthearted exploration, Velasco’s team tested AI’s ability to match music with flavors. When asked to suggest music for mulled wine, ChatGPT recommended:
Although the AI's musical pairings were unconventional, they offered intriguing insights into how music and flavor might interact, drawing on principles of sensory blending that humans naturally experience.
The exploration of AI's ability to perceive and blend sensory experiences is still in its early stages but holds immense potential for fields like product design and marketing. While AI can replicate patterns found in human sensory perception, its creative limitations mean it should be viewed as a tool for inspiration rather than a standalone solution.
As researchers continue to study AI’s sensory capabilities, they aim to uncover deeper insights into the associations between senses—opening possibilities for closer connections between humans and the machines we create.